Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

MAVC Final Report Open Content in Kalimantan

380 views

Published on

Wikimedia Indonesia and Humanitarian Open Street Map Indonesia present final report for Open Content in Kalimantan: Wikipedia & Open Street Map for Transparency. As part of Making All Voices Count grant program.

Published in: Education
  • Be the first to comment

  • Be the first to like this

MAVC Final Report Open Content in Kalimantan

  1. 1. CONTRACT NUMBER: HO CIM1008740 Reporting Period: April 2015 - March 2016
  2. 2. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 2 Preface This final report reflecting result, success, and challenges in our journey of empowering ordinary Indonesian citizen to participate in contributing to information regarding their surrounding in digital era. A total of 35 events attended by 700 people, in order to get trained throughout. The training was supported by nine other entities, and finally Eastern Kalimantan glow for activities when Open Street Map identify the world map of activity. The evaluation done by the third party describe in details on how we do it, what we achieve, how are the citizen do it, and answers the common question of what kind of quality that their contribution reflect? We thank Making All Voices count in allowing us, Wikimedia Indonesia and Humanitarian Open Street Map Indonesia, to document this important journey, we also would thank Sumana Harihareswara from Changeset Consulting LLC, Yayasan Bumi & Center of Borneo Environmental Remote Sensing, University of Mulawarman. Lastly, we would like to thank our dedicated team in making this project successful: Arief Rahman (Samarinda/ Balikpapan), Suyono Darul (Samarinda/ Balikpapan), Harry Mahardhika (Jakarta),VanessaWinona (Jakarta), Ika Rahmat (Jakarta), Yantisa Akhadi (Jakarta), John Mark Vandenberg (Australia). Siska Doviana Head of Board of Executive
  3. 3. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 3
  4. 4. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 4 Table of content 2 Preface 4 Table of content 5 Administrative data 7 Introduction and highlights 8 Summary of Activities 8 Acquiring numbers as baseline projects 8 Development of instructional toolkit 9 Development of software for Wikimedia Article Creation 10 Draft 1 and Draft II tested to local partners 11 Kickoff Meeting with Local Partners in Kalimantan 11 Initial OpenStreetMap and Wikipedia Workshop with local partners in Kalimantan 12 Follow up OpenStreetMap Workshop with local partners in Kalimantan 13 Follow up Wikimedia Content Workshop with Local Partners 15 Result 17 Numbers of Wikipedia Page created by participants trained 18 Numbers of Wikipedia page view of the pages created 19 Numbers of Edits in OpenStreetMap made by trained participants 21 Previous or potential problems and mitigations 26 Measurement of usages of Wikipedia pages overtime relative to content added 27 Outreach 28 Skills 29 Conclusion 31 Appendix
  5. 5. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 5 Perkumpulan Wikimedia Indonesia & Humanitarian Open Street Map PerkumpulanWikimedia Indonesia (WMID) established on 5 September 2008 and legally recognize as an association by the Ministry of Law and Human Right in 2011.Wikimedia Indonesia is founded to encourage growth, development and dissemination of knowledge in Bahasa Indonesia and other languages spoken in Indonesia for Free. As an association Wikimedia Indonesia have 58 members from throughout Indonesia and eight board member. Since 2010Wikimedia Indonesia has initiate and completed 15 projects and 21 sub-projects with the duration from 3 to 18 months period. Humanitarian Open Street Map (HOT) started recruiting intern in Indonesia to map in 2011 to help the Indonesian Government in assessing disaster respons. By 2014 they have coached students from 13 Universities in Indonesia and their “Train ofTrainers” program have successfuly train people to do data collection,data integration,and validation of data, which becomes a special skill or ability inbecoming OSM experts. Annual meeting of memberWikimedia Indonesia 2014 Administratif data Organizations responsible Project Name Contract Number Reporting period
  6. 6. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 6 Open Content in Kalimantan – Wikipedia & OpenStreetMap for Transparency Open Content in Kalimantan Project is a joint-project between Wikimedia Indonesia (WMID) and Humanitarian Open Street Map (OSM) in mapping and content production about Kalimantan. The project is focused on making mapping module for OSM and making bot articles in IndonesianWikipedia and BanjarWikipedia and training for organisation’s partner in Kalimantan.The toolkit is to create and using the internationally recognised toolsetsWikipedia and OpenStreetMap at the micro/local level – in this case, alimantan.T e ro ect reflects Wikimedia Indonesias and the Humanitarian OpenStreetMap Team’s commitment to the growth of open content in Indonesia and will train local communities and government to produce their own content, document their knowledge and map capacity in their localities. Rather than build a completely new toolset, this project takes existing tools and uses them to create local content. Contract Number HO CIM 18740 Reporting period This reporting period covers April 2015 – February 2016 OSMTeam 2014 Administratif data Organizations responsible Project Name Contract Number Reporting period
  7. 7. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 7 About the project Open Content in Kalimantan project is focused on making mapping module for OSM and making module for creating articles in Wikipedia and link it to OSM. A more advance idea also introduced, what about making bot articles in IndonesianWikipedia and BanjarWikipedia using OSM data. The highlight of the previous report includes the launching of OSM mapping module in Indonesian language, mapping and writing competition roll out, and recruitment of local resources in Kalimantan. Figure 1 - Mapping module in Indonesian language T e O en Content in alimantan is t e first in t e world on how to combine OpenStreetMap and Wikipedia as a tool to improve transparency.The focus of the activities is to enable local resident and government to built database of public infrastructure location and facilities. This would improve citizens accessibility on public infrastructure information w ic normall ard to access.T e first ste to reach the goal is by developing toolkits to address lack of data in Kalimantan. 1. Introduction and Highlights This section is an introduction and highlights of the past reporting period for the project.
  8. 8. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 8 2.1. Acquiring numbers as baseline projects Activity: Completed on previous report. Please refer to the previous reporting period for detail 2.2. Development of instructional toolkit A. Modification of Learn OSM into Project Materials Activity: Completed on previous report. Please refer to the previous reporting period for detail Figure 2 - Manual I and Manual II B. Modification of Wikimedia Material Activity: odification of Wikimedia aterial a ened during June 2015 to February 2016. During this time three Wikimedia Materials created; Manual I: A Guide How To Create Wikipedia Article. Manual II:PyWikiBot. Manual III: PAWS.Two of the materials were printed and can be downloaded online, while the last one was using a web format online. C. Creation of Joint Materials for Linking Wikipedia to OpenStreetMap Activity: On 26 February 2016, Wikimedia Indonesia invite Andy Mabett, from Birmingham England as a wikidata expert to talk directly to OSM team on how to link theWikimedia data and OSM data using wikidata identifier on O attributes table. As a result OSM team in Indonesia will add one slide about linking the two data betweenWikimedia and OSM to their training materials using wikidata.org. 2. Summary of Activities This section is a summary of the activities that have been implemented during the reporting period. It will relate to agreed indicators (if any) and compare with the original proposal or last annual review.
  9. 9. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 9 2.3. Development of software for Wikimedia Article Creation A. Preparation of source data Activity:There are two activities related to sourcedata. Wikimedia approach to local government for Kalimantan Government facilities (public service infrastructure i.e. ealt ser ices offices and alimatan base O data for Kalimantan endemic species. Total data compiled from the government for two reporting period is 1,584 data consist of health and education facilities in East and South Kalimantan. or our effort in obtaining O data in alimantan endemic s ecies we fail.We learn t at local O are more likely cooperate in doing campaign instead of providing data knowledge in local or Indonesian language. B. Development of Wikipedia Article Creation Software Activity:DuringthisreportingperiodManualII:Pywikibot and Manual III: PAWS was created. Both manual tested and prove successful to create article automatically on February 2016. As a result the software continued to be improve,pywikibot then being installed on the server and avaialable online as a virtual machine so user don’t have to instal it on their computer.The virtual machine called PAWS and tested on February 2016 we tried it on 15 people and sucessful in test.wikipedia.org. After the success we’re stepping back, and reassess change direction to using PAWS, including translation effort to other languages.We didn’t get to large scale test on wikipedia.org since the baseline numbers of geography articles in Wikipedia before the project roll out already reaches 7.895 articles for villages, kabupatens, cities, provinces, and islands. The existing articles mean that the information needed to be added manually, should we bot-create the articles in large scale,it will overwrite the existing articles. 2. Summary of Activities This section is a summary of the activities that have been implemented during the reporting period. It will relate to agreed indicators (if any) and compare with the original proposal or last annual review.
  10. 10. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 10 C. Wikipedia Community Consultations Activity: On February 27, 2016 Wikimedia Indonesia did Pywikibot training for Indonesian languageWikipedia community. A total of 17 people attended the session, coming from Samarinda (Kalimantan), Pekanbaru (Sumatera), Singapore, and Bandung. The consultation resulted usage of Pywikibot in Aceh Wikipedia for automatic article creation. D. Development of Wikipedia Toolkit in Indonesian Language Result: During June 2015 to February 2016 Wikimedia Indonesia develop Manual I: A Guide How To Create Wikipedia Article.A project audit for Wikimedia effort by Changeset Consulting LLC dub this manual as “Very polished in appearance and detailed in coverage“ The manual was launched in Goethe Institute on 31 March 2016 and tested to 35 participants.Thirty percent of attendees succesffuly created articles from the manual and the launch itself receives widespread coverage in Indonesian media. 2.4. Draft 1 and Draft II tested to local partners Result: For this reporting period Wikimedia Indonesia tested Manual I, Manual II and PAWS on February 26 and 27 training with the help of Andy Mabett. HOT OSM team was introduced to wikidata, andWikimedia Indonesia team was trained using pywikibot and PAWS. As a feedback all the participant feels the training was useful, PAWS was something that they mentioned is new and interested in to try,while some participant tried practice Pywikibot in other language Wikipedia. This event also marked consolidation between OSM and Wikipedia data by using wikidata. 2. Summary of Activities This section is a summary of the activities that have been implemented during the reporting period. It will relate to agreed indicators (if any) and compare with the original proposal or last annual review.
  11. 11. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 11 2.5 Kickoff Meeting with Local Partners in Kalimantan Result: Kick off meeting with local partners in Kalimantan is done during previous reporting period.We’ve done 12 days events from January to February 2015 ranging from local government meet up and upgrade to provincial level government meet up, signing Universities Memorandum of Understanding, training for Wikipedia Pywikibot, Open Street Map and Open Sea Map. On this reporting period we’d be focusing on workshop with local partners and follow up meeting and consultation (please refer to 2.7 to 2.8 for details) 2.6 Initial OpenStreetMap and Wikipedia Workshop with local partners in Kalimantan Result: Initial OpenStreetMap and Wikipedia Workshop with local partners in Kalimantan was done during previous reporting period. On February 16-18, 2015 for both workshop with 43 participant attending four workshops total and trained to contribute into Wikipedia manually, trained to contribute to OpenSeaMap, and train to contribute to OpenStreetMap. On this reporting period we’d be focusing on the result of these students after Wikipedia workshop by presenting its readership statistic after one year. (please refer to 2.7 to 2.8 for details) 2. Summary of Activities This section is a summary of the activities that have been implemented during the reporting period. It will relate to agreed indicators (if any) and compare with the original proposal or last annual review.
  12. 12. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 12 2.7 Follow up OpenStreetMap Workshop with local partners in Kalimantan Result:Based on the result of the previous workshop effort, Humanitarian OpenStreetMap switch “workshop only model” with one month competition.Within this reporting period there are total four digital mapping competition being held in Samarinda where the participant map Samarinda and Tenggarong, Balikpapan, and Kutai Kertanegara. All these workshops held in June, October, Desember 2015, and February 2016. In this reporting period we’re also going to cover the result of Wikipedia writing competition using manual labor entry. Out of 15 participant joining the works o in ebruar 201 fi e continue to contribute where we grant the one person as a surprise winner, on June 2015: a laptop.Total articles written are 205 articles b fi e contributors. Figure 3 -Wikipedia and OpenStreetMap workshop in Kalimantan For Open Street Map one of our indicator of success are the workshops creates 1.500 object in Kalimantan.When the result are in, the effort had mapped 16,157 buildings in O en treet a com ares to 1 4 buildings w en it first started. The chart above also illustrate what the competition in Eastern Kalimantan do for data addition compares to no competition in South Kalimantan.The data in January 2015 showed 2,229 buildings to 5,759 buildings in March 2016. 2. Summary of Activities This section is a summary of the activities that have been implemented during the reporting period. It will relate to agreed indicators (if any) and compare with the original proposal or last annual review.
  13. 13. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 13 2.8 Follow up Wikimedia Content Workshop with Local Partners. Result: On July 2015, we continue our effort to train students how to make a bot created article in Samarinda. T e e ent was su orted b T e Office of Communication and Informatics, East Kalimantan Province (Diskominfo Kaltim). Attended by 17 participants, the participant feels that the training is “too technical” and there are infrastructure problems in the venue where the internet doesn’t work well during the training. Figure 4 - Bot creation training in Samarinda Even though there are no bot articles created,the previous effort manage to create 203 articles manually. The workshop then continues in February 2016 where we flown in an ex ert in Wikidata nd abett and PyWikibot trainer John Vandenberg.Andy suggested using wikidata to connects Wikipedia geographic articles and OpenStreetMap data. Figure 5 -Wikidata workshop with Andy Mabett Sumana Harihareswara from Changeset Consulting LLC were brought in to assess the technical work done by Wikimedia Indonesia on the Pywikibot software project and three training manuals WMID has developed. 2. Summary of Activities This section is a summary of the activities that have been implemented during the reporting period. It will relate to agreed indicators (if any) and compare with the original proposal or last annual review.
  14. 14. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 14 Assesment result of the activities The assessment report stated that Wikimedia Indonesia has “Significantly improved Pywikibot, making it more reliable and adding features that let users use it more securely and through a much easier interface.“ This work made possible through long-term architectural and workflow im ro ements Wikimedia Indonesia made the project more sustainable and laid the foundation to successfully attract more users and developers in the future. Wikimedia Indonesia has written manuals that teach users how to edit Wikipedia manually and how to use Pywikibot to easily make large, wide-spanning improvements to Wiki edia.W ile odules II and III are not et finali ed t e are nearly complete,and it is already clear that the manuals fill a ga and a need for t e Ba asa Indonesian communit for independent study,in-person workshops,and classroom education. They use illustrations and examples as well as prose to explain what the user should do and the context behind the instructions. In a code review for improvement,Wikimedia Indonesia’s volunteer consultant, John Vandenberg’s work as a code reviewer has been crucial to the improvements Pywikibot. Over the past two years he consistently contribute to the project’s overall health. He collaborated with other skilled developers such as valhallasw (Merlijn van Deen) and xZise to maintain the project, and took the lead to troubleshoot and track down defects https://lists.wikimedia.org/ pipermail/pywikibot/2014-August/008996.html so that other developers could more easily address them. A Wikimedia technical community dashboard http:// korma.wmflabs.org browser scr-contributors. tml s ows that Vandenberg is a code review leader not only within Pywikibot but across all Wikimedia-related technical projects. He ranks # 20 of 348 maintainers, having issued 1283 code review approvals that led to code merges. Vandenberg also provides advisory review and vetoes consistently, and ranks # 15 of more than a thousand contributors, having performed 9393 code review actions. 2. Summary of Activities This section is a summary of the activities that have been implemented during the reporting period. It will relate to agreed indicators (if any) and compare with the original proposal or last annual review.
  15. 15. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 15 Result Throughouttheproject,startingonJanuary2015toFebruary 2016, both Wikimedia Indonesia and Humanitarian Open Street Map had conducted 35 events consist of launching, workshops, trainings, competitions, press conference, MoU signing, and awarding ceremonies. These events are being supported by nine other institutions and attended by total of 700 attendees. The effort also being co ered b fi e national and local media some of them are featured stories. Figure 6 - News coverage on Manual I launch Instructional Toolkit I, “How to Edit Wikipedia Manually” deployed on February 27 and reviewed by a group of Wikipedian. On March 31 the toolkit was tested with 35 public participants in Goethe Institut. Instructional Toolkit II, “PyWikiBot” deployed on August 2015. Both of the guide is accessible online (in Indonesian) via Wikimedia Commons https://commons.wikimedia. org/wiki/File:Modul_Pelatihan_1_-_Berkontribusi_di_ Wikipedia_Bahasa_Indonesia.pdf Instructional Toolkit for OSM by HOT OSM deployed in February 2015 and tested with 16 students from three different University during the training session in Samarinda, Kalimantan. From then on this instructional kit has been widely use in other trainings and competitions in Jakarta, Samarinda, Kalimantan. This guide is accessible online (in Indonesian) at http://openstreetmap.id/panduan-osm/ Joint InstructionalToolkit forWMID and HOTOSM Manual III has a complete coverage of using PAWS, including screenshot and explanatory text. Training was conducted with 17 participants, toolkit available online https://www. mediawiki.org/wiki/Manual:Pywikibot/PAWS Additionally HOTOSM and WMID also agree to use wikidata to connect spatial data and informational data. 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  16. 16. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 16 Result One of the unforseen result is Wikimedia Indonesia failed to tie in coo eration wit local O in alimantan. Before t e ro ect we belie e en ironmental O s ould have data on endemic animals and plants in Kalimantan. Mongabay end up not having resource to help with the data. Our perception was they are more “campaign ready” compares to educate local using hard data. The intent of working wit local O is to make sure bot article creation could be deployed to make automatic articles about endemic species.This is one of the agreed indicators that didn’t happened. Figure 7 - Meeting with East Kalimantan Goverment Secondly, are Wikimedia and HOT OSM meetings with ro ince le el go ernment officials. e eral meetings took place, but the effort still has no “buy-in” from provincial level government. However, on the national level, when Kalimantan mapping effort being showed to Indonesian Language Body and Creative Economy Body, the effort of combining spatial data and informational data interest them. Creative Economy Body support our effort in getting the grant for technology improvement from Ford Foundation, and the Language Body give a grant toWikimedia Indonesia to create an independent site illustrating languages vitality in Indonesian region as a cooperation between language body and Wikimedia Indonesia. The two sucesses are unforseen result and a different successful indicator for the organization. This is one of the non-agreed indicators that happened to be successful Other result after the project ends are we end up with data comparison, not only before and after data impact, but also data fromWikipedia,government,and Open Street Map.With all this data, our next question would be, how many did this data cross over? So we need to be creative in mashing up large amounts of data in order to give a meaning and be useful to the locals.We created OpenKalmap data mash up layer: http://openstreetmap.id/kalmap/ 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  17. 17. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 17 Below are result and indicators as stated in the contract. 3.1. Numbers of Wikipedia Page created by participants trained Before the project roll out, we provide baseline that stated that Indonesian language Wikipedia had 7.895 articles. During Wikipedia writing workshop, the 15 participants in total created 205 articles, 105 articles subject are about Kalimantan in Indonesian language Wikipedia.The articles vary from article regarding culture, tourism and geography. After the project roll out, the numbers of article for geography about Kalimantan becomes 7,924 articles, one of the reason is because geographic articles for Kalimantan already complete, as in, they exist but with very little information. “Wikipedia writing workshops shows that a lot of local participants are more interested in writing about culture, arts, and history comparing to writing geographic articles. Surprisingly to us, these are the most widely read articles as well“ For the participants culture, arts, and history are priority article, more than writing about geographic articles. The readership (demand) showing the trend of readership for cultural articles are higher than readership for geographic articles. Total number of articles created organically vs unorganicaly before and after the project roll out are as follows: 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  18. 18. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 18 3.2. Numbers of Wikipedia page view of the pages created The 205 article created received 241,029 readership after seven months creation.The articles manually created by 15 artici ants were di ided into fi e categories alimantan culture, Kalimantan traditional art, Kalimantan history, Kalimantan geographic articles (including place of interest/ tourism/ village/ street), and general encyclopedic articles. rom demand ers ecti e out of fi e categories istor and traditional art received most readership. The highest are Kudungga with 60 view per day and second highest is Tari Gong (traditional dance) with 54 view per day.While geographic articles readerhip vary from 2 to 7 views per day. Participants also created various encyclopedic articles unrelated to Kalimantan, however the page view of this articles score the lowest among other categories. “The most challenging part in creating Kalimantan subject articles inWikipedia by locals were the absent of reference online regarding their surrounding “ Often the source need to come from a blog, which is not a alid source and can not be erified to a is cal book that take time and transportation expense to get – if they can get one at all. 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  19. 19. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 19 3.3. Numbers of Edits in OpenStreetMap made by trained participants For Open Street Map one of our indicator of success are the workshops creates 1.500 object in Kalimantan.When the result are in, the effort had mapped 16,157 buildings in OpenStreetMap, compares to 1,476 buildings before the project started. The chart above also illustrate what the competition in Eastern Kalimantan for data addition compares to no competition in South Kalimantan. The data in showed 2,229 buildings (Jan’15) to 5,759 building (March’16) for South Kalimantan. Compares to 1.476 (Jan’15) to 16.157 buildings (March’16) in Eastern Kalimantan. Indicator of success for June 2015: Instructional toolkit for O a e been finali ed after tested in early February 2015 in Samarinda city.This toolkit has become a standard for any OpenStreetMap activities such as training materials and mapping competition in alimantan. T is toolkit for t e first time used in OpenStreetMap Training in Samarinda at 22-24 June 2016 and continue to be used for other activities. Indicator of success July - August 2015: OpenStreetMap mapping competition successfully held in Samarinda and Tenggarong city. This competition created collected 5,025 buildings and 140 buildings for 4 weeks activity (July-August 2015) into OpenStreetMap. Those buildings can be s ecified as: • 676 important objects whose have references link • 2,382 objects which considered as an important facilities but not really have to be made an article in wikipedia. T ese ob ects suc as kindergarten ost office olice office fire station and local go ernment office u to sub sub district level. 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  20. 20. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 20 Those number of objects come from 2 cities in East Kalimantan. They are Samarinda and Tenggarong City. The winner of the competition comes from Tenggarong and successfully complete the city public facilities in OpenStreetMap. Beside competition activities that been held during July - August 2016, in August, our local partner in Kalimantan map West Kutai District, East Kalimantan Province as well. From that activity, our local partner successfully map about 226 objects high quality data and 542 other objects. Here are numbers of building on OpenStreetMap in East Kalimantan before and after competition: 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  21. 21. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 21 3.4. Previous or potential problems and mitigations Based on the comparation before and after mapping activity, These are some screenshoot of OpenStreetMap in some area before and after competition; Figure 8 - Before and AfterTenggarong City in OpenStreetMap Indicator of success October 2015: Figure 9 - Before and After Samarinda City in OpenStreetMap In October 2015, closing ceremony of the competition granted three winners for OpenStreetMap Mapping Competition. Local media were present and reported the events. Our mapping activity become headline in a local newspaper, Tribun Balikpapan following closing ceremony and also published online for metro tv online news.1 Figure 10 - News coverage on OpenStreetMap competition on Balikpapan 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  22. 22. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 22 Indicator of success January 2016: At the same time closing ceremony for competition in Samarinda and Tenggarong, OpenStreetMap launches the second competition for the city of Balikpapan. The effort become 4th city/district that been mapped so far from Open Content in Kalimantan Project. Added into OpenStreetMap as a result of Balikpapan competition were 1072 objects, detail as follow: • 159 important objects with reference covering school, os ital lace of wors i and go ernment office u to district level • 913 objects considered important facilities covering ob ects suc as kindergarten ost office olice office fire station and local go ernment office u to sub sub district level. Figure 11 - Before and After Balikpapan City in OpenStreetMap Figure 12 - Before and After Balikpapan City in OpenStreetMap In addition to these data being added to OpenStreetMap, we realize that our effort to do mapping was not complete until our local partners learn to download the it and create the data compiled to become a map custom to their need.It means HOT OSM need to taught the local partner how to use QGIS, a free and open source Geography Information System. 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  23. 23. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 23 Indicator of success February 2016: or t is final re ort we would like to ig lig t t e achievements two our local partners in Samarinda. Arief a man recruited in o ember 2014 as single andedl make 287,269 changes. Figure 13 - OSM Statistic of Arief (local partner) about his mapping activity HOT OSM trainedArief to use QGIS and and as illustrated below were the map downloaded and printed to be handed to Samarinda city government. Figure 14 - Goverment offices of Samarinda Kota Sub District Map Figure 15 - Goverment offices of Samarinda Kota Sub District Map rinted ma s owing go ernment offices and facilities in Samarinda. 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  24. 24. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 24 Indicator of success February 2016: The map given to the government below only showed go ernment offices. Howe er once our local artners know how to “clean up” the map using QGIS, each map printed could show different layer, such as educational facilities, map of health facilities, map of public facilities – depending to your need, or depending to what the locals need, made by a them. Figure 16 - Handover maps from our local partner to Goverment of Samarinda City Our second project recruit is Suyono Darul. Recruited in July 2015 after Wikipedia writing competition. Suyono was in second place, and eager to help the effort. His main interest is to “spread the knowledge” on how to map. After becomes comittee in two mapping competition for universities in Samarinda and Balikpapan, Suyono was confident t at t e ig sc ool students need to know t is skill as well. Figure 17 - Suyono (local partner) and his training with high school students The effort funded Suyono to do a highschool mapping works o and com etition in utai arta egara. e en highschool participated and send19 representative to do the workshop. During the three day workshop training they were taught split way, combine way, and join note way to edit and was shown how to use building tool to create a building shape in maps. 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  25. 25. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 25 Indicator of success February 2016: To collect data the high school participants uses OSM tracking in their android cellphone as substitute of GPS. After data collection and survey they start mapping using JOSM,putting correct symbols on map and the appropriate presets based on OSM standard. On the fourth day, out of 19 representative attended the workshop, eight students decide to compete. As a result 1,086 buildings and 16 street were created during a one month competition.The mapping activities covering Summer Sari village,Mekar Jaya & Segihan village,Manunggal Daya village. Puan Cepak, Sabingung, Muara Kaman Ulu- ilir, Sidomukti, Panca Jaya, Bunga Jadi. Among this data are 115 high quality data point with reference. These data are validated by two comitees,Arief & Suyono, by doing a remote alidation and field alidation. Figure 18 - Mapping activities on Kalimantan After two mapping activities we bring in Yayasan Bumi & Centre of Borneo Environmental Remote Sensing in Samarinda to evaluate the quality of the geospatial data generated through OpenStreetMap program.They evaluate two cities: Samarinda and Balikpapan in East Kalimantan Province.Yayasan Bumi work by comparing the geospatial data from OpenStreetMap with geometrically corrected very high resolution satellite imagery and the coordinate measurements conducted in t e field using and eld lobal Positioning System (GPS) receiver.When the result in, “The accuracy are good. some errors were noted, but for most cases the percentage of point and line, and polygon features never exceeds 29% “ The metrics of polygon also show similar pattern and most of t em were not significantl different. a asan Bumi also suggested that OpenStreetMap volunteer to regularly update point and polygon geospatial dataset if possible. (Please refer toYayasan Bumi document in the appendix of this report for further details) 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  26. 26. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 26 3.5. Measurement of usages of Wikipedia pages overtime relative to content added As previously mentioned writing competitions by local participant has manage to create 205 article which received 241,029 readership after seven months creation. The articles manually created by 15 participants were di ided into fi e categories alimantan culture alimantan traditional art, Kalimantan history, Kalimantan geographic articles (including place of interest/ tourism/ village/ street), and general encyclopedic articles. rom demand ers ecti e out of fi e categories istor and traditional art received most readership. The highest are Kudungga (history) with 60 view per day and second highest is Tari Gong (traditional dance) with 54 view per day.While geographic articles readerhip which rates 3 per day. Participants also created various encyclopedic articles unrelated to Kalimantan, however the page view of this articles score the lowest among other categories using seven months readership pageview analysis. nfortunatel our effort in obtaining data from local O for endemic s ecies of flora and fauna fails. Howe er t e software to do automatic articles creation was successfully improved based on third party assesment. 3. Result Describe the result(s) foreseen and unforeseen of each activity. - be as specific as possible - relate to agreed indicators (if any) - compare with the results that were planned in the proposal or annual review - explain major differences
  27. 27. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 27 4. Direct beneficiaries, participants, visitor of websites, audiences both men and women for rural and non rural. • Total outreach effort from January 2015 to February 2016, out of 35 events, supported by nine other institutions, our outreach attended by total of 700 attendees. • We conducted fi e com etitions for Wiki edia writing and OpensStreetMap mapping in universities and ig sc ools. Out of 1 winners form t ese fi e competitions male and female incentive recepients direct beneficiaries com aration: • Four were mapping competition, comparation rural vs non rural area mapped 4. Outreach Specify the number of direct beneficiaries, participants, visitors of websites, audience, et cetera (outreach). - for both men and women - for rural and non rural
  28. 28. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 28 5. Staff & Skill Wikimedia Indonesia recruit two wikipedian volunteer as part time consultant for this projects.Vanessa Winona and Arief Rahman. While previously Arief Rahman have been an amateur OpenStreetMap contributor, the training he received by HOT experienced trainer enables him to be one of the most active mappers in East Kalimantan and to print his own map using customized layer using GIS. This project also enables him to learn advanced technique in OSM taught by Harry Machmud as the lead trainer for OpenSeaMap and OpenStreetMap in Kalimantan. Even though Harry is not new to training participants for OpenStreetMap, OpenSeaMap as project brought Harry to a different level of training experience.The opportunity to learn OpenSeaMap, and created a training materials for OpenSeaMap, has made Harry see new possibilities in mapping. Suyono Darul was recruited on July 2015 as trainer commiteeforOpenStreetMapmappingcompetitionafterhe completed second place in Wikipedia writing competition. Suyono admit that he learn alot during the project and contribute to the rural knowledge for OpenStreetMap by initiating mapping competition for highschool students. 5. Skills Describe in what ways the skills of the staff were upgraded during the reporting period.
  29. 29. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 29 6. Lesson learn and conclusion from implementation of the project • Local government was not ready to buy citizen introduced effort, however national level government was able to see the possibilities and did a different buy in instead. Currently the buy in are for for creative economy improvement and languages dataset in the form of direct grant or support for grant. • Before the project started, our baseline shows that there are more Wikipedia geography pages articles about Kalimantan than there are data in OpenStreetMap.Therefore the effort of importing data from OpenStreetMap to Wikipedia didn’t happened because the software will have to overwrite an existing article in order to create an article that didn’t exist. So eventhough the software for article creation has been improved the project need to test it using unavailable article such as Kalimantan endemic species (which not available in Wikipedia) rather than testing it using geography articles that already exist. Hence, we switch the approach using wikidata to put Wikipedia articles link in OSM instead and create Open Kalmap for mesh data layer of Wikipedia articles, OSM data, and local government data for health and education facilities. Figure 19 - Map of Samarinda • T e local O focused on en ironment in alimantan was more prepared to do campaign instead of building knowldege using data. • The local participants need different kind of “incentive” to contribute, while gadget and monetary incentive work in one area, it didn’t on another area. 6. Conclusion Describe conclusion and lesson learn.
  30. 30. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 30 Conclusion: • We learn that the key success is training the right kind of people for trainers, and we can select this individuals through wikipedia writing competitions. Once trainers are selected, more intensive training need to be done to upgrade their skill and knowledge. After that the project becomes highly replicable elsewhere through the same methodology. • We also receive a perception that the student more interested in motivational speaker type of workshop rather than an intense technical ones from a visiting visitors. From the previous report of university inter- est to Open Street Map manual, we realize that there are possibilities that the modul could be the source of teaching subject in universities for practical knowledge. However it requires further adaptation. What we do on this project period is make sure local resource avail- able to do training to the local after we did training for trainers.
  31. 31. PERKUMPULAN WIKIMEDIA INDONESIA Centerflix Boutique Office • Jl. Danau Toba no. 104 Bendungan Hilir • Jakarta 10210 • Indonesia E-mail: info@wikimedia.or.id • Website: wikimedia.or.id 31 List of Appendix: 1. Audited Financial Report 2. March 2016 report on Wikimedia Indonesia’s work pursuant to the Making All Voices Count project by Sumana Harihareswara 3. Evaluation of OpenStreetMap Indonesia Geospatial Data: Samarinda and Balikpapan
  32. 32. March 2016 report on Wikimedia Indonesia’s work pursuant to the Making All Voices Count project by Sumana Harihareswara Changeset Consulting LLC
  33. 33. 2 Table of Content Overview 3 Background 3 Pywikibot 3 Training Manual 4 Activity 4 Pywikibot 4 Bugfixes 4 New Features 5 Code Review 7 Integration 7 PAWS 7 PyPI 8 Documentation 8 Planning and Release 9 Training Manual 10 March 2016 Assesment 11 Pywikibot 11 Training Manual 11 Sugested Future Work 12 Pywikibot 12 Training Manual 13 Conclusion 14 About Sumana Harihareswara 15
  34. 34. 3 Overview This report provides an overview of the technical work done by Wikimedia Indonesia on the Pywikibot software project, and an assessment of the three training manuals WMID has developed. Wikimedia Indonesia has significantly improved Pywikibot, making it more reliable and adding features that let users use it more securely and through a much easier interface. Through long-term architectural and workflow improvements, WMID has also made the project more sustainable and laid the foundation to successfully attract more users and developers in the future. Wikimedia Indonesia has written manuals that teach users how to edit Wikipedia manually and how to use Pywikibot to easily make large, wide-spanning improvements to Wikipedia. While Modules II and III are not yet finalized, they are nearly complete, and it is already clear that the manuals fill a gap and a need for the Bahasa Indonesian community, for independent study, in-person workshops, and classroom education. They use illustrations and examples as well as prose to explain what the user should do and the context behind the instructions. Background In 2014 Wikimedia Indonesia received a grant from the Making All Voices Count project to add data to OpenStreetMap and batch-create Wikipedia articles using the Pywikibot tool, with a focus on ensuring that the tools and processes developed during the project would be beneficial to other teams wanting to undertake similar tasks. Therefore, Wikimedia Indonesia worked to improve Pywikibot’s code to make it suitable for this use. WMID and Humanitarian Open Street Map also worked together to create a set of training manuals. These manuals help the Indonesian public learn to map and to to write content in Open Street Map and the Bahasa Indonesian language Wikipedia. When the project began in early 2014, Indonesian users were unsupported by software tools and by existing training manuals. Pywikibot When Wikimedia Indonesia began this project in April 2014, the Pywikibot developers (working as volunteers on their own time) had been working on Pywikibot for upwards of seven years, and had recently been slowed down by a difficult tooling migration (switching from Subversion to the less familiar and more complex Git versioning system). Pywikibot, while written in a popular programming language (Python), did not take advantage of this opportunity to connect with and leverage the Python community; as of early 2014 Pywikibot was not available in the most popular Python package repository, PyPI, for discovery and automated updating by Python programmers https://lists.wikimedia.org/ pipermail/pywikibot/2014-March/008599.html . Pywikibot’s rewrite from “compat” to “core” (see the Activity section for further explanation) had been ongoing since 2008 https://lists.wikimedia.org/pipermail/pywikibot/2008- May/002944.html . Since Pywikibot core only robustly supported the version of MediaWiki that ran on Wikimedia sites such as Wikipedia, and it was unclear how well Pywikibot supported other wikis that ran older versions of MediaWiki, the user community was confused and many bot makers stuck to the old edition. The Pywikibot maintainer team had been discouraged for years by the user community’s lack of interest in switching to the new edition https://lists.wikimedia.org/pipermail/pywikibot/2012-April/007475.html .
  35. 35. 4 In addition to this unclear support matrix, the Pywikibot maintainers were held back in their progress by a lack of clear and agreed-upon goals for their next major release, Pywikibot 2.0. The work towards 2.0 had been progressing slowly for years and volunteer developers had drifted in and out of the project. Training manuals In 2014, the “How Wikipedia Works” training manual https://en.wikipedia.org/wiki/How_ Wikipedia_Works existed in English, but no translation into Bahasa Indonesian existed. Developers had written an English-language guide explaining how to use Pywikibot’s pagefromfile.py https://www.mediawiki.org/w/index.php?title=Manual:Pywikibot/ pagefromfile.py&oldid=809096 and a somewhat confusing slidedeck helping developers learn how to use Pywikibot to contribute to Wikidata https://www.mediawiki.org/wiki/ File:Bots_hackathon_2013.pdf , but there was no comprehensive manual usable by non- coding beginners to instruct them in using Pywikibot to contribute to Wikipedia. And, as PAWS did not yet exist, no manual for it yet existed, either. Activity Pywikibot Wikimedia Indonesia’s John Mark Vandenberg worked on fixing defects (bugs), adding and mentoring new features, reviewing other developers’ code, improving Pywikibot’s integration with other services, writing documentation, and leading the planning and release process. Bugfixes Vandenberg started on the Making All Voices Count work on 24 May 2014, submitting a “patch” (fix) to a bug on 24 May 2014. The other “maintainers” (technical leaders) of Pywikibot reviewed and accepted his suggestion the same day. He continued to fix hundreds of individual defects to strengthen Pywikibot and closely related components, such as mwparserfromhell (https://github.com/earwig/mwparserfromhell/commits?author=jayvdb), which are reusable by and useful to the larger developer community. In addition to individual bugfixes, Vandenberg created mechanisms for Pywikibot to protect itself from developing certain types of bugs in the future (a sort of immune system). As an illustration: one drain on Pywikibot maintainers’ energy was the number of support requests they received for an edition of Pywikibot (known as “compat”) that lacked many newer features but worked well with very old versions of MediaWiki. Wikimedia runs on MediaWiki; it always runs on the newest version, which is updated more than a dozen times each year. But it’s often useful to run (and test) pywikibot against other wikis, including local wikis which have information to gather and then reuse in contributing to Wikipedia. Those installations of MediaWiki are sometimes older -- the software isn’t as up-to-date. To help reduce the maintenance cost of supporting compat, Vandenberg worked to remove the barriers that kept Pywikibot users from moving from “compat” to the well-supported “core” edition of Pywikibot. For instance, there were many bugs running Pywikibot on older versions of MediaWiki. He fixed or helped fix those bugs, so that Pywikibot now officially
  36. 36. 5 supports versions of MediaWiki as far back as 1.14 (which was released in January 2009). Automated tests make it possible for developers to move faster and write more innovative code, because the developers know that the automated tests will speak up to alert them if their new improvements have inadvertently introduced certain kinds of bugs. A “test suite” is a set of automated tests that check whether the defects emerge when the tool is used in various ways and contexts. Pywikibot needed an investment in future robustness, and thus under Vandenberg’s leadership, Pywikibot’s test suite grew from 4073 lines long on May 24, 2014 to 26803 lines as of 28 February 2016. For instance, the new tests include regression tests that ensure Pywikibot core continues to work properly when interacting with old versions of MediaWiki. The test suite is also protecting Pywikibot contributors from potential problems on multiple sites. Before Wikimedia Indonesia’s work started, the test suite only ran tests on English Wikipedia and on Wikidata. The suite now runs on WMF Beta labs, Wikipedia Test, and Wikidata Test (a set of official Wikimedia Foundation staging servers that help the technical community find and fix bugs before they can affect ordinary users). And the test suite also runs on Wikisource, Wiktionary, Wikia, and Musicbrainz, thus helping free knowledge contributors whose interests include sites beyond Wikipedia and Wikidata. In addition, Pywikibot’s codebase now enjoys automated “continuous integration” testing on the Windows platform via an Appveyor server. This adds a new layer of protection against future bugs, since many Pywikibot users use Windows but the Pywikibot maintainers, for the most part, use other operating systems and are thus less likely to find Windows-related bugs in the ordinary course of their work. To prevent future bugs and to make it easier for Pywikibot’s maintainers to review and manage code, John Vandenberg also created and improved multiple standalone tools that act as automated gatekeepers for new code that developers write, checking both for substantive correctness and for stylistic consistency. He improved a document style checker (https://github.com/PyCQA/pydocstyle/commits?author=jayvdb) and contributed to pyflakes (https://github.com/pyflakes/pyflakes/graphs/contributors), which provides better static code checking of new contributions. He created flake8-putty (https://github. com/jayvdb/flake8-putty) and rewrote flake8-print (https://github.com/JBKahn/flake8-print/ commit/8ecf72eff417cdc5cadbf956fd9a5b33a0800aa1) so that these automatic checkers could be more easily quieted when they were falsely flagging innocuous code as wrong. New Features Since 24 May 2014, in terms of writing and revising code, Vandenberg has been the top contributor to Pywikibot. I used the version control tool Git to count the number of improvements Vandenberg has made to the Pywikibot codebase with the command: git log --oneline --since=2014-05-24 --author=”Vandenberg” | wc -l and found that he has authored 763 “commits” that have been merged into Pywikibot since 24 May 2014. This is 185 more commits than any other Pywikibot maintainer (Fabian Neundorf, whose tally for the same period is 578 commits). For a visual representation of this technical leadership, see https://github.com/wikimedia/pywikibot-core/graphs/contributors, which counts Vandenberg’s number of commits slightly differently (758 commits) and which subdivides his contributions into 39,292 lines added and 15,268 lines removed.
  37. 37. 6 Some of these features stand alone and improve the experience of a Pywikibot user -- for instance, Vandenberg improved a user’s ability to use Pywikibot while using a proxy to access the Internet https://lists.wikimedia.org/pipermail/pywikibot/2014-June/008767.html , as many privacy-conscious users do. Some new features aid the rest of the Wikimedia community in collaborating with Pywikibot users -- as an example, Vandenberg reviewed and publicized code https://gerrit.wikimedia.org/r/#/c/147381/ that helps Pywikibot users specify a “user-agent” label their bots use in their interactions with wikis https://lists.wikimedia.org/ pipermail/pywikibot/2014-August/009015.html . Thus, if a user’s bot accidentally causes difficulties through excessive editing or by making changes that have unintended effects on site performance (perhaps because of a misconfiguration of a cache or an interaction with another user’s script), website administrators can find the user’s contact information in their access logs and contact her to request modifications. Perhaps more importantly, Vandenberg worked on long-term improvements to Pywikibot that serve as the foundation for future user-visible features. One large improvement of this type was changing Pywikibot to run on a newer version of Python. As the programming community converts from Python 2 to Python 3, the newer version of Python is now the default version installed on the most popular distributions of Linux (Debian, Ubuntu, and Fedora), and Pywikibot will now work smoothly on those machines. Although future proofing Pywikibot is a benefit, it’s not the main reason the team switched to Python 3 when it did. Python 3 is much more suited for multilingual applications. Pywikibot had been written in Python 2, which by default assumes that “strings” (text) are in the Latin characterset, as English is. In contrast, Python 3 uses Unicode strings by default, which means that it makes fewer assumptions about the language of strings that the program takes in or emits. This reduces the frequency of bugs in Python programs that take in or output text in Bahasa Indonesian and other non-English languages, especially programs interacting with international and interlingual sites such as Open Street Map, Wikidata, and Wikipedia. In a similar vein, Wikimedia Indonesia’s work in adding OAuth support to Pywikibot served multiple goals. The Wikimedia servers, as of a few years ago, support OAuth -- a way for a user to only have to authenticate themselves once to the Wikimedia servers, then delegate that authentication to multiple applications (such as their Pywikibot scripts and bots). This is more secure than passing a password around multiple times. In order to support OAuth, Pywikibot had to switch from using the httplib2 library to using the much better-supported and broadly-featured Requests library. Vandenberg reviewed code by a volunteer, Justin, who worked to entirely switch Pywikibot’s HTTP request handling to use Requests https://lists.wikimedia.org/pipermail/pywikibot/2015-June/009271.html . The version of Pywikibot that has completely switched to using Requests is still in progress, but the OAuth support that has been implemented (https://phabricator.wikimedia.org/project/ view/1246/) was critical, as PAWS requires it. (See the “Integration” section below for more on PAWS.) Volunteer VcamX successfully added OAuth support, his code mentored and reviewed by Vandenberg. Vandenberg mentored multiple interns and reviewed code by them while working on Pywikibot during the grant period, and their progress included long-term improvements to the tool. Alexander Jones https://www.mediawiki.org/wiki/Google_Summer_of_Code_past_ projects#Implement_Flow_support_in_Pywikibot successfully implemented Pywikibot’s ability https://www.mediawiki.org/wiki/Manual:Pywikibot/Flow to read from and write to the new MediaWiki discussion system, “Flow”. Pywikibot users often we use bots to post to discussion pages to post notifications of what bots have done or noticed https://www.
  38. 38. 7 mediawiki.org/wiki/Flow/Bots . As Wikimedia sites convert from the old “talk page” model to the new Flow discussion boards, bots need to be able to post to them. After Jones’s 2015 internship https://phabricator.wikimedia.org/project/view/1247/ , Pywikibot can now load Flow boards, create discussion topics, reply to existing comments, lock and unlock topics, and fetch the revision history of posts and topics. https://www.mediawiki.org/wiki/Google_ Summer_of_Code_past_projects#Implement_Flow_support_in_Pywikibot Vandenberg also mentored a woman, Priyanka https://www.mediawiki.org/wiki/ User:Prianka/Pywikibot_:_Compat_to_Core_Migration/_Progress_Report , in a 2015 internship project as part of the Outreachy initiative, which brings new women and people from other underrepresented groups into the open source software community. Priyanka aided in rewriting bots (“scripts”) from the old “compat” edition so that they were available in the new, better-maintained “core” edition. Starting in 2014, Vandenberg worked on a Pywikibot script for adding new articles to Wikipedia using Open Street Map data. A review of the script’s functionality and effects in December 2014 gave him feedback that he has been integrating into the in-progress tool. It successfully asks the OpenStreetMap API for data on institutions in specific regions (such as Samarinda) and outputs data that would be of interest for reuse on Wikimedia sites, and -- in order to avoid violating the notability criterion -- it limits its dataset to institutions that have websites and that serve more than 150 people. Code Review As I discussed in the “Features” section above, Vandenberg’s work as a code reviewer has been crucial to the improvements Pywikibot has seen over the past two years, and to the project’s overall health. He collaborated with other skilled developers such as valhallasw (Merlijn van Deen) and xZise to maintain the project, and took the lead to troubleshoot and track down defects https://lists.wikimedia.org/pipermail/pywikibot/2014-August/008996. html so that other developers could more easily address them. A Wikimedia technical community dashboard http://korma.wmflabs.org/browser/scr- contributors.html shows that Vandenberg is a code review leader not only within Pywikibot but across all Wikimedia-related technical projects. He ranks # 20 of 348 maintainers, having issued 1283 code review approvals that led to code merges. Vandenberg also provides advisory review and vetoes consistently, and ranks # 15 of more than a thousand contributors, having performed 9393 code review actions. Integration Pywikibot’s newfound stability also makes it possible to integrate it with other software and build new gateways to using it. PAWS Starting in November 2015, Wikimedia Foundation developers (especially Foundation engineer Yuvaraj Pandian) have been working with the Pywikibot team https://phabricator. wikimedia.org/project/profile/1648/ to build an easier way to run bots on the Foundation’s infrastructure. PAWS (Pywikibot: A Web Shell) is not yet polished enough to release and
  39. 39. 8 publicize widely, but it is polished enough that Vandenberg has been able to use it in workshops he has led in February 2016 in Jakarta. Collaborating with Yuvaraj Pandian at the Wikimedia Foundation, Vandenberg was able to make it far easier for nontechnical people on a variety of computers to try out Pywikibot, and to use it on a more continuing basis. The PAWS platform provides a known uniform environment for workshops and other educational environments. This is particularly salient to Windows users; most non-coders who have personal computers use the Windows operating system, and installing Pywikibot on Windows computers has long been difficult. Wikimedia Foundation’s infrastructure (Tool Labs) has consistent support from the Foundation’s operations team; my experience with Tool Labs has been that it is far more robust and reliable for bot hosting than is the average developer’s personal server or home computer. PAWS also leverages an additional component that is rapidly becoming a standard among scientific computing researchers: Jupyter, a browser-based interface to create, share, and run code in multiple languages. I spoke in late February with Nathaniel Smith, leader of NumPy and member of the steering council of SciPy.org. Like NumPy, Jupyter is fiscally sponsored by NumFOCUS, a US nonprofit, and is being actively developed by engineers at the University of California at Berkeley’s Institute for Data Science. These scientific computing tools enjoy hundreds of thousands of users and a thriving ecology of data science conversations and conferences. By working with Jupyter and harnessing its momentum, Vandenberg has set PAWS on a solid foundation, more futureproof than a similar application he wrote himself would be. PyPI The Python Package Index, or PyPI, is where the Python community looks for software, like a central phonebook or bulletin board. The work that Vandenberg and other contributors performed in listing Pywikibot on PyPI maximizes its discoverability to people who had not previously considered contributing to Wikimedia with automated tooling. The team has now released four release candidates to PyPI (https://pypi.python.org/ pypi/pywikibot), making it easy to install via the standard “pip” package installer. They immediately started receiving positive feedback from the Python community, as Python developers reflexively look on PyPI for relevant software. https://lists.wikimedia.org/ pipermail/pywikibot/2015-May/009269.html Documentation A successful software project needs multiple kinds of documentation. In Pywikibot’s case, developers of the framework itself and of bots using the framework need a systematic reference guide detailing how every function and method works. Vandenberg was heavily involved in setting up online documentation for Pywikibot during the grant period. He used an industry-standard approach called “docstrings” -- next to the declaration of a function, below its signature, programmers write a brief comment explaining the function. Vandenberg built a custom plugin called sphinx-epytext that provides basic support for a certain style of docstrings in the Sphinx automatic documentation generator
  40. 40. 9 in automatically generated documentation https://github.com/jayvdb/sphinx-epytext . Following that, he set up Readthedocs https://lists.wikimedia.org/pipermail/pywikibot/2014- November/009135.html the first online documentation for version 2.0rc2. http://pywikibot. readthedocs.org/en/latest/ While better than the status quo ex ante, the new documentation site was not on a Wikimedia server, which made it harder for the Wikimedia community to find and trust. Thus, during an in-person hackathon with other Pywikibot maintainers, Vandenberg and volunteers set up a Wikimedia-hosted documentation site https://doc.wikimedia.org/pywikibot/ which is now automatically updated after every revision to the code. Pywikibot also requires manually written guides to the system as a whole, to scripts (helping users understand what pre-written bots to use and why), and to other aspects of the software’s architecture and future. Vandenberg exercised leadership in documenting scripts https:// lists.wikimedia.org/pipermail/pywikibot/2014-August/009001.html and in documenting the migration from compat to core https://lists.wikimedia.org/pipermail/pywikibot/2014- June/008801.html , and wrote and updated several other pages of documentation, such as the English-languages PAWS guide. Planning and Release Successful software projects require both execution and alignment: individual contributors must provide new work product, and they must agree on the goals and workflow of their own and each other’s work. With Wikimedia Indonesia’s support, Pywikibot’s developer community is now much more aligned on the project’s schedule, architecture, and goals, which enables individual volunteers to work more confidently. Vandenberg contributed greatly to the planning of a proper release management schedule for pywikibot, to ensure that future versions come out on a reliable cadence. In May 2015, the first release candidate (“RC1”) for Pywikibot 2.0 came out https://lists.wikimedia. org/pipermail/pywikibot/2015-May/009268.html , after years of delays. Several release candidates followed that year, including 2.0rc4 in December 2015 https://lists.wikimedia. org/pipermail/pywikibot/2015-December/009359.html . The pywikibot team has now feature frozen Pywikibot 2.0 (which is to say, it has decided that no new major functionality will come into that version); it had been called RC5, and is nearly ready to be released https://phabricator.wikimedia.org/T121948 . The 2.x series https://www.mediawiki.org/wiki/Requests_for_comment/pywikibot_2.0_packaging will iteratively improve and serve as a reasonably smooth upgrade of Pywikibot for compat users to upgrade. Vandenberg and the other maintainers have already started work https:// lists.wikimedia.org/pipermail/pywikibot/2016-January/009382.html “on the 3.0 release which uses requests and supports OAuth.” The more innovative 3.0 series includes new architecture that is more futureproof, performant, and uses industry standard technologies. Building a roadmap requires many decisions, and Vandenberg has led or participated in those conversations several times over the past two years: in deciding on minimum supported new versions, in changing packaging methods https://lists.wikimedia.org/pipermail/ pywikibot/2014-November/009160.html , in dropping support https://lists.wikimedia.org/ pipermail/pywikibot/2014-June/008855.html for an older version of Python https://lists. wikimedia.org/pipermail/pywikibot/2014-August/009024.html , in supporting multilingual operation better https://lists.wikimedia.org/pipermail/pywikibot/2014-August/009035.html
  41. 41. 10 , in properly managing users’ secrets and security https://lists.wikimedia.org/pipermail/ pywikibot/2014-June/008856.html , in ease of use and installation https://phabricator. wikimedia.org/T100109, and in dozens of other discussions. Vandenberg particularly valued the in-person collaboration among Pywikibot developers attending the Lyon hackathon in the spring of 2015 https://www.mediawiki.org/wiki/Wikimedia_Hackathon_2015 . Vandenberg, xZise, and van Deen collaborated to schedule future work and veto or approve architecture and feature choices. Vandenberg also led the way on more immediate needs that focused other developers’ attention on urgent issues, as when he pointed out https://lists.wikimedia.org/pipermail/ pywikibot/2014-August/008996.html that new changes merged into the Pywikibot codebase had caused new bugs https://lists.wikimedia.org/pipermail/pywikibot/2014-July/008969. html . And he took care of administrative details and participated in time-saving meetings with other groups, as when he set up a category in the group task-tracker https://lists. wikimedia.org/pipermail/pywikibot/2014-July/008966.html , or when he helped reduce duplication of effort among groups of developers writing similar code https://phabricator. wikimedia.org/T97950 . Training manuals Wikimedia Indonesia developed three manuals (“Modules”), some in Bahasa Indonesian and some in Bahasa Indonesian as well as English. • Module I: A100-page manual helping beginners learn how to edit Wikipedia manually -- available chapter-by-chapter in https://commons.wikimedia.org/wiki/Category:Books_produced_by_ Wikimedia_Indonesia as “Pedoman Penyuntingan Secara Manual” (Editing Guidelines Manual). Wikimedia Indonesia wrote this Bahasa Indonesian translation of existing English material (“How Wikipedia Works”), and it is illustrated and polished. One final editing stage awaits before it is finalized. • Module II: a 40-page manual teaching basic usage of pywikibot and pagefromfile.py to contribute to Wikipedia: “Training Module II: Contributing in Wikipedia using Pywikibot: Installation and Implementation Guide” https://docs.google.com/document/d/1K94Gkimc-SGLv q9TKj5kQfS9j2XMsi4cNCzeAFgfDHg/edit?usp=sharing&ts=56ca363b (in Indonesian - “Modul Pelatihan Pywikibot” https://docs.google.com/document/d/1K8w85LOod8ZEhhH3_bG- A9GAzwt8QSwpmWyd2SgWctk/edit?usp=sharing&ts=56ca3670 ). Wikimedia Indonesia wrote this as an original work, both in English and in Bahasa Indonesian. It is finalized and polished and print copies are available. • Module III: a guide to using PAWS (Pywikibot: A Web Shell) as an easier way to use Pywikibot to contribute to Wikipedia (in Indonesian: “Modul Pelatihan III: PAWS (Pywikibot: A Web Shell): A brief overview and usage guidelines” https://docs.google.com/document/ d/1W3QE6cQxkjYX22yDze5Jan-pSwD_NVZVzu5HXFtxvOc/edit?usp=sharing&ts=56ca365a ). John Mark Vandenberg initially wrote the English-language outline and shared it on the Wikimedia technology wiki https://www.mediawiki.org/wiki/Manual:Pywikibot/PAWS ; the Wikimedia Indonesia team translated that outline into Indonesian and expanded it into Module III. Vandenberg will use those improvements to improve the English language version, and Wikimedia Indonesia is revising and proofreading the Indonesian version before finalizing and printing it.
  42. 42. 11 March 2016 assessment Pywikibot Wikimedia Indonesia’s support has significantly improved Pywikibot, in the short term and the long term, for users around the world. The upcoming release of Pywikibot -- intended to be packaged as Pywikibot 2.0 and suitable for general use -- will be a stabilizing force in the community, as currently those users who cannot use compat have had to constantly update their versions without the aid of proper package management in order to keep their bots working. PAWS with Jupyter is a tremendous step forward in usability for Pywikibot and is a gamechanger along the lines of Dreamhacks http://hack.dreamwidth.net/ and Khan Academy. In-browser support, including a workspace where the user can store and edit files such as scripts, makes workshops, cross-platform collaboration and teaching, and other activities much more possible. Wikimedia Indonesia’s John Mark Vandenberg added important features and fixed several bugs in Pywikibot over the grant period -- moreover, his leadership and mentoring stabilized the Pywikibot project and set it on a more solid footing for the future. Because of WMID’s support during the grant period, Pywikibot is significantly closer to being a part of a robust Open Street Map-Wikipedia workflow. Training manuals Training modules I and II with Ichsan Mochtar (WMID board member) and Suyono Darul (Board member of an education institute in East Kalimantan). Module I is very polished in appearance and detailed in coverage, and replicates the long-appreciated How Wikipedia Works manual. Module II is currently not as polished on the sentence/grammar level (in the English version) as Module I. It succeeds at explaining skills and concepts that the target reader does not already have experience with, such as the command line, illustrating the process with relevant screenshots while also providing instructions in the text (which makes it easier for the reader to copy and paste those instructions as needed). It will be reusable outside of Indonesia, but it clearly explains why Pywikibot-assisted editing is ideal for the Indonesian context:
  43. 43. 12 “Do you realize the ease given by Pywikibot? Contributing to Wikipedia without having to open Wikipedia site directly helps us to avoid losing our contribution when internet connection suddenly gets interrupted. If you are an active contributor, you will likely be able to make several articles at once and want to upload all of them to Wikipedia. By using Pywikibot, you can upload all of your articles in one process.” Overall, the instructions are suitable for a user who has edited Wikipedia but who is new to Pywikibot. I note one exception, which should be easy to remedy. Section 4.2 includes a suggestion that looks promising, involving using Excel to organize data and then using Pywikibot and article templates to generate useful articles. But a crucial step is unclear: “By utilizing this feature and data of University in East Kalimantan and South Kalimantan, we will try to make stub article.” Where should the user get the data, and how should she set up the columns to succeed at this exercise? Module III has complete coverage of using PAWS, including screenshots and explanatory text. It successfully explains the background, context, and paradigm the user will need to understand in order to comprehend why she is performing these actions, and helps the user understand how to use PAWS to make automated edits. It will be reusable outside of Indonesia. The prominent gap in Module III’s coverage is that it does not mention Open Street Map nor explain how to use OSM data to enrich Wikidata or Wikipedia. Ideally, in combination with the existing HOT Indonesian OSM manual http://openstreetmap.id/panduan-osm/ , Module III would teach the user to decide on data to reuse from OSM, fetch it programmatically, use Pywikibot to upload that data to Wikidata, and then decide how to enrich one or more Wikipedias with that data. On a grammar and stylistic level, the current English-language draft could use additional work; the prose is comprehensible but not polished. One phrasing choice in the Indonesian version of Module III that could be improved on its own is the use of the word “laptop.” The instructions call the Python 3 interface in Jupyter a “’laptop’ interaktif,” for instance, and refer to the PAWS host server as a laptop. Those components and applications are not laptop computers, and calling them such may interfere with users’ learning after the workshop. Suggested future work Pywikibot The most urgent tasks are: • Polish PAWS to a releaseable state, addressing the issues listed in https://phabricator.wikimedia. org/project/board/1648/query/all/?filter=all&hidden=true • Polish and release Pywikibot 2.0, also known as RC5, addressing the issues listed in https:// phabricator.wikimedia.org/T121948#1892243 • Improve the in-progress tool for OpenStreetMap integration such that it could be used in Training Module III. Currently the code is rough, with insufficient comments and slightly messy coding practices. The developer needs to limit its data gathering to schools, and implement functionality to create the Wikidata items that will form the basis of new Wikipedia articles. Important in the near future but not as urgent: • As the Pywikibot community already is discussing https://www.mediawiki.org/wiki/ Project:Pywikibot/Documentation_RFC , the autogenerated documentation is a good start but is insufficient to their needs. Developers, installers, users, and the larger Wikimedia community
  44. 44. 13 need more documentation, slide decks, videos, in-line and interactive help, and other sorts of documentation. Additionally, the English-language PAWS documentation is sparse and needs more detail, especially on how to use Jupyter. • Wikimedia sites are available in Indonesian, and the MediaWiki software supports that. Pywikibot and Python 3 also support localisation into Indonesian, However, Jupyter and the PAWS interface are currently only available in English. To better serve the Indonesian community, developers should internationalize the Jupyter & PAWS interfaces, changing the software to allow for it to speak to users in multiple languages, and translators should translate the Jupyter and PAWS interface messages into Bahasa Indonesian. Important for the long-term success of this project: • Developers will need to continue to work on Pywikibot, to deliver a 3.0 version that includes OAuth support, and to continue to keep up with MediaWiki’s changing API. As they publish new releases, the Pywikibot maintainers should publish concise release summaries on the announcement mailing list https://lists.wikimedia.org/mailman/listinfo/pywikibot-announce and link to comprehensive release notes. • As the bot community already discusses regularly https://meta.wikimedia.org/wiki/International_ Pywiki_Project , one blocker to the progress of automated improvement to Wikimedia sites is that botmakers find it difficult to share their scripts with each other and with the larger public. This obstacle may be newly soluble with the advent of PAWS; Jupyter lowers the barrier to sharing code with other bot-makers. It may be advisable to integrate Jupyter with Git so as to provide an easier user interface to Wikimedia’s shared code hosting (Gerrit). Collaboration with NumFOCUS or with UC Berkeley’s Institute for Data Science may shed new light on this opportunity. • As these to-do notes indicate https://www.mediawiki.org/wiki/API:Client_code/Evaluations/ Pywikibot#Suggested_TODOs , Pywikibot developers struggle to keep up with the need to review new proposed patches. The community health charts indicate http://korma.wmflabs. org/browser/repository.html?repository=gerrit.wikimedia.org_pywikibot_core&ds=scr a growth in open changesets (patches awaiting review) and a slow turnaround on reviews. As existing research from Mozilla indicates https://wiki.mozilla.org/Contribute/analysis reducing the average turnaround time on patch submissions will help contributors stay inspired and avoid demoralization, especially among new open source contributors. I recommend that funders invest in training more of the Pywikibot volunteer developers to join Vandenberg as co-maintainers, and pay multiple maintainers to perform regular code review. Training manuals The most important improvement to the manuals would be -- as indicated in the “Assessment” section above -- an addition to Module III to cover Open Street Map integration, as specified in the Making All Voices Count grant and to complement existing materials http://openstreetmap.id/ panduan-osm/ . Additional suggestions for future work follow. In technical manuals, consistent styles for typography and illustration help learners quickly comprehend whether an item is meant as explanatory, imperative, or cautionary. The high standard of graphic design set in Module I ought to be the standard Wikimedia Indonesia follows in Modules II and III. As the underlying software changes, Pywikibot maintainers and Wikimedia Indonesia ought to work together to regularly update the phrasing and screenshots in the training modules. For instance, Module III at one point instructs the reader to use the “Run” option within the “Cell” menu in PAWS’s Python 3 console. That option now has a different name and icon. Similarly, my experience
  45. 45. 14 following the instructions for connecting to test.wikipedia.org in PAWS yielded slightly different numbering in the labels of output and steps: instead of Out[3]: APISite(“test”, “wikipedia”)’ I saw Out[4]: APISite(“test”, “wikipedia”)’ As Jupyter (the PAWS framework) continues to evolve as its own open source project, and as Pywikibot improves, users will need updated instructions, to avoid confusion. Conclusion Using tools like Pywikibot to help add local Indonesian content to Wikipedia is a promising avenue, but in early 2014, several pieces of the pipeline to success were missing, including curriculum and technology. Wikimedia Indonesia’s work over the past two years has built most of those pieces, and if continued, the work is on track to finish the pipeline. With a stronger innovative digital infrastructure in the form of Pywikibot and PAWS, and with a stronger social infrastructure in the form of the three training manuals, this initiative has empowered citizens -- not just in Indonesia -- to speak with a new kind of voice. Sumana Harihareswara 7 May 2016
  46. 46. 15 About Sumana Harihareswara Sumana Harihareswara is an open source software contributor, teacher and project manager with over a decade of experience in the software industry. Her past leadership in nonprofit, academia, industry, and volunteer organizations earned her an Open Source Citizen Award in 2011. She lives in New York City and is the founder of Changeset Consulting. ShewasmostrecentlySeniorTechnicalWriterandEngineeringCommunityManagerattheWikimedia Foundation, where she managed Wikimedia Foundation’s open source software community. While working at the Wikimedia Foundation, she coordinated group learning and work events where volunteers learned new ways to contribute to Wikipedia and Wikimedia projects, and mentored a project to evaluate and improve Pywikibot and similar tools that interface with the MediaWiki web API. Beyond Wikimedia, Harihareswara has contributed to several open source and open data projects, such as GNOME, AltLaw, Mailman, and Bicho (part of the MetricsGrimoire suite). In late 2013 Harihareswara assessed the state of Bicho’s codebase and the project’s friendliness to new developers, producing a list of issues to fix; MetricsGrimoire’s maintainers have since begun rearchitecting the application. Harihareswara has also created open source Python web applications such as Missing from Wikipedia, which queries Wikimedia to find opportunities for editors to add new articles. Harihareswara frequently speaks and writes about open source and management; she was keynote speaker at Open Source Bridge in 2012, code4lib in 2014, Wiki Conference USA in 2014, and FLOSS Metrics Community Meeting in 2016. Her poster at PyCon 2014, “Be A Better Mentor: What Hacker School Taught Me About Community Mentoring,” focused on approaches for designing effective educational institutions and initiatives. She has also presented, taught, and spoke at several hands- on OpenHatch workshops introducing new contributors to the open source software community and teaching them skills such as Git and IRC. Harihareswara has also managed projects at Collabora, GNOME, QuestionCopyright.org, Fog Creek Software, Behavior, and Salon.com. From mid-2014 to early 2015, Harihareswara served as a member of the board of directors of the Ada Initiative. She holds an MS in Technology Management from Columbia University and and a BA in political science from the University of California, Berkeley, and participated in the Recurse Center (a peer-led engineering education institution) in 2013 and 2014. AT UC Berkeley, she taught three semester-long courses to undergraduates via the DE-Cal (Democratic Education at Cal) program.
  47. 47. Evaluation of OpenStreetMap Indonesia Geospatial Data: Samarinda and Balikpapan Yayasan Bumi & Center of Borneo Environmental Remote Sensing, University of Mulawarman March 2016
  48. 48. Evaluation of OpenStreetMap Indonesia Geospatial Data: Samarinda and Balikpapan (Yayasan Bumi & CeBEReS UNMUL) ii Evaluation of OpenStreetMap Indonesia Geospatial Data Final Report March 2016 Yayasan Bumi & Center of Borneo Environmental Remote Sensing, University of Mulawarman Humanitarian OpenStreetMap Team (HOT) Project Team Principal Investigator : Y Budi Sulistioadi, Ph.D Co-PI : Ali Suhardiman, Ph.D Project Administrator : Adi Supriadi, M.Si Field Team : Chaidir Arsyan Adlan, S.Si Seftiawan Samsu Rijal, M.Sc Wisnu Kinanjar Azis Anjari Yayasan Bumi Jl. Suwandi I No 72 RT 24 Samarinda, Indonesia Tel: (0541) 748163 Email: bumi@hijaubiru.org In collaboration with Center of Borneo Environmental Remote Sensing University of Mulawarman Jl. Krayan, Gedung Pasca Sarjana Pertanian Lt 4, Samarinda, Indonesia Email: ceberes@unmul.ac.id
  49. 49. Evaluation of OpenStreetMap Indonesia Geospatial Data: Samarinda and Balikpapan (Yayasan Bumi & CeBEReS UNMUL) iii Abstract This project evaluates the quality of the geospatial data generated through the OpenStreetMap program for the cities of Samarinda and Balikpapan in East Kalimantan Province. The evaluation is conducted by comparing the geospatial data from the OpenStreetMap with the geometrically corrected very high resolution satellite imagery and the coordinate measurements conducted in the field using handheld Global Positioning System (GPS) receiver. The point features were evaluated by their attributes, while the line features were evaluated by their geometric accuracy against the referenced datasets. The polygon features were evaluated through a set of metrics, i.e. perimeter, extent, circularity ratio and the precision of their centroids. In general, the level of accuracy of the OpenStreetMap datasets for Samarinda and Balikpapan are good. There are some errors here and there, but for most cases, the percentage of point and line and polygon features with “high error” state never exceeds 29%. The metrics of the polygon features also show similar pattern and most of them were not significantly different. Considering this evaluation result, we encourage the OpenStreetMap program to regularly update the point and polygon geospatial datasets especially with current information such as business status and update the geometry of the line features to match with the large scale municipal topographic maps or very high resolution satellite imagery.
  50. 50. Evaluation of OpenStreetMap Indonesia Geospatial Data: Samarinda and Balikpapan (Yayasan Bumi & CeBEReS UNMUL) iv Table of Contents Abstract..........................................................................................................................................iii Table of Contents........................................................................................................................... iv List of Tables ................................................................................................................................. vi List of Figures............................................................................................................................... vii List of Appendices ........................................................................................................................ vii 1. Introduction................................................................................................................................. 1 1.1. Background.......................................................................................................................... 1 1.2. Objectives ............................................................................................................................ 1 1.3. Methodology........................................................................................................................ 1 1.3.1. Technical Terms............................................................................................................ 2 1.3.2. Evaluating the OpenStreetMap Geospatial Data .......................................................... 2 2. Project Implementations ............................................................................................................. 4 2.1. OpenStreetMap Datasets and Geographic Boundaries........................................................ 4 2.2. Sample Pre-Screening.......................................................................................................... 5 2.3. Sampling Design.................................................................................................................. 6 2.4. Generating Random Sample ................................................................................................ 6 2.5. Uploading Samples to the GPS Receiver............................................................................. 7 2.6. The Use of Very High-Resolution Imagery......................................................................... 7 2.7. Validation Process ............................................................................................................... 7 2.8. Decision on the General Level of Accuracy........................................................................ 8 3. Project Results ............................................................................................................................ 9 3.1. Samarinda ............................................................................................................................ 9 3.1.1. Point Datasets................................................................................................................ 9 3.1.2. Line Datasets................................................................................................................. 9 3.1.3. Polygon Datasets......................................................................................................... 10 3.1.4. Extraordinary Cases.................................................................................................... 13 3.2. Balikpapan ......................................................................................................................... 16 3.2.1. Point Datasets.............................................................................................................. 16 3.2.2. Line Datasets............................................................................................................... 16
  51. 51. Evaluation of OpenStreetMap Indonesia Geospatial Data: Samarinda and Balikpapan (Yayasan Bumi & CeBEReS UNMUL) v 3.2.3. Extraordinary Cases for Line Datasets of Balikpapan................................................ 17 3.2.4. Polygon Datasets......................................................................................................... 17 4. General Level of Accuracy....................................................................................................... 22 4.1. Samarinda .......................................................................................................................... 22 4.2. Balikpapan ......................................................................................................................... 22 4.3. Recommendation ............................................................................................................... 23 Appendices.................................................................................................................................... 24
  52. 52. Evaluation of OpenStreetMap Indonesia Geospatial Data: Samarinda and Balikpapan (Yayasan Bumi & CeBEReS UNMUL) vi List of Tables Table 1. Technical terms used in this project.............................................................................. 2 Table 2. Metrics calculated to support evaluation of the OpenStreetMap geospatial datasets... 2 Table 3. Geographic Boundaries of the OpenStreetMap Geospatial Datasets Evaluated for this project ........................................................................................................................... 4 Table 4. The Number of Pre-Screened and Total Samples for the City of Samarinda............... 5 Table 5. The Number of Samples ............................................................................................... 6 Table 6. Number and percentage of correct and incorrect attributes of point dataset for Samarinda ..................................................................................................................... 9 Table 7. Error levels of the OpenStreetMap line dataset for Samarinda (in meters).................. 9 Table 8. t-test results between the pairs of GPS vs RS, GPS vs OSM, and RS vs OSM generated geospatial data for polygon extent for Samarinda...................................... 11 Table 9. t-test results between the pairs of GPS vs RS, GPS vs OSM, and RS vs OSM generated geospatial data for polygon perimeter of Samarinda.................................. 11 Table 10. t-test results between the pairs of GPS vs RS, GPS vs OSM, and RS vs OSM generated geospatial data for polygon circularity ratio for Samarinda City............... 12 Table 11. t-test results between the distance from the RS generated and GPS measured polygon centroids to OpenStreetMap polygon centroids for Samarinda.................................. 13 Table 12. Extraordinary cases of polygon features with high error in Samarinda...................... 14 Table 13. Mean and maximum error values for all evaluation parameters and pairs of polygon datasets after removing the samples with outstanding errors ..................................... 15 Table 14. Number and percentage of correct and incorrect attributes of point dataset for Balikpapan .................................................................................................................. 16 Table 15. Error levels of the OpenStreetMap line dataset for Balikpapan (in meters)............... 16 Table 16. Extraordinary cases of line features with high error................................................... 17 Table 17. t-test results between the pairs of polygon extent for GPS vs RS, GPS vs OSM, and RS vs OSM generated geospatial data for Balikpapan ............................................... 18 Table 18. t-test results between the pairs of GPS vs RS, GPS vs OSM, and RS vs OSM generated geospatial data for polygon perimeter of Balikpapan ................................ 19 Table 19. t-test results between the pairs of GPS vs RS, GPS vs OSM, and RS vs OSM generated geospatial data for polygon circularity ratio for Balikpapan...................... 20 Table 20. t-test results between the distance from the RS generated and GPS measured polygon centroids to OpenStreetMap polygon centroids for Balikpapan................................. 21

×