AMIA 2013 - Open Research Networking Gadgets (ORNG)


Published on

Published in: Technology
1 Like
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • Note the orange and the circles The .info is NOT a typo, that’s a real web site Hope you like the flying orange, you’re going to see more of that before we are done.
  • RNT A way for researchers to find and learn about one another for collaboration Software ranges from Open Source (profiles, vivo) to Commercial (sci val) to home grown builds (LOKI or CAP). Installations go from local individual deployments (ucsf) to a centralized SaaS model (KNODE, one of our ORNG partners) LOD is a way of publishing semantic data on the web. The data structure is defined by an ontology OpenSocial. Came from industry and supported by Google, LinkedIn, nature network. OpenSocial has standards for data persistence, people and connections, security (oauth) and user interface, publish-subscribe We like to call OpenSocial applications “gadgets” They are VERY easy to build and install (just point to them). HTML and Javacscript in an XML package.
  • There’s an industry around this. Salesforce Chatter, Jive and Huddle are all targeting the “social networking as a mechanism for online collaboration” space. Social Networking is not just a toy In the pre Google+ days iGoogle and OpenSocial were used by Google as a strategy to attack facebook. The’re now following other strategies to fight that battle and OpenSocial (with many members from Jive on the board) focused on the enterprise space (office productivity tools). Facebook has a similar API to OpenSocial, but it’s proprietary. The enterprise space does not have as much visibility as the consumer space but it’s closer aligned to our needs. This change is better for us. We keep in touch with the opensocial group, now mentions our work on their website. We’re pretty proud of that. LOD and the VIVO ontology solved an issue we had with OpenSocial in a slam dunk kind of way.
  • Nerding out into details on this could eat up an hour of our 22 minutes You can see this info through an ORNG gadget on my web page (eating our own dog food) Glued a bunch of software together Shindig (java reference implementation of open social) as our OpenSocial engine. Same as LinkedIn and iGoogle MIT SIMILE/Epimorhpics added as an OpenSocial feature “slick” to convert LOD into JSON so that they could work with gadgets Added this to both Profiles (.NET) and then VIVO (Java)
  • Notice the “An”. Don’t want to get into the details of the quantitative data behind the smileys. Don’t want to get sued. Describe the “columns of what they do” tell world about me, do things personally for me when I’m engaged with the site, allow me to work with others on the site. Start with Facebook because they legitimized Social Networking in a way that MySpace couldn’t quite do, and part of this (we think) is because they intentionally avoided (want you to know the person in real life unlike myspace) changing a bit with facebook pages as they look for revenue. Linsey Lohan Strong on targeted content (see what my friend fed their cat)) Fairly strong on communication and collaboration, even if it’s for gathering friends to slay an evil dwarf king. Note that Facebook has its own API that competes with OpenSocial LinkedIn Very strong on the first point, strong on the second with their emails, 3 rd is not a big use case for them (but they do have OpenSocial apps that do this) (use LinkedIn in the pre-hire page) Way you can tell a colleague is about to leave is via LinkedIn invite RNT Super strong on first, weak on 2 nd and more or less non existent on last Here comes the calvary! Note that we’ve made our strong points even stronger. A little crowded even 
  • This is where RNT’s are already strong, but we can make them stronger. This first point is critical, and it took us a while to realize we were addressing this but once we did it seemed like such an obvious thing to do. Some data will NEVER be sourced from my RNT. In particular things that are very dynamic (tweets) and things that are not textual (multi media). These things take up a lot of disk space, they often require some fancy flash object to render on a page, and people are already doing for free. Bringing in live data to show on a profile page is cool This ability of using ORNG gadgets to bring in external content has been a sweet spot for us
  • I know for a fact that researchers tweet because I’ve seen in on my phone today at this conference! There’s probably someone in this room tweeting right now, but please pay attention to the presentation. Anirvan 
  • A good video may get more viewership than all a researchers publications have readerships combined. Its more accessible, and it’s good to have both in the profile. Mention UC tv. With the fancy researchers, it’s pretty easy to find them on YouTube. I know that some of the VIVO folks can be found on youtube. Mention katy when we show our dude?
  • Wake Forest , brave enough to deploy our code. They addressed personal content with Recommended Readings and Grant Opportunities. (They took our modified version of Profiles) Gets data from PubMed and some government web site that publishes funding sources. Want to know how easy it is to build gadgets? Andy’s a business school student  . You don’t even need a real nerd to do these things. (Andy is actually a rock star developer) Their gadgets are based on pre LOD Profiles and ORNG, and he needed to code up a custom web service to pipe “researcher interests” into the gadgets. With VIVO LOD, this customization would no longer be necessary. Wake Forest emailed us last week on this subject An idea we have is to build a gadget that will highlight relevant items as you click around on the site. Could help fix that crowding issue with so much data on a researchers page. The person has 200 publications, which ones are relevant to me. We don’t have any screenshots of the Wake Forest gadgets but we’ve seen them and they’re beautiful.
  • The next frontier RNT has the rich data and connections, commercial tools have the functionality. Don’t want to pull all of our data into their systems, don’t want to have to rebuild their functionality in ours systems. Integrate the 2, with ORNG. Similar to the Detailed User Profile technique. Instead of using ORNG to bring in external content, we’re using ORNG to bring in external functionality. The more we do to bring online collaboration into our research networking tools, the easier it will be to measure the value of our RNT’s
  • Built on top of our “researcher shopping cart” that we’ve had for quite a while for dropping out lists. It’s a highly functional gadget, but a little on the ugly side. Used by a select few for generating email list. Now any one of our users can use it to create a chatter group Let’s say this group goes on to do some work that maybe gets some recognition. ANIMATE Humble people that they are. Now we have a data trail that shows “hey, this group was created by Profiles and ORNG” We helped our institution win a Nobel Prize! That would be nice The more we do to bring online collaboration into our research networking tools, the easier it will be to measure the value of our RNTs Chide Brian for being ambitious 
  • Can’t go into details on all of this but just want to show, we have a lot going on. Work with Recombinant and Harvard for next release. Help WakeForest refactor code to remove custom web service Make data searchable (in VIVO and Profiles). Add metadata (very cool). Here about something on CNN about a UCSF professor. Someone comments on his youtube video (being on CNN), you search for CNN on profiles and she pops up! Using crowdsourced data from an external site to make our local RNT smarter. Very googlie of us. Look into an hopefully adopt JSON-LD as RDF to JSON engine. Retool gadgets as needed Add JSON-LD to VIVO Current OpenSocial “RDF” feature works with any rdf (and would LOVE to see adopted by other industries or even Google+ which uses Shindig under the covers). Is there a compelling need to add VIVO specific convenience features so that programmers don’t have to “walk” the RDF tree to get things like a list of publications. All code eventually dies, good code pays for itself first. We bid a warm farewell to some. Opens up idea of “visualization” gadgets that work in Profiles and VIVO to dress up existing RDF
  • AMIA 2013 - Open Research Networking Gadgets (ORNG)

    1. 1. Clinical and Translational Science Institute / CTSI at the University of California, San FranciscoOpen Research Networking Gadgets (ORNG)
    2. 2. But First….• What is a Research Networking Tool? – “at UCSF, it’s basically like LinkedIn for biomedical researchers”• What is Linked Open Data? – Web pages that computers and people can read and understand• What is OpenSocial? – An API for adding web applications (gadgets) into a social or research networking web site
    3. 3. Evolution of Social and Research Networking • Social Networking is recognized as a powerful means of online collaboration • Facebook dominates the consumer space, OpenSocial moves to the enterprise space • Research Networking Tools adopt Linked Open Data (LOD) and the VIVO ontology • We combine the VIVO data standard with the OpenSocial application standard -> ORNG
    4. 4. What is ORNG?• An extension made to both the Profiles RNS and VIVO research networking tools to allow them to run Linked Open Data/OpenSocial applications (ORNG gagets)• And of course, the collection of ORNG gadgets themselves:
    5. 5. How did we build ORNG? An Ontology Driven Approach to Improve the OpenSocial Standard Eric Meeks (UCSF), Leslie Yuan (UCSF), Griffin Weber (Harvard), Maninder Kahlon (UCSF) Clinical and Translational Science Institute, University of California, San Francisco Harvard Catalyst, The Harvard Clinical and Translational Science CenterIntroduction NIH Grant Match* and Recommended Reading* Gadgets using DIRECT Match Gadget using researcher Solution• Science 2.0 is happening, and Research Networking Tools such as researcher data obtained with different custom API’s data obtained with VIVO RDF via JSON • An open source product called Babel which was developed by the Profiles, VIVO, SciVal Experts and others have become MIT Simile Project was discovered. Babel provides many data commonplace throughout our institutions. translation services, including RDF/XML to JSON.• Our Research Networking Tools fulfill a need that can not be met by • A proof of concept system was created by integrating the production commercial social networking sites such as LinkedIn, Facebook and UCSF Profiles code with pre-release VIVO compliant Profiles code, Google+ because we need institutional provenance for our data and integrating Babel with Apache Shindig. content and first class support for our data model. • The DIRECT Match Gadget was built to test the proof of concept• Commercial social networking sites have become platforms. This system. It worked! It has also been successfully unit tested with allows them to leverage numerous development communities and RDF/XML from various external VIVO compliant sources. more rapidly deliver innovative functionality.• Our Research Networking Tools should also become platforms. We believe that delivering more functionality more quickly to our RDF/XML converted to JSON for Griffin Weber researchers will increase productivity and accelerate science.Problem• Converting a web site to a web platform is not trivial. It can be done * Built by Andy Bowline of the independently by a large software development team, or it can be Wake Forest School of Medicine. done by leveraging existing open source solutions such as Apache Shindig.• The software resources available to our institutions are limited as compared to a recognizable commercial site such as Facebook or LinkedIn. Apache Shindig, which is based on the OpenSocial OpenSocial with RDF/XML converted to JSON via Simile Babel standard, is the more attractive if not only viable solution.• The OpenSocial standard does not have first class support for our data model. Fortunately OpenSocial is extendable and this gives us an opportunity to address our specific data model needs. Browser Backend ServicesApproach OR*• Manually extending OpenSocial with custom fields to match our data HTML Content Next Steps model was always an option but an expensive one from a development perspective and a flawed one because customization • Integrate our solution into the RDF based version of Profiles and breaks interoperability. make our code available to the open source community.• Convergence towards RDF and the VIVO ontology across our • Promote our solution to the OpenSocial Foundation. Other verticals Research Networking Tools presented an opportunity. With a M D R are suffering from the same domain based data model issues with X L F standard ontology we now have a standard way to express our data, OpenSocial that we encountered in bioinformatics. / but how can we integrate the VIVO ontology into OpenSocial? • Get you to help us build our community for Open Research• OpenSocial works well with JSON but not with any standard Domain Object Request Request Proxy Networking Gadgets (ORNG) at! serialized forms of RDF such as RDF/XML or Turtle. A standard Babel means of converting RDF to JSON was required. JSON Domain Data Acknowledgments The . This project was supported by NIH/NCRR UCSF-CTSI Grant Number Gadget Content Gadget UL1 RR024131 and Harvard Catalyst Grant Number 1 UL1 RR025758- Specification Gadget Hosting Servers 01. Its contents are solely the responsibility of the authors and do not http://anywhere/gadget.xml necessarily represent the official views of the NIH. We would like to thank Andy Bowline of Wake Forest, MIT Libraries and Ontology. MIT CSAIL as well as all other contributors to the SIMILE Project. * Successfully tested with VIVO (ask for demo!) but not yet implemented. We also want to thank Andy Smith and the OpenSocial Foundation.
    6. 6. An Analysis of Social and Research Networking Detailed User Personalized Online Profile Content CollaborationOpen Source RNTs
    7. 7. Detailed User Profile• Researchers do many things: their data is dynamic and all over the web Tweets Presentations on slideshare Videos on YouTube Code on GitHub• ORNG strengthens RNTs by aggregating and presenting researcher data in “real time” from multiple web sources
    8. 8. Personalized Content• Andy Bowline of Wake Forest has done a great job with ORNG in addressing this need – Recommended Readings Gadget – Funding Opportunities Gadget• UCSF partnered with Ying Ding of Indiana University to have students prototype an ORNG “relevancy” engine
    9. 9. Online CollaborationRNTs have the data and connections; industrytools have the collaborative functionality.ORNG brings the two together.•UCSF integrated Salesforce Chatter intoProfiles•Added “Follow in Chatter” and “Create a groupin Chatter” capabilities to Profiles – The data trail from creating a group is fantastic
    10. 10. Next Steps• Bring ORNG into formal Profiles release• Store OpenSocial data in local RDF (makes it searchable)• Store external metadata in local RDF (searchable)• JSON-LD for gadgets and beyond• Make RDF feature an official part of OpenSocial• Extend RDF feature with VIVO ontology specific feature• Honorable discharge for redundant gadgets• Explore Apache Rave• Build more gadgets! => You and us!
    11. 11. Join the ORNG movement! •Mailing list •App store •Support Eric Meeks, Brian Turner, Anirvan Chatterjee, Leslie Yuan