A talk given at the Semantic Reasoning workshop held at the National Museum of Natural History September 6, 2012. The audience included computer scientists and biological scientists interested in using EOL for their research.
8. EOL has Global Partners and is
internationalized
Norway
Dutch
USA Taiwan
Mexico China
Egypt
India
Costa
Rica Colombia
Peru
Australia
South Africa
9. From Moorea Biocode
EOL summarizes knowledge
Erosaria caputserpentis
Serpent's Head Cowrie
Depth range based on 51 specimens in 2 taxa.
Water temperature and chemistry ranges
based on 40 samples.
Environmental ranges
Depth range (m): -5 - 67
Temperature range (°C): 23.011 - 28.496
Nitrate (umol/L): 0.048 - 0.923
Salinity (PPS): 33.821 - 35.837
Oxygen (ml/l): 4.349 - 4.825
Phosphate (umol/l): 0.088 - 0.228
From GBIF Silicate (umol/l): 0.983 - 4.026 From OBIS
11. http://eol.org/pages/704102
Richness scores
Cynthia Parr Global Content Summit
Species Pages Group 17-19 Jan 2011
12. Whirlwind tour
• What kind of information we have
• How we assemble that information
– Big picture
– Subject semantics
– Names infrastructure
– Curation
– Richness score
• How machines and people interact with EOL
• Next steps
13. EOL aggregates and curates
Scientific Databases, including
BHL, GBIF, ALA, INBio, COL,
Scratchpads, LifeDesks
Scientific Journals Curate
Aggregate
Comment
Rate, Collect
eol.org
Quality control
14. Sharing process adds semantics to content objects
SPM
DwC infoitem
description
Plinian
Core
using
Darwin Core Archive
flat files as
transport mechanism
EOL v2
15. Number of text objects
0 100000 200000 300000 400000 500000 600000 700000 800000
Distribution
Multiple topics
Subject of text object
Habitat
Threats
Conservation
Trends
Associations
TrophicStrategy
PopulationBiology
Migration
LifeExpectancy
Behaviour
Diseases
16. Content objects are associated with taxon
names
Wikimedia Commons: Physeter macrocephalus
(note we actually have over 3.3 million named pages)
21. EOL curation
• Trust or untrust taxon associations
• Add new taxon association
• Set preferred hierarchies
• Set preferred common names
• Leave comments
Coming: Taxonomic concept curation
22. EOL is not Wikipedia
…though we have more than 212,000 Wikipedia
articles and 115,000 Wikimedia images
Can’t currently edit within text objects
23. Whirlwind tour
• What kind of information we have
• How we assemble that information
• How machines and people interact with EOL
– API
– Third party apps
– Collections and communities
• Next steps
24. EOL enables machine interaction
Curate
Aggregate
Comment
Rate, Collect
eol.org
API
Third party apps
27. Studies currently underway
with University of Maryland
• Cross-cultural study on
motivation to engage in citizen
science – Dana Rotman
• Interaction among scientists
and non-scientists on EOL’s
social network – Jae-wook Ahn
• Website traffic analysis to aid
conservation communication –
Yurong He and Bill Fagan
28. Whirlwind tour
• What kind of information we have
• How we assemble that information
• How machines and people interact with EOL
• Next steps
29. Using EOL collections
to get computable data
Step 1: Search on EOL for
organisms with characteristics
of interest. Add each one to an
EOL collection.
Step 2: Write a program using
EOL API methods to retrieve the
external database identifiers for
the species in that collection.
Step 3: Add to your program
code to retrieve data using
external database APIs.
Step 4: Analyze, rinse, repeat.
From Arthur Chapman
31. Efforts underway
Phylogenetic trees: Collaboration with Open Tree of Life project
for draft tree
Computable data challenge
http://eol.org/info/data_challenge
Rod Page’s Bionames project
Alexandria Archive Institute
Devries and Thessen using DBPedia Spotlight to extract
associations among taxa and add to Linked Open Data cloud
Sloan 2 project: Marine computable data
TraitBank ABI proposal
32. Research wishes
• Collecting nominations for research idea
where EOL can help:
http://eol.org/info/wishes_for_research
DUE 15 SEPTEMBER
• Will follow with Rubenstein Fellows call for
proposals
33. Thanks to
Our funders
John D. and Catherine T. MacArthur Foundation
Alfred P. Sloane Foundation
Smithsonian Institution
Marine Biological Laboratory
Harvard University
David Rubenstein
and other funders and donors
All our content providers and global partners
Volunteer curators and individual contributors via
Flickr, Wikimedia, and members of EOL
34. Summary of EOL page richness
Overall Hot List
• 950,000 have content • 30 % of 75K are rich
• 2 % are rich • Average richness = ~30
• ~22 % have only links
• to literature • Red Hot List
• 56 % of 3K are rich
• Average richness = 43
35. Long Tail in databases contributing to EOL
600000
Number of taxa for which content is contributed to EOL
500000
400000
300000
200000
100000
0
1 11 21 31 41 51 61 71 81 91 101 111 121 131
… viewed on log scale
1000000
100000
10000
1000
100
10
1
1 11 21 31 41 51 61 71 81 91 101 111 121 131
Partners in order of # taxa contributed to EOL
36. Taxon page richness algorithm
a (Breadth) + b (Depth) + c (Diversity)
60% 30% 10%
Breadth: Images, topics of text
objects, references, maps, videos, sounds, conservation
status
Depth: # words per text object, # words total
Diversity: Sources (partners) 0 – 100, Threshold 40
Editor's Notes
Whirlwind tour to EOLAs you may know, Encyclopedia of Life is a web site providing global access to knowledge about life on earth.Global – the whole worldAccess – free, and freely re-usableKnowledge – synthesized, not rawLife on Earth – biological diversity
My goals are to give you the whirlwind tour with enough information to ring some bells in areas that might be of interest to you, and inspire you to ask deeper questions
I want to emphasize that EOL deals in summarized knowledge, not raw specimen data. For example, for the serpents head cowrie, we have images like this from the Mooreabiocode project, but instead of serving the individual specimen data, we get the overall distribution of specimen data on a map from GBIF. We also get a summary of environmental data associated with specimens in the Ocean Biogeographic Information System database. Imagine if we could do a summary like this across databases.
This is a graphical way of presenting the summarized data from OBIS, which Jen Hammock on my staff worked on with Edward Van den berghe and our team at the Marine Biological Lab. The salinity range for the species is shown here as just a smal, specific slice of the global ocean minimum and maximums.Looking just at 15 content providers we already work with, it is possible that numeric data such as lifespan or average body weight is already available for more than 800,000 species
EOL takes information from about 200 sources so far, mostly scientific databases, but also including Flickr and Wikipedia, and automatically sorts it onto on taxon pages. Our curators can then trust or untrust it, or anybody can provide comments or ratings. About a thousand credentialed scientists have already volunteered to help with quality control. Actions and comments get fed back to the original providers, and the material on EOL is also available to other applications via an Application Programming Interface, which I’ll talk more about in a moment.We’re partnering with over two hundred scientific databases as well as public conribution sites like Flickr and Wikipedia.100+ partner databases700 curators/1000s contributors/46,000 members2.8 million pages500 thousand pages with Creative Commons contentOver 2 million data objects and >1 million pages with links to research literatureTraffic in past year: 1.7 million unique users, 6.2 million page views
ExtensionLeveraging strengths
EOL takes information from about 200 sources so far, mostly scientific databases, but also including Flickr and Wikipedia, and automatically sorts it onto on taxon pages. Our curators can then trust or untrust it, or anybody can provide comments or ratings. About a thousand credentialed scientists have already volunteered to help with quality control. Actions and comments get fed back to the original providers, and the material on EOL is also available to other applications via an Application Programming Interface, which I’ll talk more about in a moment.We’re partnering with over two hundred scientific databases as well as public conribution sites like Flickr and Wikipedia.100+ partner databases700 curators/1000s contributors/46,000 members2.8 million pages500 thousand pages with Creative Commons contentOver 2 million data objects and >1 million pages with links to research literatureTraffic in past year: 1.7 million unique users, 6.2 million page views
Free for third party applications, as long as licenses are respectedField guidesMobile applicationWeb page widget
Please see me afterwards if you are interested in any of these topics
We have a feature where users can create customized collections of pages or objects on EOL.A scientist could search for a characteristic, say, red flowers, and create a collection of those taxa. Actually, we’ve been doing this with blue coloration in the “Life is blue” collection. If you wanted to test what might be driving the evolution of coloration, you could write a program that uses EOL to get all the Genbank IDs for those species identifiers or some other EOL partner that we’ve mapped to each of those taxon pages, and then use those to go to that database and pull raw data to analyze. For example, genetic sequences, or specimen locations. In the future we hope to make step 2 and step 3 even easier, so you might just be able to click a button and download lots of raw data for your collection from certain data sources.
You can also use EOL for crowd-sourcing. For example, Jennifer Hammock has started a collection called “Mystery associates” and asked people to try to identify the partners shown in photos that have some sort of ecological association. When they’ve been identified, like this sea star and anemone predation interaction, then she moves the image to the “known associates” collection. This adds to the information we have from a bunch of partners on food web interactions, and then would be available for foodweb modelers. There are many other possible ways that the large crowds on EOL could be harnessed to generate new datasets from EOL content. And this is all possible to some degree now.
For the future, we are working on a few new angles. First, we are working to get a more phylogenetic organization available on EOL, because that will definitely help those who are doing comparative analyses and who want a true evolutionary framework. The deadline for submitting a large tree is this weekend, Monday really. The second challenge is to propose research work using computable data and EOL in some concrete way. Perhaps as I suggested with using collections to harvest computable data or perhaps using text mining. Here the deadline is next month for the idea, and then we’re providing funds to accomplish the pilot project over the next year.Finally, in September here in Washington we’re bringing in computer scientists and biologists who have an interest in broad scale data-intensive science using biodiversity data. We expect this to lead to other projects and enhancements of the EOL platform.All this could, in my personal opinion, lead up to EOL beginning to serve as The Smithsonian’s phenotype repository. Parallel with genbank, we could be the initial point of entry for ecologists or other biologists seeking large-scale structured information about the observable characteristics of organisms.
Also note that there is an implication that a “rich page” is a “high quality page” – not necessarily true but often it is.As EOL goes forward with our version 2 we’ll be gathering other inputs that can tell us if a page is successful – ratings of its objects, for example. The numbers in yellow are definitely out of date
Inspired by community ecology & measures of species diversity, which of course were originally inspired by information theory, but we haven’t used those measures. Instead we put together these factors in a way that we could assign weights to different factors based on how well they capture “a rich page”We sampled dozens of pages and had team members assess them for their gestalt “richness” based on their own criteria. Then we compared those scores to those generated by the algorithm, and iteratively changed weights until we achieved a set of weights that appeared to reflect human perception of “richness.”Note that there’s a penalty that unvetted material is only worth about 75% of vetted materialAlso there are maximums for many of these input values – having 200 images may not make a page much more rich than having 25 images.Reserve the right to change this to ensure that the index is as useful as possible. Like Google PageRank, want to ensure that nobody can game the system.