Network Visualization for Encyclopedia of Life

Network Visualization for Encyclopedia of Life

HCIL Brownbag11/27/2011

Jae-wook Ahn, Jenny Preece, and Ben Shneiderman

Goals

• Build a latent social network from implicit relationships of a Technology Mediated Social Participation project

• Find out what if the latent network can reflect real participation network

• Find out what we can learn from the latent network

• Find out if the network can encourage future participation

Encyclopedia of Life

• Effort to gather and share scientific knowledge about all living things in a single online resource

• http://eol.org• Multiple information sources• Flickr Photo Images– This group allows anyone to provide images and

videos for the Encyclopedia of Life web site. – Contributors tag their images for EOL use– http://www.flickr.com/groups/encyclopedia_of_life

Flickr Dataset

• Crawled : 7/5/2011

• Machine Tags

– “taxonomy:binomial=Dipsacus fullonum”

Owners # Photos #Photos #

with Machine

TagsPosted Date

1,415 84,543 78,508 (93%) 9/3/2002~

Number of Uploaded Photos per Year

Taxonomy of TagsTag Count Percentage

binomial 53,544 68kingdom 6,084 8family 4,151 5genus 4,018 5

common 3,256 4order 2,558 3class 2,421 3

trinomial 1,562 2phylum 357 <1species 99 <1

Normalizing Tags by Biological Classification

• ITIS (itis.gov)• Catalog of Life (www.catalogoflife.org)• Convert binomials to higher levels– Homo sapiens• Homo (species, genus)• Hominidae (family)• Mammalia (class)• Animalia (kingdom)

Latent network discovery

1. If two people share interests– If Flickr photo tags overlap

2. We assume there are implicit connections– Generate the social network visualization

3. Owner (photo uploader) network visualization4. Co-occurrence of tags by spreading activation– Connect owners with high co-occurrence score (>=

0.9)

12

Birds, Mammals

Insects

Magnolia

Reptiles, Amphibians

Fish

People connecting birds/mammals to reptiles/amphibians

People connecting birds/mammals to insects

Latent network discovery

• Questions:1. Can the implicit network teach us about

the real network?2. Will it contribute to the future behavior

changes?

Evaluating the network

• Contact the people by email• Randomly selected Connectors (2) and Centers

(3)• Questions

1. Have you had any on/off-line relationships with other EOL Flickr members?

2. If you have, what kind of activities did you have, especially, regarding what life forms?

Connectors

Email correspondences

Brian Gratwicke (Birds/Mammals – Reptiles/Amphibians)

“I am a fish guy by training, I am employed as an amphibian biologist by training and used to work for Save the tiger fund. I have numerous online and offline communications with various groups and work at a zoo.”

Peter Edin(Birds/Mammals – Insects)

“I just post to various groups but have no interaction with users in any way except the occasional comment(s) on the photos.”

Centers

Email correspondences

Donald Hobern (Insects)I’ve had a number of relevant contacts. Here is an attempt to summarise:1. I am Director of the Atlas of Living Australia (ALA, http://www.ala.org.au/) which is a national biodiversity data management activity that is aligned with and shares data bidirectionally with GBIF, EOL and other international projects.2. The ALA has adopted the EOL Flickr group as one of the channels for photographers to make photos available for use by ALA (since it benefits many more if we do it this way). We have encouraged many Australian photographers to contribute and I have personally spent time helping several to understand how to tag the images, etc.3. I have had numerous email contacts or face-to-face meetings with EOL staff and representatives and will in fact be meeting with Cyndy Parr today at the Smithsonian to discuss future collaborations between ALA and EOL.

Donald Hobern (cont.)4. I have also spent time commenting on and in some cases suggesting corrected identifications for images in the EOL Flickr group.5. I have on two occasions corresponded with EOL Curators when I have disagreed with some of their comments on my photos. In one case, they agreed that their proposed reidentification was incorrect. In the other they had correctly indicated a problem with one photo of a species but also queried a valid photo.6. I had one of my photos selected as one of the monthly competition winners.• My contacts were generally focussed on photos of species groups for

which I have some expertise (birds, Lepidoptera and some other insect orders) and for photos from Australia and, to a lesser degree, Europe, since I am mostly familiar with these faunas.

Todd Pierson (Amphibians)

(1) Yes.(2) Mostly administrators, but also some amphibian specialists.

Arthur Champman (Birds)• This is a hard question to answer as I have been involved with EOL since

before it began. Indeed one of my emails advocating the EOL was used as the front page of the EOL site for some time prior to its first release.

• As a photographer, I am very much a generalist. I am a botanist by training.• As far as interactions with other EOL people and bird people - I am a member

of TDWG and other organisations , and frequently interact with Cyndy Parr, David Patterson and other EOL staff members. I also interact with a number of other EOL contributors from time to time Iin particular - to use their Flickr ids - treegrow (various), dhobern (birds and moths), Sciadopitys (bids and others), Tony Rodd (plants), Robert Whyte (spiders), mirmycian (ants and spiders), servitude (insects and spiders), Barbol (birds and fish) and many others . I also work a lot with the Atlas of Living Australia people on all sorts of species as well as on Data Quality issues. With photographs, perhaps my biggest interactions are with insect people and entomologists.

Arthur Champman (cont.)• Photography is just part of a bigger environmental arena I work in,

and in many ways it is difficult to separate out the photographic part from those other activities - such as Data Quality, Crop Wild Relatives, Environmental Modelling, Citizen Science promotion and my other consulting activities as they all overlap.

• As a botanist, I also interact with other botanists and mycologists on identifications, creation of sets for various areas in Australia (see for example my Werribee Gorge set), groups (other than EOL) associated with Field Guides in Australia for example - Field Guide to Insects of Australia, Field Guide to Rainforest Plants in Australia, Field Guide to Australian Plants as well as non-Flickr Groups on Fungi, Lichens, etc. I also contribute images for the Reptile Bioblitz and other groups.

Temporal network visualization

• Show the evolution of network visually• Using colors/shades to represent current/past

activities for nodes/vertices– Monthly– Current month (red)– Past month (low intensity colors)

brian.gratwicke – connection with Brds/Mammals

Online demo is accessible at

http://www.cs.umd.edu/~jahn/eol

Next Steps

• Compare with moderators and admins from the Flickr EOL group

• Compare with explicit social interactions gathered from EOL

• Reveal the online temporal visualization to users– Find out how it can encourage future

participations