1. Description of the experiment
It is common for Semantic Web conferences to run some social tools and experiments for the duration of the conferences. They are intended to (a) encourage deployment of semantic web technologies, (b) gather and share semantic data, and (c) provide useful services to conference attendees.
For ESWC 2009, we are launching an ambitious social experiment, where the Semantic Web, the Social Web, and the Physical World will come together to create a rich integrated network of information (figure below). Acquiring and integrating these heterogeneous, but overlapping, data sources will enable us to provide a range of novel services to conference attendees.
This experiment has a great potential of becoming a default platform for further deployment and development in many other upcoming conferences in a wide variety of domains. In the following we will briefly describe how each of the three information networks will be collected and used.
We strongly believe that the full potential of these data sources can only be realised with integration. This experiment will be the first step towards reaching this goal, and will hopefully be one of many experiments to come.
2. Real-Time Social Contacts
Very small RFID devices will be placed in the badges of conference attendees who agree to participate, to mine real-time physical social contact information (e.g. face-to-face contacts). The hardware and software platform for doing this have already been demonstrated in the context of the SocioPatterns project, which builds on the Open Beacon active RFID platform.
The SocioPatterns project will share with ESWC its expertise in setting up the RFID data collection infrastructure, as well as the required software (RFID firmware, data analysis software, Flash visualizations, and APIs). Similar RFID-based experiments have been successfully deployed recently at a number of conferences, but without any semantics or online profiling.
RFID contacts network
The figure on the right is a screenshot from a live web-based visualisation that was available to the attendees of the 2008 25C3 conference in Berlin. Attendees used these Flash pages to view live data about their present contacts (yellow edges) and their past contacts (green-blue edges). In the visualisation, the central node is the targeted attendee. Edges get thicker the more time the two attendees spend together.
3. Semantic Web Linked Data
In contrast to the full-fledged Semantic Web vision, Linked Data is mainly about publishing structured data in RDF using URIs rather than focusing on the ontological level or on inference. This simplification lowers the entry barrier for data providers, hence fostering a wider adoption. The Linking Open Data (LOD) project is an open collaborative effort aimed at bootstrapping the Web of Data by publishing datasets in RDF on the Web and creating large numbers of links between these datasets. Currently, the project includes over 50 different datasets with several billion of interlinked RDF triples ranging from rather centralised ones (such as DBpedia) to those that are very distributed (e.g. FOAF files).
Data from several past conferences (e.g. ESWC, ISWC, WWW) are already stored as linked data and can be accessed from data.semanticweb.org. Additionally, several publication databases (e.g. CiteSeer, DBLP) have been collected and published as linked data by RKB Explorer. People's communities of practices will be retrieved from RKBExplorer and used in this experiment.
In ESWC this year, we are continuing the trend of collecting metadata on papers, sessions, authors, keynote speakers, and conference attendees, and publishing it as linked data. Additionally, much of the other data we are collecting from the social profiles (see following section) is also stored as linked data.
4. Online Social Profiles
Many of the conference participants are likely to have profiles in one or more of the popular social networking sites, such as Facebook, Flickr, Delicious, or Last.fm. Adding such data to the other two networks above (physical and semantic) can greatly enhance the conference experience, by providing a new rich source of data that is not usually tapped on at such events.
Linking participants to their online profiles creates a good opportunity to extrapolate the extra-academic connections between people. In the TAGora project, we developed an architecture to automatically generate 'profiles of interest' for individuals based on their tagging activity within various folksonomy sites. The basic premise is that the tags used most often will correlate with the concepts a person is most interested in. Our profile building technology includes novel algorithms to align the sense of a particular user tag with the correct DBpedia concept. Such profiles will enable us to calculate various relationships between individuals, based on overlap in their interests, which can be presented alongside more conventional methods, such as their communities of practice. Additionally, explicit 'online social friendships' will be collected and used for recommending talks and for suggesting new contacts.
Profile of interest
5. Services
Combining data from the three rich networks above opens new doors for various types of services to be developed and deployed at the conference. For example, examination of the real-time social interactions of a participant during the conference can reveal the other individuals with whom they spent a certain amount of time in face-to-face contact. Similarly, analysing linked data can reveal a person's community of practice (e.g. based on co-authorships or project memberships). If those individuals are not present within their online social networks (e.g. Facebook), recommendations can be made for adding them. This kind of recommendation would alleviate the common problem in which a person spends some time talking to people, but later forgets their name or how to get in contact with them. It would also enable participants to know in advance those people within their social network that are attending the conference.
Another example service is enriching the visualisations of real-time face-to-face contacts with background information from the other two networks. New edges will be added in to reflect relations of co-authorship, or links from social networking systems.
One of the most valuable experiences of attending a conference is social networking. To this end, services to boost and encourage people to network and to make new contacts can be extremely useful. Calculating the amount of overlap between the attendees' profiles of interest (from online social sites) can reveal new interesting contacts for the people in question. Such suggestions will be made available to individuals on the experiment web page. They can use the real-time physical location visualisation to locate those recommended people.
In addition to the above, we will make recommendations on which particular session or talk to attend. For example, people might be quite interested in attending talks given by others with whom they interact with a lot, in the physical world (RFID data), scientific world (linked data), or online (profiles of interest). Such recommendations will be listed on the website and will be updated daily.
6. Setup
Privacy of participants is top priority. A release form is given to all conference attendees to obtain their permission to collect and use their data. The form explains what the data is, how it will be used, and for how long. No data about any participant will be collected without their signed consent.
Each participant who agrees to take part in this experiment will be given an RFID device to place in their conference badge. They will be shown how to switch the device on/off to gain control on when and where they would like their face-to-face contact data to be collected.
With respect to building profiles of interests, participants can register their Delicious, Flickr, and Last.fm accounts (if they have any) on the experiment website. Facebook Connect will be used to grab social contacts and use it to identify friends at the conference, or to suggest adding new contacts based on the RFID data.
People attending ESWC 2009 will be emailed prior to the conference with information about this experiment, and will be given several contact points to raise any questions, issues, or concerns about this event, before, during, or after the conference.
