This proposal explores in a novel fashion an understudied social phenomenon - the nature and dynamics of short-term professional communities. Modern community-based systems such as social networking and recommender systems have to date been effectively limited in applicability to large, long-term communities. The research goal addressed by this proposal is to maximize information- and people-finding in short-term professional communities. The subject of the investigation is professional research conferences (a specific kind of small short-term community). The project will explore new methods to leverage information about user interests and contacts, which is already available from multiple external sources, and develop cross-cutting techniques for existing social technologies.
The modern Web has introduced a range of social tagging, social networking, and recommender systems that are now utilized by millions of regular users. While quite different from each other, these systems share a very important feature: their success is based on the contributions of their users. To be truly successful, these systems require a large volume of contributions: the more items, tags, links, or ratings in a specific system, the greater its ability to help users. Unfortunately, experience with many online communities shows that a relatively small proportion of their members actually become active contributors. This drawback to social systems effectively limits their area of applicability to only large, long-term communities. The fact that social systems are currently of service only to large, long-term communities does not mean that smaller, short-term communities would not benefit from these systems. On the contrary, additional support for information sharing, connecting people, and finding the most relevant information is even more important for smaller short-term communities precisely because of their short-term nature. This makes it harder to establish valuable connections and to find critical information using regular means. Take, for example, a community of people brought together for a short time to deal with a critical disaster such as the Haitian earthquake or Gulf oil spill. These people are coming from different backgrounds and have different skills and valuable pieces of knowledge. This small community will have access to a large volume of information, which can be helpful for finding a solution. To solve the problem at hand, these people have to establish working connections, share their knowledge, and find the most pertinent information in a short period of time. Their very success as a community may depend upon finding and sharing critical pieces of information or identifying a subset of members (those with complementary ideas) to collaborate. This situation is exactly when social systems could be of great help. This need defined the global research goal addressed by our project: to maximize information- and people-finding in small short-term professional communities. We selected professional conferences as the subject of our one-year pilot investigation, as they are an important and feature-rich type of small short-term community. In this specific context, the immediate goal of our project was to develop and evaluate a set of techniques, which will maximize the ability of research conference attendees to discover new information and find useful contacts in this short-lived environment. To explore the innovative techniques for information- and people-finding in the context of professional conferences, we developed Conference Navigator, an integrated personalized social bookmarking and social networking system. Over the course of the project, we incrementally expanded the set of Conference Navigator features while deploying and evaluating new versions at academic conferences. Starting with a basic set of talk bookmarking, we gradually added social linking features and advanced recommendation functionalities. By the end of the project, the system had been used at five top-ranked academic conferences: iConference (Feb. 2011), Hypertext (June 2011), UMAP (July 2011), ECTEL (Sept. 2011), and TPRC (Sept. 2011). The results of our work were very encouraging. First, we were able to demonstrate the feasibility and the value of social systems in the context of small short-term communities. By offering a rich integrated social platform, we were able to actively engage a much larger proportion of potential users. In contrast to the normal 1% of active participants in regular social systems, over 25% of conference attendees (301 out of 1197) in our studies emerged as contributors by bookmarking and tagging conference talks and thus adding to the pool of "collective wisdom". As a result, even at those conferences with a relatively small number of delegates, the system was able to harvest a sufficient number of contributions to provide a rich social interface and to offer good-quality recommendations. Moreover, we discovered that offering a more complete set of social and recommendation features can increase the proportion of actively-engaged users from about 15% (a conference where the system offered a minimal set of features and no recommendations) to about 23% (two conferences where the system featured advanced talk recommendations) to about 40% (two conferences where the system added advanced social networking and co-attendee recommendation functionality). The analysis of user surveys also confirmed that the integrated system provides an important value in the conference context and that advanced features are strongly appreciated by the attendees. In addition to our primary findings, the Conference Navigator framework enabled us to run several studies of various personalized recommendation and engagement approaches. The results of the studies were used to improve the system. While the project is completed, we continue our work with enhancing the system and plan to make it available at 4-7 conferences per year. The most recent version of system can be explored at http://halley.exp.sis.pitt.edu/cn3/portalindex.php
