What do you think? - A social networking site for academia.
 |
||||
Interdisciplinary collaboration is an important component of academic research. It serves as a means to pool physical and intellectual resources, gain social status and knowledge, and bridge ideas from different areas. However, it is difficult to find such collaborators because many researchers do not know what others are working on, nor how their interests might overlap. This problem is paramounted by the fact that office spaces usually consist of people with homogenous sets of interests. In this work, we attempt to support collaboration amongst the various subfields within the computer science department at the Univerity of Maryland (UMD). To this end, we introduce UMDRecorder, a Firefox browser plugin that captures user web activity on "educational" domains. We attempt to use web activity to automatically generate user profile pages which summarize user interests, and match users to other similar users based on these interests. We also employ data mining techniques on top of this data to structure the data into relevant "topics". Our tool was used by 9 graduate students at UMD for 6 weeks. The results of our evaluation study suggest that web activity is a useful indicator of a user's research interests, but more sophisticated data mining techniques are required to filter and organize the information in a meaningful way.
Downloads
 Final Paper |
 umdrecorder1.png |
 umdrecorder-common.png |
 umdrecorder-home.png |
 umdrecorder-home1.png |
 umdrecorder-papers.png |
 umdrecorder-topic.png |
 umdrecorder-updates.png |