Abstract:
Popular programming languages offer thousands of add-on libraries or modules. Popular applications offer thousands of add-on plug-in modules. We consider the problem of recommending a small set of related modules when someone visits a page describing a module. There are three promising sources of information for these recommendations. The first is textual similarity among the descriptions of modules. The second is co-mention of the modules in forum conversations. The third is co-installation of modules in public sites that report on which modules are used. I will report on a deployment on drupal.org, comparing the effectiveness (in terms of user click-throughs) of recommenders that use the three sources. I will also describe a further experiment, about to be released on drupal.org, that dynamically updates the recommendations based on users‘ click behaviors, using a Multi-Armed Bandit learning algorithm. I will describe how we are using the three sources of information to set „priors“ for the bandit algorithm, and how we will assess the amount of improvement in recommendations from using the

priors as compared to a prior-free version of the bandit algorithm.
Background paper: http://portal.acm.org/citation.cfm?id=1639714.1639737

Paul Resnick is a Professor at the University of Michigan School of Information. He previously worked as a researcher at AT&T Labs and AT&T Bell Labs, and as an Assistant Professor at the MIT Sloan School of Management. He received the master’s and Ph.D. degrees in Electrical Engineering and Computer Science from MIT, and a bachelor’s degree in mathematics from the University of Michigan.
Professor Resnick’s research focuses on SocioTechnical Capital, productive social relations that are enabled by the ongoing use of information and communication technology. His current projects include analyzing and designing reputation systems, ride share coordination services, and applying principles from economics and social psychology to the design of on-line communities.
Resnick was a pioneer in the field of recommender systems (sometimes called collaborative filtering or social filtering). Recommender systems guide people to interesting materials based on recommendations from other people. His articles have appeared in Scientific American, Wired, Communications of the ACM, The American Economic Review, Management Science, and many other publications.