Andrew Tomkins

"Analyzing Dynamics of Choice among Discrete Alternatives"

(Vortrag im Rahmen der "MPI Distinguished Lecture Series" in Kooperation mit dem Fachbereich Informatik)

In this talk we'll consider two problems in which users must select from a set of alternatives. In the first scenario, a user consumes a class of item repeatedly, such as listening to a sequence of songs, or visiting a sequence of restaurants over time. Our goal is to understand the dynamics of repeated consumption of the same item. We present a model related to Herbert Simon's 1955 copying model, and analyze its effectiveness. In the second scenario, a user traverses a directed graph whose nodes represent items, and whose arcs represent related items recommended by the system. In this setting, we develop a model and algorithm for determining the underlying quality of each node based on traversal data. Our result provides a well-motivated unique solution to the problem of "reverse engineering" a markov chain by finding a transition matrix given the graph and the steady state. Bio: Andrew Tomkins is an engineering director at Google working on web analysis, search, and personalization. His research has focused on measurement, modelling, and analysis of content, communities, and users on the World Wide Web. Prior to joining Google, he spent four years at Yahoo!, leading search research and serving as chief scientist of the search organization. He also spent eight years at IBM's Almaden Research Center, where he co-founded the WebFountain project and served as its chief scientist. He has published over 100 technical papers and submitted over sixty patents. Andrew received Bachelors degrees in Mathematics and Computer Science from MIT, and a PhD in CS from Carnegie Mellon University. He recently co-chaired the program of WWW2008 and KDD2010, and serves on the editorial board of ACM Transactions on the Web.