Social Network Analysis

The Social Network Analysis group works on statistical models for analyzing relational data. One focus lies on machine learning methods for predicting unobserved relations between entities, e.g. which users are supposed to be friends, which products a customer might want to buy in the future, etc. Such problems often deal with relations of higher-order, are multi-relational and the setting is highly sparse. Our research deals among others with matrix and tensor factorization models, sequential models and advanced learning techniques.

Members Assistant Prof. Dr. Steffen Rendle [more]

Teaching Lecture Factorization Models for Machine Learning Seminar KDD Data Mining Cup 2012 [more]

Publications Steffen Rendle (2010): Context-aware Ranking with Factorization Models, in Series on Studies in Computational Intelligence, Springer, ISBN: 3642168973. Steffen Rendle, Christoph Freudenthaler, Lars Schmidt-Thieme (2010): Factorizing Personalized Markov Chains for Next-Basket Recommendation, in Proceedings of the 19th International World Wide Web Conference (WWW 2010), ACM. Best Paper Award. [more]

Software libFM is a library for factorization machines (FM). It can mimic several state-of-the-art factorization models including SVD++, PITF or FPMC just by specifying the input data. Tag Recommender implements the Pairwise interaction tensor factorization PITF for recommending personalized tags. MyMediaLite is a lightweight, multi-purpose library of recommender system algorithms. [more]