SC²S Colloquium - December 16, 2015
|Date:||December 16, 2015|
|Time:||3:00 pm, s.t.|
Matthias Fischer: Simulation of a Recommender System with Adaptive Sparse Grids
An essential part of most Q&A websites is a system to recommend suitable questions to the user. As part of a project with the website “gutefrage.net”, the goal of this thesis is to build a recommender system based on a clustering of the questions in contrast to the existing system, which is based on a text similarity. Methods to cluster the questions that use sparse grids density estimation are already available. The main steps of this thesis are to improve the current clustering, build the recommender system itself and then optimize it via a user simulation using available log data.