Towards Integrating Collaborative Filtering in Visual Data Exploration Systems

Abstract

Visual data exploration assists users in investigating data by providing recommendations. These recommendations take the form of queries that retrieve these data for the next exploration step, paired with suited visualizations. This paper extends the content-based recommendation techniques adopted so far in EVLIN for query recommendations with collaborative-filtering recommendation techniques. For that, we propose a merge approach that fuses evolution provenance graphs representative of individual user’s exploration session into a global multi-user graph. This merged graph is then used by our collaborative recommendation approach that searches similar queries for a given user query in the multi-user graph to then recommend queries that were previously explored in exploration steps adjacent to these similar queries.