Building MultiView Analyst Profile From Multidimensional Query Logs: From Consensual to Conflicting Preferences
This work addresses the challenge of summarizing user preferences in domains like data analysis, but it appears incremental as it builds on existing preference summarization methods.
The paper tackles the problem of constructing user profiles from OLAP query logs to better meet analyst needs by clustering preferences into consensual, semi-conflicting, and conflicting categories, and enriching them with personal, professional, and behavioral views to build a generic model.
In order to provide suitable results to the analyst needs, user preferences summarization is widely used in several domains. In this paper, we introduce a new approach for user profile construction from OLAP query logs. The key idea is to learn the user's preferences by drawing the evidence from OLAP logs. In fact, the analyst preferences are clustered into three main pools : (i) consensual or non conflicting preferences referring to same preferences for all analysts; (ii) semi-conflicting preferences corresponding to similar preferences for some analysts; (iii) conflicting preferences related to disjoint preferences for all analysts. To build generic and global model accurately describing the analyst, we enrich the obtained characteristics through including several views, namely the personal view, the professional view and the behavioral view. After that, the multiview profile extracted from multidimensional database can be annotated.