Impressions from ACM SAC 2010

Fred Durao's picture

The 25th ACM SAC conference was held in Sierre, Swiztherland, a very privileged area surrounded by the stunning Swiss Alps. Comparing to my previous attendance, it seems the number of attendees has decreased (maybe just an impression). On the other hand, the quality of works and the level of plenary discussion seem to be improved. Because the conference is multi-track, I could attend sessions of related areas to my current work such as data mining and semantic web.

Our talk took place in the traditional IAR track (Information and Retrieval) involving researchers and practitioners from other fields such as artificial intelligence and data mining. I presented the paper entitled “Extending a Hybrid Tag-Based Recommender System with Personalization”, authored by Fred Durao and Peter Dolog, where we describe how semantic similarities between tags can improve the precision of a tag-based recommender system. The talk was okay and triggered two questions from the audience: one about constraining the queries to a given domain and the other about adding a learning model to the recommendation model. The answer to the first was to constraint the search to specific domain ontologies and rewrites the Sparql queries regarding the roles of concepts for the chosen ontologies. This however reduces considerably the reusability of the approach. The answer to the second question was to maintain a knowledge base containing the set of successful semantic similarities but aware to the cost needed to provide such maintenance.

The 26th ACM SAC 2011 will be held in Taiwan, China.
 

Fred Durao.

Peter Dolog's picture

about reducing reusability

I think it does not have to reduce reusability, if the ontology concepts and roles/properties are chosen dynamically in runtime according to some heuristics.  You can again look for such an approach to the paper I co-authored with Heiner Stuckenschmidt, Holger Wache and Joerg Diederich http://www.springerlink.com/content/f67x042445030v05/:

 

Peter Dolog