Research

Diversification

A great amount of information becomes available to users every day through a number of on-line sources. However, locating valuable or important information can prove out to be an overwhelming task, due to the great volume of accessible data. Therefore, in addition to traditional web search methods, proactive search models are becoming increasingly popular and used by a continuously growing portion of Internet users. In these models, users specify their interests via subscriptions and the system automatically forwards all new information (or events) that matches their subscriptions to them. We face the challenge of handling very large, continuous flows of information, a fact that demands the ranking of the forwarded information, so that users receive only what is most useful to them. Our work focuses on the development, implementation and evaluation of models, algorithms and techniques for supporting the ranking of information being forwarded towards the users of large-scale network-centric information management systems. Ranking is based on the importance of each piece of information. We consider that importance is influenced by two main factors: (i) relevance to user interests and (ii) diversity. Relevance is important so that users are only notified about the most interesting events according to their specified subscriptions, while diversity ensures that the received notifications are not referring to the same or similar events.

Recommendations

Typically, users interact with database systems by formulating queries. However, many times users do not have a clear understanding of their information needs or the exact content of the database, thus, their queries are of an exploratory nature. We propose assisting users in database exploration by recommending to them additional items that are highly related with the items in the result of their original query. We propose a novel exploration mode of interaction: we present to the users additional items which, although not part of the answer of their original query, may be of interest to them. This way users see information that they may be unaware that exists. For instance, when asking for movies directed by F.F. Coppola, we guide exploration by recommending movies by other directors that have directed movies similar to those of F.F. Coppola, i.e., with similar characteristics, such as, genre or production year. We also consider expanding the original query with additional attributes, by finding correlations with other relations. For example, when asking for the title of a movie, we also look into its genre or other characteristics.

Useful links

• DBLP
• mdrosou@DBLP
• mdrosou@CiteULike Library
• The Fixed Time World Clock