Andrea Bernardini

Discovering and recognizing student interaction patterns in exploratory learning environments

In a Exploratory Learning Environment users acquire knowledge while freely experiencing the environment. In this setting, it is often hard to identify actions or behaviors as correct or faulty, making it hard to provide adaptive support to students who do not learn well with these environments. In this paper we discuss an approach that uses Class Association Rule mining and a Class Association Rule Classifier to identify relevant interaction patterns and build student models for online classification. We apply the approach to generate a student model for an ELE for AI algorithms and present preliminary results on its effectiveness.

Full discrimination of subtopics in search results with keyphrase-based clustering

We consider the problem of retrieving multiple documents relevant to the single subtopics of a given web query, termed “full-subtopic retrieval”. To solve this problem we present a novel search results clustering algorithm that generates clusters labeled by keyphrases. The keyphrases are extracted from the generalized suffix tree built from the search results and merged through an improved hierarchical agglomerative clustering procedure. Our approach has been implemented into KeySRC (Keyphrase-based Search Results Clustering), a full web clustering engine available online at http://keysrc.fub.it. We discuss how the keyphrase-based clustering algorithm can be used not only for browsing through the clustered search results but also for producing a re-ranked list of results emphasizing the diversity of top hits. Using a novel measure for evaluating full-subtopic retrieval performance, called “Subtopic Search Length under k document sufficiency”, and a test collection specifically designed for evaluating subtopic retrieval, we found that our approach was able to discriminate between the different subtopics present in search results in a very effective manner, with a clear improvement over other subtopic retrieval systems. In particular, browsing through KeySRC clusters was the best method to retrieve more documents per subtopic (i.e., k>1), whereas using the re-ranked list formed from KeySRC clusters was more effecive for retrieving just one document per subtopic (i.e., k=1).

Analyzing the behavior of professional video searchers using RAI query logs

A large number of studies have investigated the query logs of Web search engines, but there is a lack of analogous studies for multimedia database management systems (MDBMSs) used by professional searchers. In this paper we perform an extensive analysis of the query logs of the RAI multimedia catalogue, both at the query level and at the session level. Based on the observation that a large proportion of the queries returned zero or, conversely, too many hits, we identified three query reformulation strategies to reduce or enlarge the set of results. Our study indicates that the desire of controlling the amount of output may have a relatively limited (moderate-to-little) impact on the users behavior, while at the same time some counter-intuitive findings suggest a suboptimal utilization of the system. The findings are useful for MDBMS developers and for trainers of professional searchers to improve the performance of interactive searches, and for researchers to conduct further work.