We examine the question of information presentation of (possibly large) ranked datasets in two contexts: recommendations in social content sites and online search of structured datasets. We argue that despite the intuition that structure could be used to improve linear ranking, it is not always available, or sufficient, to help users explore ranked content online. On social sites such as Facebook and Yahoo! Travel, effective content recommendation is becoming increasingly important. A well-known side effect of maximizing accuracy is over-specialization. We discuss the limitations of using structure for diversifying recommendations and formalize explanation-based diversity. We show that our algorithms achieve a good compromise between accuracy and heterogeneity on Yahoo! Movies recommendations. In online applications such as Match.com and Trulia.com, users define structured profiles which are used to retrieve a ranked list of matches. We argue here that neither linear ranking nor attribute-based grouping is adequate for effective data exploration. We formalize rank-aware clustering and evaluate our algorithm over large datasets from Yahoo! Personals.
Sihem Amer-Yahia is a research scientist at Yahoo! Labs since June 2006. Before that, she was member of technical staff at AT&T Labs for seven years. Her research is on data management and query processing in search, recommendations and advertising. Sihem was co-editor of the W3C Recommendation on XML Full-Text search. This year, she chaired SIGMOD 2009 Tutorialsa and the VLDB 2009 Industrial track. She is currently chairing the Social Networks and Personal Information track at ICDE 2010, the Structured and Unstructured Data Track at WWW10 and the Undergraduate Program at SIGMOD 2010.
