Finding frequent co-occurring terms in relational keyword search

Given a set Q of keywords, conventional keyword search (KS) returns a set of tuples, each of which (i) is obtained from a single relation, or by joining multiple relations, and (ii) contains all the keywords in Q. This paper proposes a relevant problem called frequent co-occurring term (FCT) retrieval. Specifically, given a keyword set Q and an integer k, a FCT query reports the k terms that are not in Q, but appear most frequently in the result of a KS query with the same Q. FCT search is able to discover the concepts that are closely related to Q. Furthermore, it is also an effective tool for refining the keyword set Q of traditional keyword search. While a FCT query can be trivially supported by solving the corresponding KS query, we provide a faster algorithm that extracts the correct results without evaluating any KS query at all. The effectiveness and efficiency of our techniques are verified with extensive experiments on real data.

• Type: Conference
• Year: 2009
• Where: EDBT
• Authors: Yufei Tao, Jeffrey Xu Yu
• Keywords: Database, EDBT 2009, FCT Query, Keyword Search, KS Query