Path-oriented keyword search over graph-modeled Web data

Keyword based search systems are becoming increasingly popular and are considered a key feature in many information management systems. Keyword based search approaches have the significant advantage of not requiring users to know how data is organized or stored. Typical approaches assume the dataset to be modeled as a graph, where answers to queries are sub-graphs ranked according to some criteria. Exploring the graph and building and ranking quality pose a number of challenges. In this paper, we discuss Y aanii , an approach for effective Keyword Search over graph-modeled Web data. Y aanii contains a novel approach to keyword search, by extracting the best results from the first set of answers and then combining a solution building algorithm with a ranking technique. In addition to the algorithms and the processes for building result sets, we provide a detailed study of the computational and ranking complexity of Y aanii and compare it with other approaches. We show that Y aanii is superior in terms of efficiency and quality of returned results from both the experimental and theoretical aspects.