Feedback-Driven Structural Query Expansion for Ranked Retrieval of XML Data

Abstract

Relevance Feedback is an important way to enhance retrieval quality by integrating relevance information provided by a user. In XML retrieval, feedback engines usually generate an expanded query from the content of elements marked as relevant or nonrelevant. This approach that is inspired by text-based IR completely ignores the semistructured nature of XML. This paper makes the important step from content-based to structural feedback. It presents an integrated solution for expanding keyword queries with new content, path, and document constraints. An extensible framework evaluates such query conditions with existing keyword-based XML search engines while allowing to easily integrate new dimensions of feedback. Extensive experiments with the established INEX benchmark show the feasibility of our approach.