A query transformation framework for automated structured query construction in structured retrieval environment

Abstract

Current research and development of structured retrieval systems allow search communities to tap into structured resources over the web. These resources have advantages compared with the normal web resources as they have additional annotations that can be utilized when composing a search request. However, structured query languages used by these retrieval systems are not meant for normal searchers. This article addresses this problem by automating the construction of a structured query from a keywords query that is more familiar to the web search user. As a result, a query transformation framework is proposed. Different from the existing framework, which is a one-time solution, our proposed framework has greater flexibility in accommodating new structured query languages and different collection types. A novel intermediate query representation is designed to handle transformation to multiple types of query languages, whereas a context-based probabilistic model is designed to capture the structural features of both homogeneous and heterogeneous collections. Experimental results show that the structured queries generated from our framework give promising results compared with their source queries. This finding suggests that the framework can potentially be used in search applications to improve their search features.