Developing base domain ontology from a reference collection to aid information retrieval

摘要

Information Retrieval (IR) is a common technique used to manage a growing technical document collection. Owing to the complexity of technical documents, the direct application of IR often leads to unsatisfactory results. Therefore, many semantic approaches, such as ontology, are applied to enhance IR performance. However, ontology development is often a labor-intensive process and the availability of ontologies significantly influences the applicability of IR. Consequently, many efforts are dedicated to the automation of the ontology development process for reducing the human labors. In addition, reference collections, which are developed as the golden standards to evaluate IR performance in many IR research, can be regarded as an available resource for ontology development. To ease domain ontology development for supporting IR, this research proposes a semi-automated approach to develop a base domain ontology from a reference collection. This research also validates the base ontology on an Earthquake Engineering reference collection, called the NCREE (National Center for Research on Earthquake Engineering) collection. The results reveal that the human workload of the proposed approach is affordable. Furthermore, the base domain ontology can help achieve a satisfactory IR performance.