Ontological approach for engineering information retrieval

Zhanjun Li, Purdue University

Abstract

Engineering design is a decision making process in which the basic sciences, mathematics, and engineering sciences are applied to convert resources optimally to developa product. In current practice, textual documents and computer-aided design (CAD) drawings are two of the most prevalent data resources in which products and processes are described. In information science areas such as information retrieval, text classification, question answering, and information extraction; statistics-based methods and keyword-based input have been prevalent in research. However, their applications in engineering information retrieval are in early stages of exploration. Most of these applications lack the engineering context such as function, form, material, production, and design processes. Current information retrieval tools either retrieve irrelevant results for engineering or are not in a form that the user can navigate and explore. Engineering context demands the system allow exploration for knowledge discovery and creation of new ideas. "Delivering the right information to the right people at the right time" plays an important role in supporting engineers' memory extension, design concept exploration, design reuse, and in the learning process of especially novice engineers. Current engineering practices ignores reuse of previous knowledge because appropriate information retrieval tools have not been developed. The redundant effort per employee is increasing, with the complexity of enterprises and products. This is in particular critical to the current stage of the economy where design and innovation is becoming more important in view of globalization and manufacturing decline. Attempts have not been made to formalize the well-established contextual knowledge in a specific domain, such as design and manufacturing, to leverage the engineering information retrieval process in the context of product development environment. This thesis proposes and develops a new computational framework that includes an ontological basis and algorithms to retrieve unstructured engineering documents while handling complex queries through a novel orienteering user interface. The results from the preliminary performance test and usability study demonstrate that the proposed method outperforms the traditional keyword-based method and reduces engineers' information seeking efforts under various design tasks. We also propose the combinational approaches which 1) integrate the ontological approach with traditional keyword-based approach to tackle the ontology uncertainty issue, and 2) incorporate the shape-based part retrieval into the framework in order to achieve a more comprehensive and more flexible engineering information retrieval environment.

Degree

Ph.D.

Advisors

Ramani, Purdue University.

Subject Area

Linguistics|Mechanical engineering|Information science

COinS