| Product family design is one of the prevailing approaches for product realization in mass customization paradigm. With the ever-increasing product offerings in consumer market, a good product family modeling scheme is imperative for incorporating various design aspects, e.g. product function, for better product family analysis and design. Previously proposed product family modeling schema generally stress the component structural aspects, with limited capability to model complex relationships amongst physical components under various design facets. Due to this limitation, existing schemes are constrained to support intelligent PFD tasks, e.g. faceted component analysis based on contextual design requirements like customer voices. In order to cope with these issues, ontology-based representation is identified as a promising solution. Nevertheless, ontology development in design engineering demands a great deal of time commitment and human effort to digest complex design information. This thesis proposes a multi-faceted semantic tagging approach that is able to automatically suggest semantically related annotations based on design and manufacturing corpus. Technically, a semantic relatedness based ranking approach, FacetRank, is introduced in this thesis. Evaluation results using an annotated dataset indicate that FacetRank is capable of extracting salient terms from a collection of documents. Using such an annotation approach, this thesis suggests a semi-automatic methodology of developing a semantically annotated multi-faceted product family ontology (MFPFO). Utilizing the aforementioned ontology, this thesis also presents a new perspective towards ontology supported product family design. Specifically, a faceted information search and retrieval framework based on a semantically annotated MFPFO is proposed. Faceted concept ranking (FCR) approach for ontology-based faceted component search is suggested to generate rank values for component search results corresponding to complex design requirements. From the component ranking results, a platform selection approach is proposed to evaluate the suitability of platform choices and platform change assessment. Utilizing the search ranking results and new commonality metrics, optimal selection of components via a multi-objective design optimization is presented. A case study of three laptop computer design that involves four laptop computer families is demonstrated with promising outcomes. Finally, the advantages, issues and future works possible are also critically discussed in this thesis. |
