(A) study on recommendation approach for engineering information using the personalized retrieval techniques

Abstract

Engineers create engineering documents with their own terminologies, and want to search existing en-gineering documents quickly and accurately during a product development process. Keyword-based search methods have been widely used due to their ease of use, but their search accuracy has been often problematic because of the semantic ambiguity of terminologies in engineering documents and queries. The semantic ambiguity can be alleviated by using a domain ontology. Therefore, this study proposes a semantic search framework to retrieve engineering documents with less semantic ambiguity. Based on the framework, two advanced techniques, ontology-based query expansion and relation-based document ranking, are proposed to improve the accuracy of the search result. For this purpose, a domain ontology is developed based on product structure information and engineering documents. Using the domain ontology, terminologies in documents are disambiguated and indexed. Meanwhile, user’s preference information is exploited in the query expansion process and document ranking process to incorporate the engineer’s personalized information needs. Also, at the same time, both processes consider user’s searching intent which is implicitly inferred from the user’s task context. The experimental results show that proposed approaches can substantially outperform existing approaches in retrieving engineering documents. Reflecting user’s information needs precisely is considered to be the most important factor underlying the notable improvement. The experimental results also provide empirical evidence that the proposed approaches are useful additions to the existing document retrieval practices in the engineering domain. In addition, these approaches will be usefully exploited to recommend task relevant documents, when user’s information needs can be automatically represented as a query. For this purpose, a recommendation approach is also proposed. The recommendation approach tries to find engineering documents directly or indirectly related to the target task, and then analyze the gathered documents to represent user’s information needs of the task for recommending task relevant information. In the approach, PLM usage data is exploited for identifying related tasks and gathering related design information. An empirical analysis using engineering documents is also provided to show the potential benefits of the proposed recommendation approach.