In an era of rapidly expanding digital content, the number of e‐documents and the amount of knowledge frequently overwhelm the R&D teams and often impede intellectual property management. The purpose of this paper is to develop an automatic patent summarization method for accurate knowledge abstraction and effective R&D knowledge management.
This paper develops an integrated approach for automatic patent summary generation combining the concepts of key phrase recognition and significant information density. Significant information density is defined based on the domain‐specific key concepts/phrases, relevant phrases, title phrases, indicator phrases and topic sentences of a given patent document.
The document compression ratio and the knowledge retention ratio are used to measure both quantitative and qualitative outcomes of the new summarization methodology. Both measurements indicate the significant benefits and superior results of the method.
In order to implement the methodology with practical success, the accurate and efficient pre‐processing of identifying key concepts and relevant phrases of patent documents is required. The approach relies on a powerful text‐mining engine as the pre‐process module for key phrase extraction.
The methodology helps R&D companies consistently and automatically process, extract and summarize the core knowledge of related patent documents. This enabling technology is critical to R&D companies when they are competing to create new technologies and products for short life cycle marketplaces.
This research addresses a new perspective in R&D knowledge management, particularly in solving the knowledge‐overloading issue. The methodology helps R&D collaborative teams consistently to summarize the core knowledge of patent documents with efficiency. Efficient R&D knowledge management helps the firm to take advantage of IP positioning while avoiding patent conflict and infringement.
