Detecting signals of new technological opportunities using semantic patent analysis and outlier detection

• Published in: Scientometrics Vol. 90; no. 2; pp. 445 - 461
• Main Authors: Yoon, Janghyeok, Kim, Kwangsoo
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.02.2012; Springer
• Subjects: Bibliometrics. Scientometrics; Bibliometrics. Scientometrics. Evaluation; Computer Science; Data mining; Exact sciences and technology; Information and communication sciences; Information science. Documentation; Information Storage and Retrieval; Library and information science. General aspects; Library Science; Research and development; Sciences and techniques of general use; Scientometrics; Semantic analysis; Subject–action–object (SAO) structure; C63; Technological opportunity; Research and development (R&D) planning; Patent mining; Outlier detection; Semantic patent similarity; C82; Multidimensional scaling (MDS); Patents; Correlation analysis; Semantic analysis; Scientific research; Research and development
• ISSN: 0138-9130; 1588-2861
• DOI: 10.1007/s11192-011-0543-2

In the competitive business environment, early identification of technological opportunities is crucial for technology strategy formulation and research and development planning. There exist previous studies that identify technological directions or areas from a broad view for technological opportunities, while few studies have researched a way to detect distinctive patents that can act as new technological opportunities at the individual patent level. This paper proposes a method of detecting new technological opportunities by using subject–action–object (SAO)-based semantic patent analysis and outlier detection. SAO structures are syntactically ordered sentences that can be automatically extracted by natural language processing of patent text; they explicitly show the structural relationships among technological components in a patent, and thus encode key findings of inventions and the expertise of inventors. Therefore, the proposed method allows quantification of structural dissimilarities among patents. We use outlier detection to identify unusual or distinctive patents in a given technology area; some of these outlier patents may represent new technological opportunities. The proposed method is illustrated using patents related to organic photovoltaic cells. We expect that this method can be incorporated into the research and development process for early identification of technological opportunities.