5. Conclusion
This paper presents an approach to identify emerging technologies, and applied it to a case of mechanisms of electron transfer in electrochemical glucose biosensors. The proposed method uses advanced algorithms to automatically select keywords by applying a TF-IDF function, and to quantify relatedness between pairs of keywords, rather relying on domain experts as did previous keyword-based patent analysis. The algorithm to automatically select keywords replaces previous subjective judgments of domain experts with decision of high rank based on TF-IDF value, and contributes to reduction of domain experts' work by just choosing a cutoff of TF-IDF values. The algorithm to quantify relatedness provides objective evidence for keyword relatedness by identifying pairs of keywords that have high similarity values, and improves semantic processing to compare patent documents. Due to the complexity and fusion of emerging technology, two or three principal keywords are not sufficient to characterize this technology. For example,glucose biosensors entail biotechnology, electronics, and nanotechnology, so emerging technologies could not be expressed using two or three main keywords. For this reason, automatic technical keyword selection that conveys contents of patent is useful, and the relatedness between keywords database allows semantic processing of the similarity of patent documents. Consequently, our algorithms are expected to enhance reliable and objective keyword-based patent analysis to identify emerging technologies.
