Hybrid Method of Keyword Extraction for English Patents

• Main Authors: Lee,Po-Hui, 李柏輝
• Other Authors: Bian, Guo-Wei
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/75925528236434131876

碩士 === 華梵大學 === 資訊管理學系碩士班 === 100 === Patent is not only the technical indicators of the company or individual, but also a very important asset. In the development of new technologies or products, we need to consider the basis of its own patents or technology, and whether the similar technology in patent documents had been registered. In addition to the management of own patents, the searching for related domestic and foreign patent documents is another important task for development of products. 　　In this study, the hybrid technology of keyword extraction is adopted for patent documents in English. It can help the indexing and related terms subtasks for information retrieval of patents. 　　The results showed that the words extracted by the statistical extraction method are resulting in the correct rate by the articles, adverbs, and conjunctions. The rule-based extraction method retrieves a small number of words with the vocabulary term as the end. And the correct rate of this method is relatively high, because the vocabulary-oriented approach in the formulation of rules to avoid the common errors. However the rule-based extraction method required the processing of POS tagging and the matching of rules, the processing speed is slower. The proposed method improved the statistical method, and the experimental results show that it can remove approximately 50% of the errors of the statistic-based method.