关键词: 文本分析, 文献计量学, 文本挖掘, 主题词簇, 光伏电池, Text Analysis, Bibliometrics, Text Mining, Term Clumping, Photovoltaic Cell

Abstract: How to retrieve potential information and improve the capability for knowledge discovery is one of the hottest topics in the research of science of science. Most related approaches are based on the “terms” delivered from Natural Language Processing. However, phrases and terms retrieved in this way are really huge and “noise,” which is impossible for individuals to deal with and also incredible for software processing. Based on bibliometric and text mining techniques, this paper constructed the semi-automatic “term clumping” steps, involving with stopwords thesaurus, fuzzy matching, association rules, term frequency inverse document frequency analysis and cluster analysis. These steps combined quantitative and qualitative methodologies for term cleaning, consolidation and clustering, and generated better term list for competitive technical intelligence. This paper applied the term clumping steps in the dataset from Derwent Patent database on China’s Photovoltaic Cell, and demonstrated these steps effectively.