Early Identification Method of Interdisciplinary Topics based on Text Clustering and Multi-Label Classification
[Research purpose]Using patents as research data,this paper proposes an early identification method for interdisciplinary top-ics by combining text clustering and multi-label classification.[Research method]Taking"quantum computing"as the research field,a large number of non-interdisciplinary patents are filtered out from the patent collection through two methods:selection based on clustering results and selection based on multi-label classification.Then,the topic identification method is adopted on a small dataset with a high proportion of interdisciplinary patents to achieve early identification of interdisciplinary topics.Subsequently,empirical research is conduc-ted on the Derwent patents to verify the effectiveness of the proposed method.[Research conclusion]Some interdisciplinary topics such as"quantum encryption technology"and"quantum computing technology and quantum computers"are found.Compared with existing methods,the method can discover interdisciplinary topics in the literature when the interdisciplinary field is still in its embryonic or growth stage and the number of relevant literature is small.
万方数据
维普
