Term extraction and correlation analysis based on massive scientific and technical literature
Wen Hua Zeng, Hongjiao Xu, Junsheng Zhang · International Journal of Computational Science and Engineering · 2017
Scientific and technical term is the basic unit of knowledge discovery and organisation construction. Correlation analysis is one of the important technologies for the deep data mining of massive, different scientific and technical literature. Based on the freely available digital library resources, this study adopts the technology of natural language processing to analyse the linguistics characteristics of terms, and combines with statistical analyses to extract the terms from scientific and technical literature. Using the results of term extraction, the paper proposes the algorithm of improved VSM towards correlation calculation for analysing different scientific and technical literature. According to the experimental results, it proposes a new way and possibility to automatically extract terms and realise correlation analysis for different literature from massive scientific and technical literature. Our method is superior to the method of unadopting linguistic rules and MI calculation. The accuracy of terms is about 73.5%. Compared with the traditional VSM based on terms, the correct rate of correlation calculation is increased by 12%.