Font Size: a A A

The Research On Topic Evolution For Chinese Literature Of Science And Technology Based On LDA

Posted on:2016-12-13Degree:MasterType:Thesis
Country:ChinaCandidate:S W YuanFull Text:PDF
GTID:2308330464954810Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
With the continuous development of the Internet, the researchers can easily get a large number of literature of science and technology from the Internet, however, faced with such massive amounts of data and information of science and technology, researchers, to be found in this vast scientific and technological information they want tend to feel helpless So, how to get useful knowledge from the vast literature of science and technology information, currently becomes an important problem to be solved.LDA model is widely used at present, and used to a basic technology of data mining and natural language processing, by a great number of scientific researchers analyzing field hotspot and trend Based on the existing defects of the LDA model and the excavation of the practical need of science and technology literature, exploring the useful knowledge in science and technology literature information. The LDA model is used to collect the implied topic, by topic filtering method of filtering to get hot topic, and then according to generate topics in discrete distribution of time coming up from the macro research hotspot and trend analysis in the field of Chinese science and technology literature, in order to help researchers to quickly understand and grasp, in general, research status and development trend in the field of science and technology. Main work is as follows:First, In terms of subject evaluation, this paper is producing artificial evaluation on the topic, then scoring the topics on the basis of evaluation grades from the subjective, and filtering out some low topics scored.Second, because the evaluation of personnel has certain limitation, namely subjectivity is too large. It is difficult to correct evaluation for researchers, and It is proposing topic filtering method based on entropy and the contribution, to filter the topic. and comparing with manual evaluation.Last, using the method of document approval ratings to the topic evolution in the intensity; and using the methods of KL distance and cosine formula to the content evolution analysis, in order to help researchers to quickly understand and grasp, in general, research status and development trend in the field of science and technology.In this paper, we select the title, abstract, key words provided by the authors from the Journal of computer to experiment, the results show that the topic filtering method which proposed in this paper, to a great extent, improves the quality of the topics; Topic evolution method is able to show the change of a topic on the strength and content in time.
Keywords/Search Tags:LDA Model, Topic Filters, Evolution of the Topic, Topic Models
PDF Full Text Request
Related items