Title: Research on the evolution of public opinion and topic recognition based on multi-source data mining

Authors: Zeguo Qiu; Baiyan He

Addresses: School of Computer and Information Engineering, Harbin Commercial University, Harbin, Heilongjiang Province, China; China and School of Management, Harbin Commercial University, Harbin, Heilongjiang Province, China ' School of Computer and Information Engineering, Harbin Commercial University, Harbin, Heilongjiang Province, China; China and School of Management, Harbin Commercial University, Harbin, Heilongjiang Province, China

Abstract: In the era of rapid network development, the internet has become the main medium for the spread of online public opinion. Users can express their views on hot issues through text, pictures, and videos anytime and anywhere, analyse the evolving trend of netizens' emotions in emergencies to discover the law of public opinion evolution and potential risks, and provide decision support for public opinion guidance and control. In this paper, Python is used to pre-process the collected text data, and proposes a method based on spectral clustering algorithm, using the Latent Dirichlet Allocation to extract text topics. From the perspective of multi-language and multi-source data, it mines high-value public opinion topics in online public opinion, and uses data visualisation methods to study the emotional tendencies of netizens. The research results can clearly reveal the content of discussion and the emotional attitude of netizens in the period of public opinion transmission.

Keywords: internet public opinion; sentiment analysis; multi-source data; visualisation; natural language processing.

DOI: 10.1504/IJCAT.2022.127816

International Journal of Computer Applications in Technology, 2022 Vol.69 No.3, pp.219 - 227

Received: 23 Jul 2021
Accepted: 06 Oct 2021
Published online: 19 Dec 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article