Topic Recognition and Correlation Analysis of Articles in Computer Science

Abstract

Topic identification and similarity detection are two related essential task in data mining, information retrieval, and bibliometric data analysis, which aims to identify significant topics and to find similarity between text collections.It is an essential activity to identify research papers according to their research topics to enhance their retrievability, help create smart analytics, and promote a range of approaches to evaluating the research environment and making sense of it.The proposed frame work deals with three main steps: text extraction, topic identification, and similarity detection.The PyPDF2 module is used to extract text from pdf file. CSO classifier is used for topic identification and similarity between documents is calculated using different models, such as Tf-Idf, Bert, Glove, Word2vec, and Doc2vec.and compared these models with respect to cosine similarity and Eucleadian distance obtained from these models.