iJournals: International Journal of Software & Hardware Research in Engineering (IJSHRE) ISSN-2347-4890 Volume 9 Issue 5 May 2021 Sara Mohammed; Tarek El-shishtawy; Walaa Medhat, Volume 9 Issue 5, pp 73-83 May 2021 A Stem-Based Classification Approach for Identifying Author Specialty Sara Mohammed 1 ; Tarek El-shishtawy 2 ; Walaa Medhat 2,3 Information System Department, Faculty of Computers & Artificial Intelligence, Benha University, Benha, Egypt 1 ; Head, Department of Information System 2 ; Information System Department, Faculty of Computers & Artificial Intelligence, Benha University, Benha, Egypt 2 ; Information Technology and Computer Science, Nile University 3 sara.ibrahim17@fci.bu.edu.eg 1 ; t.shishtawy@bu.edu.eg 2 ; WMedhat@nu.edu.eg 3 DOI: 10.26821/IJSHRE.9.5.2021.9519 ABSTRACT Researchers and readers of scientific articles face the problem with identifying the articles and scientific research papers categories and hence the difficulty in determining authors' specialty. Many researchers face the problem of selecting a journal that is suitable for publishing his/her scientific research paper. Many experiences assist researchers in choosing the appropriate journal. However, no one addresses the problem of determining the publisher's specialty of the scientific paper according to his / her article. This paper proposes a solution to identify the author's specialty through abstract comparison. Also, it suggests a new method to help choose the appropriate journal. That finds the appropriate journal according to the abstract of the article that is required to be published. A classification model designs to find the correct category of a given article. Accordingly, the author's specialty is determined. The classifier also finds the Scimago journal categories according to the journal's scope. We built the classifier using a vector space model based on a cosine similarity measure. Also, we use M-TF-IDF weight which is a TF IDF, but we have suggested a modified method that helps us with the measurement. After classifying the article category, a second classifier based on the Levenshtein algorithm selects the appropriate journal for publishing an article. Our dataset is divided into three groups: the scopes of journals, the abstract of articles, and the title of the journal and its scope datasets—all datasets in the main category fromthe Scimago website. The proposed measure shows good performance of results. Keywords: Classification, Vector Space Model, Cosine Similarity, Modified TF-IDF, Levenshtein Edit Distance. 1. INTRODUCTION Due to the diversity of scientific disciplines, the reader or researcher has a problem determining the publisher's specialty according to his published article, i.e., to any branch of science the article belongs to. There is no specific measure to determine the publisher's specialty just on reading the article. This is because of the convergence of phrases and words in most science branches. Also, the increasing of number of journals are available for researchers to publish their research made a problem in choosing the best journal for publishing. Therefore, the second problem addressed in this paper that faces the author is choosing the appropriate journal to publish the research papers and articles. This is because the process of publishing research papers and articles requires much time and effort to read and select the appropriate journal for publication. This work focuses on solving the previous problems. Firstly, we propose a text classification technique that automatically classifies authors and