A Journal Article Selection based on a Combination of Scanning and Skimming Techniques

Authors

รศ.ดร.นิธินันท์ ธรรมากรนนท์, Mr.์Nantapong Keandoungchun

Published

Naresuan University Journal: science and Technology

Abstract

Typically, an academic writing style of journals is complicated because those journals use complex vocabularies and complex sentences. It might be hard for students to read documents in detail in short period of time. Therefore, this research proposes a Journal Article Selection (JAS) based on a combination of scanning and skimming techniques to reduce number of documents sufficient to support unskilled readers. JAS is developed by python on Google Colaboratory (Bisong, 2019). The designed JAS is assessed in three aspects which are the optimal number of topics and subtopics by considering the coherence value and the relevant terms within topic, the correctness of model and the percentage of the document reduction. The experimental results revealed that there are three layers of topic model: Medical, Business Law and Computer are in layer 1, Social Network and Information System are in layer 2, and Process Model, General Problem and Artificial Intelligence are in layer 3. On other aspects, the proposed model is achieved in 77.36 average percentage of F-measure and 94.15 average percentage of unnecessary document reduction. In conclusion, it can be concluded that JAS can reduce documents sufficient for readers to read in short period of time.

(2022). A Journal Article Selection based on a Combination of Scanning and Skimming Techniques. Naresuan University Journal: science and Technology, -(1), 30-41.