Show simple item record

A DESIGN OF A JOURNAL ARTICLE SELECTION MODEL BASED ON ABSTRACT USING COMBINATION OF SCANNING AND SKIMMING TECHNIQUES

dc.contributorNantapong Keandoungchunen
dc.contributorNantapong Keandoungchunth
dc.contributor.advisorNithinant Thammakoranontaen
dc.contributor.advisorนิธินันท์ ธรรมากรนนท์th
dc.contributor.otherNational Institute of Development Administration. School of Applied Statisticsen
dc.date.accessioned2022-03-25T04:32:06Z
dc.date.available2022-03-25T04:32:06Z
dc.date.issued7/1/2022
dc.identifier.urihttps://repository.nida.ac.th/handle/662723737/5692
dc.descriptionDoctor of Philosophy (Computer Science and Information Systems) (Ph.D.(Computer Science and Information Systems))en
dc.descriptionปรัชญาดุษฎีบัณฑิต (วิทยาการคอมพิวเตอร์และระบบสารสนเทศ) (Ph.D.(Computer Science and Information Systems))th
dc.description.abstractTypically, an academic writing style for journals is complicated because those journals use complex vocabularies and sentences. It might be hard for students to read journal articles in detail and in short periods of time. Therefore, this research proposes a Journal Article Selection Model (JASM) based on article abstract using combination of scanning and skimming techniques. JASM aims for two objectives. Firstly, JASM aims to find out knowledge within journal articles in the form of multi-layer topic model, which is the hierarchy of topics and subtopics. Secondly, JASM aims to use that knowledge to reduce number of journal articles sufficient to support unskilled readers for reading only important journal articles in limited period. The experimental results revealed that the designed multi-layer topic model consists of 4 layers which each layer contained 1, 5 ,13, and 2 topics, respectively. Each topic of the multi-layer topic model shows the direction or trend of current knowledge journal articles in ScienceDirect, an online journal database.  The knowledge can be used to classify unseen journal articles in this research to help readers who lack of reading skills or new readers in the field of computers to reduce the number of journal articles that need to be read. The designed JASM is assessed in two aspects which are the accuracy of model and the percentage of the journal articles reduction. The accuracy of proposed model can achieve in an average of 83.31 percentage of F-measure. In other aspect, JASM can achieve in an average of 99.20 percentage of journal articles reduction, which can approximately reduce 595 journal articles out of 600 journal articles. It means that readers need to read only 5 journal articles due to limited time. In addition, the existing research found that most people spent 29.3 minutes per article reading, and also spent 17.25 average hours for reading weekly. Therefore, most people must read approximately at least 34 articles per week. Hence, by reducing the number of articles in JASM to only 5 articles, the readers will spend approximately only one day per week to read those articles. Therefore, this research concluded that JASM could reduce journal articles sufficient for readers to read in their course. It supposes to increase number of knowledge transferring to be resided within people.en
dc.description.abstract-th
dc.language.isoen
dc.publisherNational Institute of Development Administration
dc.rightsNational Institute of Development Administration
dc.subjectJournal Articleen
dc.subjectScanningen
dc.subjectSkimmingen
dc.subjectMulti-Layer Topic Modelen
dc.subjectMulti-Layer LDAen
dc.subject.classificationComputer Scienceen
dc.titleA DESIGN OF A JOURNAL ARTICLE SELECTION MODEL BASED ON ABSTRACT USING COMBINATION OF SCANNING AND SKIMMING TECHNIQUESen
dc.titleA DESIGN OF A JOURNAL ARTICLE SELECTION MODEL BASED ON ABSTRACT USING COMBINATION OF SCANNING AND SKIMMING TECHNIQUESth
dc.typeDissertationen
dc.typeดุษฎีนิพนธ์th


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record