Master'sOpen Access

Topic modelling of TOJDE Journal with LDA

2017
0 views
0 downloads
Advisor: Doç. Dr. Özgür Yılmazel

Abstract (EN)

In order to record various information, advantages of computer systems such as security, cost, accessibility as well as the provision of access to rapidly growing data in the information age, the issue of extracting the information sought from these data has caused difficulties. Topic modeling algorithms such as Latent Dirichlet Allocation and topic modeling tools developed on these algorithms is often used to determine the topics mentioned between thousands of documents. In this thesis, the study aims to generates searchable texts from the articles registered by The Turkish Online Journal of Distance Education (TOJDE) journal and perceive the topics with using Latent Dirichlet Allocation algorithm on these searchable texts. Along with the detection of the topics, a system that presents visual user-friendly analysis charts developed and reached graphical outputs which show the distribution of the topics according to years. The words in the article archive have been simplified with operations such as stemming before analyzing texts with the Latent Dirichlet Allocation, thereby increasing the success of the topic modelling process.

Author

Yusuf Kartal

How to Cite

Yusuf Kartal (Master Thesis). Topic modelling of TOJDE Journal with LDA, 2017, Anadolu University.

Keywords

License

Tüm Hakları Saklıdır

This work is shared under the specified license terms.

More theses from Anadolu University