Application Of Text Mining In Grouping Thesis Topics Using TF-IDF Method Based On Thesis Abstract

Abstract

UPT (Technical Service Unit) Dehasen University Bengkulu Library documents student theses based on study programmes using the Slims 9 (Bulian) application. One of them is the Informatics Study Programme, Faculty of Computer Science, where there is a specialisation in software engineering and networking that distinguishes one thesis from another. This sometimes makes it difficult for the library to provide information to students who are looking for references, due to the large amount of thesis data that is managed and the difficulty in obtaining information on the percentage of the number of thesis topics from each thesis document. The application of text mining in grouping thesis topics using the TF-IDF Method based on thesis abstracts at the Dehasen University Library Bengkulu can help library staff in grouping theses in more detail for the same topic or theme, help library staff quickly find theses that are relevant to user needs, and can find out information on the percentage of the number of thesis topics from each title and abstract that has been submitted, especially in students of the Informatics Study Program, Faculty of Computer Science. The system has uploaded training data as much as 25 data which is used as the basis for the calculation of the TF-IDF Method. In the TF-IDF Method there are several processes that occur, namely Tokenizing, Filtering, Stemming and the final value of tf-idf.  Based on testing that has been done on testing data as much as 3 thesis data, the results show that 2 theses belong to the data mining thesis topic group with a percentage of 66.77% and 1 thesis belongs to the decision support system thesis topic group with a percentage of 33.33%.