Open AccessText Mining for Internship Titles Clustering Using Shared Nearest Neighbor
Author(s) -
Lisna Zahrotun
Publication year - 2017
Publication title -
computer engineering and applications journal
Language(s) - English
Resource type - Journals
eISSN - 2252-5459
pISSN - 2252-4274
DOI - 10.18495/comengapp.v6i3.214
Subject(s) - internship , computer science , theme (computing) , cluster analysis , informatics , value (mathematics) , information retrieval , similarity (geometry) , outlier , k nearest neighbors algorithm , world wide web , artificial intelligence , machine learning , medicine , electrical engineering , medical education , engineering , image (mathematics)
An Internship course becomes one of many compulsory subjects in Under graduate Program of Informatics Engineering in Ahmad Dahlan University, Yogyakarta.In the last few semesters, we found that some students were failed in taking this subject. After being identified, they were facing some obstacles such as determining the main theme for their job description. During this study, we proposed an application to classify the internship titles by using a technique in text mining called Shared Nearest-Neighbor and Cosine Similarity. From the result, we got values from the parameter K is 7, the epsilon value is 0.5, and the value of Mint t is 0.3 with 22 clusters and 0 outlier. These values presented that all data titles of internship activitiesareclassified into each cluster. 7 topics whichtook by majority of students are:1) Information Systems (7 titles);2) Instructional Media (5 titles);3)Archiving Applications (4 titles);4) Web Profile Implementation (3 titles); 5)Instructional Media for University Courses (3 titles); Multimedia (3 titles) and 6)Workshop & Training (3 titles).