ABSTRACT
The research data can be grouped by the similarity of theme, object and method of the research. The results of grouping the research data can show how patterns of similarities and variations of the theme of research studies from time to time. The results of grouping can also reveal themes that many taken by students and themes that rarely taken by students at a certain time. These information is expected to help professors in evaluating the learning method that has been done. This research classifying paper documents of Chemistry Department, Faculty of Mathematics and Natural Sciences, University of March. Department of Chemistry has been chosen because it has pretty much amount of the data research.
Grouping of data research that is typically in the form of text can be done with text mining with clustering method. Clustering method used in this study is a combination of Hierarchical Clustering and K-Means Clustering methods. Reswarch data that were selected was paper documents. Part of the documents processed are part of the abstract.
Document clustering generate 16 clusters. Results were analyzed by the relevant between the documents of each clusters and the expected theme of each cluster. Results of cluster analysis showed that the research in the Department of Chemistry in the year 2009-2011 is limited to a few themes. In 2009 the theme that widely studied is complex anorganic, organic focusing on antifungal and antioxidant, and physical chemistry. In 2010 the theme that widely studied were organic. In 2011 the theme that studied were organic and anorganic. In 2012 and 2013 the theme of research in the Department of Chemistry is more varied than in previous years.
Keyword: thesis abstract, clustering, hierarchical clustering, k-means clustering, text mining.
ABSTRAK
Data penelitian dapat dikelompokkan berdasarkan kemiripan tema, objek maupun metode penelitian. Hasil pengelompokkan data penelitian dapat memperlihatkan bagaimana pola kemiripan penelitian dan variasi tema penelitian dari waktu ke waktu. Hasil pengelompokan juga dapat memperlihatkan tema yang banyak diambil mahasiswa dan yang jarang diambil mahasiswa pada waktu tertentu. Informasi tersebut diharapkan dapat membantu dosen dalam mengevaluasi metode pembelajaran yang telah dilakukan. Penelitian ini mengelompokkan dokumen skripsi Jurusan Kimia, Fakultas Matematika dan Ilmu Pengetahuan Alam, Universitas Sebelas Maret. Jurusan Kimia dipilih karena jumlah data penelitiannya cukup banyak.
Pengelompokan data penelitian yang umumnya berbentuk teks dapat dilakukan dengan text mining dengan metode clustering. Metode clustering yang digunakan pada penelitian ini adalah kombinasi antara metode Hierarchical Clustering dan K-Means Clustering. Data penelitian dipilih dokumen skripsi. Bagian dari dokumen yang diolah adalah bagian abstrak skripsi.
Clustering dokuman menghasilkan 16 cluster. Hasil cluster dianalisa keterkaitan antar dokumennya dan diperkirakan tema dari tiap cluster. Hasil analisa cluster menunjukkan bahwa penelitian di Jurusan Kimia pada tahun 2009-2011 terbatas pada beberapa tema. Tahun 2009 tema yang banyak diteliti adalah anorganik kompleks, organik dengan fokus antijamur dan antioksidan, dan kimia fisik. Tahun 2010 tema yang banyak diteliti adalah tema organik. Tahun 2011 tema yang diteliti adalah organik dan anorganik. Pada tahun 2012 dan 2013 tema penelitian di Jurusan Kimia lebih bervariasi daripada tahun-tahun sebelumnya.
Kata kunci : abstrak skripsi, clustering, hierarchical clustering, k-means clustering, text mining.
