APLIKASI PENGELOMPOKAN HASIL PENCARIAN DOKUMEN BERITA MENGGUNAKAN ALGORITMA BISECTING K-MEANS; APPLICATION OF NEWS DOCUMENT SEARCH RESULTS CLUSTERING USING BISECTING K-MEANS ALGORITHM

ALFAJRI, ERRIC, Anny Kartika Sari

2016 | Skripsi | FMIPA

Search for news document is common activity of internet users. To reach the information, user can use internet search engines, which can return search results depends on the keyword. Finding the information needed on the search results can be a problem, because several of subtopic are mix together in a long list of search results. Clustering feature can be implemented in search engine to present the search results in several clusters. In this research, a search engine application of news document is build. A search result clustering feature is added to the search engines. There are news sites which are used as sources for searching news documents. Google Custom Search is used for searching documents, then clustering is implemented on the search results using bisecting k-means algorithm with a library supplied by carrot2. This research produces a search engine application of news documents which can present search results based on clusters. Each cluster is represented by a label cluster. The evaluation was done on the clustering results, found that the precision is 0.57 and the cluster label quality is 0,78.

Kata Kunci : clustering, bisecting k-means, carrot2, Google Custom Search.