Search Query Clustering Comparation On E-Commerce Using K-Means And Adaptive DBSCAN
Darwin, Ronsen Purba, Muhammad Fermi Pasha · 2020
Search queries on electronic commerce vary greatly depending on user behavior and preference. They are funneled through queries that imply buying interest. The accuracy of the search query clustering comparison can be enhanced through either k-Means or adaptive DBSCAN which had not been conducted by previous studies. Clustering helps to derive business knowledge models, especially taxonomic search features. This study used dataset which consisted of 2.074 records as a result of pre-processing, the accuracy was obtained 92.63% using adaptive DBSCAN and 91.75% using k-Means. The taxonomic results help to produce more informative query outputs that can be used to improve the appearance of electronic commerce and search features. Thus, increase of customers satisfaction and conversions can be achieved.