Abstract
Recently, online shopping is woven into the fabric of daily life.
In order to make these sites more customer-friendly, sites has recommender system.
These system oftenly uses collaborative filtering approach.
However, this approach has problem with recommending diverse categories, so that will make user boring.
In this research, to make a discoverable recommendation in comprehensive perspective such as category, defines dissimilarity between all 2 items on the basis of Browse Node ID, and use k-medoids method to make new categories.
Furthermore, makes a weighted complete graph, which uses categories as a node.
This indicates trend between different categories.
Based on weighted complete graph information, proposed system will show a recommendation and subserves conventional one.
To evaluate the effectiveness of the proposed method, subjective experiment will be conducted.
