CONVERSATIONAL RECOMMENDER SYSTEM IN CHOOSING SMARTPHONE WITH FUZZY MULTI-CRITERIA DECISION MAKING APPROACH

• Main Author: Indra Nainggolan, Bobby
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/39182
• Institution: Institut Teknologi Bandung

A conversational recommender system has been developed that is able to mimic conversations like buyers and professional sales support in product search. This system uses ontology as its knowledge base. This recommendation system utilizes the concept of navigation by asking and navigation by proposing in generating interactions. However, the system turned out to only use the mandatory input on product selection. Mandatory input is not used in the product sorting process. The system only calculates optional input / utility values to sort the product, this results in the system being unable to sort products in several conditions. Fuzzy multi-criteria decision making is one of the solutions that can be used to overcome this problem. All user’s input can be used to build a matrix of relative importance criteria that can be computed with smartphone specifications to produce values that can replace utility values. From the test results it was found that this concept succeeded in optimizing all user inputs in providing recommendations. The order of recommendations produced is also better than the previous recommendation system.