Abstract:
This research aimed to (1) create a semantic knowledge base using ontology to recommend designs for processed tamarind packaging, (2) develop an algorithm and web application for recommending tamarind packaging designs using content-based recommendation techniques based on ontology, and (3) evaluate the algorithms performance and user satisfaction with the packaging design recommendation
web application. The research is divided into three phases: (1) a survey of 384 consumers preferences regarding processed tamarind packaging and synthesizing knowledge from color theory and design principles to build the ontology knowledge base; (2) the development of a content-based recommendation algorithm on ontology using Python, connected to a web application developed with PHP, HTML, CSS, Ajax, and jQuery with the algorithm implemented through a RESTful API developed with the Flask library; and (3) an evaluation of the algorithms performance by comparing the packaging design recommendations with those from experts and assessing the satisfaction of 30 users, with results analyzed through means and standard deviations. The results found that the ontology knowledge consists of a total of 26 classes, divided into 3 levels of knowledge. The web application consists of functions for receiving product images and descriptions to create meaningful packaging design knowledge. It also provides recommendations by ranking packaging designs that are suitable for the product. The algorithm can recommend packaging designs accurately according to expert suggestions, achieving an accuracy rate of 85.18%. User satisfaction with the web application is rated at a high level (=3.52 SD=0.646).