Designing a cross-language comparison-shopping agent

This research pertains to the design and development of a shopbot called WebShopper+. This shopbot is intended to help customers find and compare e-tailers that market their wares using different languages. WebShopper+ is built with a multilingual ontology to overcome the language barriers that arise with global e-commerce. This research proposes a semi-automatic method of constructing a multilingual ontology by using the formal concept analysis and association analysis. It also proposes an automatic method for the categorization of product data into predefined classes, with the aim of alleviating administrators' task load. Additionally, a semantic search mechanism based on concept similarity is designed to assist customers in finding more desirable products. The experimental results show that these methods perform well and the shopbot can help customers find real bargains on the Web and to find products that cannot be bought locally.

Research Highlights
► We develop a shopbot that can help customers to locate real bargains on the Web and potentially purchase products that are not available in their own countries.
► The semi-automatic method of constructing a multilingual ontology is capable of achieving high precision and coverage.
► The data classification method is able to automatically and correctly classify product data into said ontology.
► The semantic searching mechanism can help customers search for products using their native languages.