Implementation of web product search engine

Abstract

In the recent years, web product search engine competition has been on the rise and each aims to provide users a better searching platform. Traditional search engine has been improved with the option to allow users to search using images and provide text suggestion to assist users in searching. In this project, a hybrid of textual and visual web product search engine is proposed.

A full-text search engine is equipped with comprehensive functions to handle various search queries. The search engine is further improved to handle multi-lingual queries, multiplicity search criteria and providing sorting mechanism. Basic concepts, such as feature extraction, features selection and nearest neighbour search, have been studied to assist in the implementation of image search. In improving the usability of search engine, an effective approach of using a joint textual-visual suggestion has been implemented.

In this project, the author shows that the proposed search engine is able to handle textual and image queries. When a user types a text query, a joint textual-visual suggestion list containing keywords and a collection of representative images is shown in a dropdown menu. When a user submits a text or image query, the search results are presented and user will be able to refine the search results according to their preference.