An Intelligent Shopper’s Assistant

Partha Sarathi Dutta, Master’s Student, Computer Science, Sandip Debnath, Sandip Sen

 

Shopping on the Internet has become a convenient way of purchasing commodities of choice. With the boom of e-commerce, online sellers are flooding the market with their products. Personal agents have been developed that assist users by providing need-specific information, that are obtained by generating, filtering, collating, or transforming information available on the Internet. We believe in an alternative approach where such personalized software agents can assist users to modify queries. This maximizes the quality of retrieved information and could be considered to be a key enhancement of agent technology in online shopping, which in most of the cases merely relies on simple information retrieval. A consumer interested in a particular product class, e.g., a consumer electronics item, can use such an agent to locate a desired product fitting a specified budget from one of several online stores. Such an assistant can not only suggest the cheapest price for a product, but also can suggest a higher quality alternative for a slightly higher price than the limit specified. We call this functionality “query reformulation”: modification and augmentation of a user query with the objective of maximizing user satisfaction.

 

The highlights of this research project are the following:

q                   Identifying the types of query reformulations: We have identified three different categories of context-sensitive reformulation techniques, which guarantee increased customer satisfaction than without employing them.

q                   Constructing domain ontology: We have designed an ontology representing the domain of consumer electronics goods. Quality of query reformulation is enhanced using the ontology as a buttress to the system logistics.

q                   Identifying domain types: Our comparative analysis of different domains with different characteristics reveals some important domain features that are conducive for the usage of reformulations a powerful tool to enhance online shopping experience.

 

We have a working system that implements the above-mentioned functionalities. In the present form, it supports queries in the domain of consumer electronic goods. The system is available for evaluation at: http://www.mcs.utulsa.edu/~partha/CHAYANI