SIR Computer Science Investigation Abstract

DEVELOPMENT OF INTELLIGENT SEARCH SOFTWARE

Presenters:

Ovidiu Hentea, Illinois Mathematics and Science Academy, 1500 W. Sullivan Road, Aurora, IL 60506

Daniel Moorehead, Illinois Mathematics and Science Academy, 1500 W. Sullivan Road, Aurora, IL 60506

Brian P. Sawicki, Illinois Mathematics and Science Academy, 1500 W. Sullivan Road, Aurora, IL 60506

Mentor:

Dr. Jay Budzik, Intellext

Abstract:

One of the most rapidly advancing areas of study in information sciences today is intelligent search and search algorithm design. At Intellext we have been given insight into modern search technologies while actively developing and implementing new search tools. Development and research was done for Intellext’s Watson search, a proactive search assistant that interfaces with common desktop applications to provide information relevant to a user’s task. Research included comparison of elements of the effects of Watson’s statistical search tool by means of an ablation study in which one or more of the heuristic search components had been removed. Subjects were asked to complete a web-based survey backed by an SQL database (used to store results) comparing results for documents generated by the differing search algorithms. This allowed for an accurate judgment of the contributions as to different functions within the search algorithm. The study was crucial to the development of the search-tool due to the information it would offer, such as the relevance of each of the elements in the searches sent out by the tool. Other projects included a detailed study of potential competition in today’s search tool market, in the interest of acquiring information regarding competitors and potential partners that might be able to offer the technology needed to complement the stature of the Watson search engine in the marketplace segmentation. We also did a brief research project on the variety of desktop search-tools on today's market and their versatility in terms of integration with Watson. A final work in progress involves writing a plugin to allow Watson to interface with the Mozilla Firefox © web browser.