University of Chicago Infolab FAQFinder

Overview

The University of Chicago's Infolab develops intelligent web agents designed to help users find information. Their applications, developed by using Allegro CL, range from Web-based rental property locators and movie selectors to a complex FAQ system.

Example: FAQFinder

A knowledge-based information retrieval application developed by using Allegro CL, FAQFinder is a question-and-answer system that uses a natural language interface to search for information on FAQ lists found on the web. The FAQ lists are derived from USENET News Groups and comprise the groups' collective knowledge regarding popular topics. A user may ask a question in natural language, and FAQFinder will provide the FAQ file most likely to yield the answer, search through the file for similar questions and present the answers ordered according to likelihood of a match.

For example, a user might want to know what the current discussion is regarding the addictive nature of caffeine. The inquirer would simply type: "Is caffeine addictive?" and click on "Find Answer." The system retrieves the following series of files most likely to yield the desired information:

• Caffeine_FAQ
  coffee and caffeine's frequently asked questions
• Drink_tea_FAQ
  rec.food.drink.tea FAQ
• Drugs_rec_drugs_FAQ
  rec.drugs [FAQ]
• Etc.

Then, by clicking on the first item, the system presents a series of questions and answers:

1. How do you pronounce mate?
2. Caffeine and your metabolism
3. Studies on the side-effects of caffeine
4. How much theobromine/theophylline there is in...?
5. How does caffeine taste?

By further selection, a final answer is achieved. For example, choosing # 3 yields a press release concerning the effects of caffeine on one's health.

Try your own questions at http://www.ics.uci.edu/~burke/faqfinder/.

© 2008 Franz Inc - Privacy Statement
[ Consulting Services | Packages/Pricing | Allegro NFS | Certification Program ]