Computer systems that can directly and accurately answer peoples' questions over a broad domain of human knowledge have been envisioned by scientists and writers since the advent of computers themselves.

Consider, for example, the computer on Star Trek – how it understands questions and can quickly provide accurate, customized answers and can engage in a fluent information seeking dialog with the user. We call this technology open domain question answering and it has tremendous promise for impacting society and business. 

Growth of Natural Language Text
With the staggering growth of human knowledge expressed in natural language text, the possibilities and challenges are now immense. Imagine interacting with an expert computer system as fluently and naturally as you would a human expert to gain just-in-time knowledge, precise answers to relevant questions and to help discover key insights based not on what one human can recall but rather on the entirety of human knowledge. 

While researchers in computer science and Artificial Intelligence (AI) have investigated this area for over 40 years, general purpose solutions remained elusive. Recent technical advances, however, are giving new promise to the quest. Applications in business intelligence, healthcare, customer support, enterprise knowledge management, social computing, science and government would all benefit from such technology.

The DeepQA project at IBM research (www.ibm.com/deepqa) is developing a computer system intended to greatly advance the start-of-the-art in automatic opendomain question answering. The project is developing massively parallel architecture capable of dealing with enormous expressivity and complexity of natural language through the novel integration and advancement of natural language processing, information retrieval, statistical machine learning and formal knowledge representation and reasoning.

Watson takes Jeopardy! Challenge
Currently driving the project is the Jeopardy! Challenge. The goal is to apply DeepQA to develop a computer system, code-named Watson, that can successfully compete against a champion player in the well-known Jeopardy! quiz show. The challenge demands extraordinary levels of precision, confidence and speed over an incredibly broad domain of knowledge. 

Competing against humans at Jeopardy! will require Watson to accurately answer richly expressed open-domain questions, distinguish between relevant and irrelevant content, score and justify answers with pin-point precision and understand what it knows so that it can determine the likelihood that its answers are correct (before it buzzes in). Watson’s confidence is computed as a accurate probability that its candidate answer(s) are correct. This is key for winning Jeopardy!, as wrong answers are severely penalized, and a tremendous advantage  in real business applications.

Natural Engagement
Ultimately I believe that DeepQA will inspire and enable a remarkable new class of business applications that will change the way computers interact with people. It promises to engage users more naturally and to deliver precise answers and explanations through interactive and intelligent dialog based on deeper understanding of free-form natural language queries and content.