From what I see, US patent 6,904,428 (by the Illinois Institute of Technology) is a search algorithm process. Any ideas why IBM thought they needed it?

An intranet mediator for obtaining direct answers to natural language questions allowing users to search both a data warehouse of integrated/structured data sources and unstructured data sources. The intranet mediator allows the user to obtain an answer to a natural language question without having to surf the data sources in which the answer might be contained, or without being limited to one specific factual item return. The intranet mediator operates on the supposition that most answers to business queries are contained within structured data sources which have been integrated into the data warehouse thereby having common schema and known contents. Preselection of the most relevant data source(s) is thus possible before query output. Search of unstructured data is also performed for additional context surrounding either the question or the answer. A direct answer is given in response to the question. If desired, the intranet mediator may also display a list of data sources where additional relevant information may be found.

2 Answers 2

  1. A method of digital data gathering for providing a direct answer to a natural language question, comprising:

a) accepting input of a natural language question;

b) identifying the relevant concepts of the natural language question;

c) assembling the relevant concepts of the natural language question into a query

d) identifying, via a meta-data source for a physical data warehouse, a data source in the physical data warehouse likely to contain an answer to the query;

e) performing a first search of the query in the physical data warehouse;

f) performing a second search of the query in an unstructured data source not contained in the physical data warehouse;

g) integrating the results of the first and second searches and selecting a direct answer to the natural language question; and

h) displaying the direct answer to the natural language question.

Sounds an awful lot like IBM's Watson.

(For those of you who don't know, it's IBM's natural language processing AI that IBM showcased on Jeopardy about two years ago. The computer beat Jeopardy allstar Ken Jennings handily. From what I've heard, IBM has plans to use the service for processing all sorts of things, including diagnoses in medicine and document review in lawsuits.)


I was studying Natural Language Processing in 2002 so I remember that there was a lot of feverish work being done in those years due to Google's sudden emergence and popularity. Neither of the authors are particularly notable so although their patent sounds pretty cutting edge for those years it doesn't mean that their results were necessarily the best. One thing I notice about the patent is that it seems to cover an overall design and not methods that produce improved outcomes. In NLP, it's really all about the outcome. So basically, the patent says "hey! we're claiming NLP searches on a combination of structured and non-structured data! Who care that our software only has an 80% accuracy rate!" I'm going to guess that IBM purchased the patent because of what it claimed and how early it claimed it, not what it actually produced. To IBM's credit, they actually backed up the claim with a whiz-banger product a few years later.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .