by:  Frederique Segond, Xerox Principal Scientist, XRCE

Millions of people were able to enjoy the benefits of natural language processing when they saw IBM’s Watson star on Jeopardy.  It begs the question – what is natural language processing? And, what else can linguistics do that justify corporations like IBM and Xerox investing their R&D in it?

As  a researcher in the field for many years at Xerox’s research lab in France, I could go on and on about how NLP works and  the decades of significant breakthroughs that have led us to where we are today.  Yet,  I’ll save that for another post for another day. What I would like to share is how NLP can be used in business today. Let’s take the case of text analysis and opinion mining in the fields of business, medicine and law (e-discovery).
NLP is helpful to analyze the opinions of Internet users on products, political events, or the introduction of new laws or candidates, to give a few examples.   Or let’s say manufacturers of video games want to know what users like and dislike about their games so that marketing can respond at lightning speed. Or they may want to know what features have technical issues to provide feedback to the development teams. What applies to video games applies to cars, printers, and any other product or service.
In the medical field, NLP tools are beginning to play an important role and this will significantly increase with the spread of electronic patient records, and archived medical records that have been digitized. Examples of applications include the ability to detect side effects of drugs, Hospital Acquired Infections or even links between diseases to advance epidemiological studies.

NLP is very effective for e-discovery in the legal domain, where attorneys often deal with tens of millions of documents to prepare for a case.  NLP helps filter those documents and  extract significant information that then be stored in a knowledge base to  support the case.

While NLP was behind Watson’s win – it is also behind a number of future applications that will help us get down to our real business at hand.  Perhaps I can explore some of these future NLP applications in future posts.