Pachacutec
Veteran X
IBM researchers are developing what they call the third generation in search technology that will search according to the underlying meaning, and not simply by keywords, page rank, or other peripheral indicators. IBM plans to employ natural language processing for "discovery systems" that can search through files no matter whether they are structured as Web documents, databases, images, video, or audio. Although search technology has advanced significantly in just the past several months, especially with Google committed to new features and products every few weeks, the inherent limitations of Google-style search are apparent when trying to answer nuanced questions: For instance, though most Web searches are proficient in finding Mozart's birthplace, trying to find out how many Web pages are published in each language requires some extra effort on the part of the user. IBM researchers recently showed off their developing technologies at the Thomas J. Watson Research Center, including a system called Piquant that was able to answer questions from an article, even though the answers were not phrased exactly within the article. Piquant analyzed the semantic structure to answer "Who is Canada's prime minister?" even though those exact words were not used in the text. The Semantic Analysis Workbench automatically translates data to reveal useful patterns, and the researchers gave several examples to show how this would be useful. Customer service centers could analyze customer service notes to identify trends or problems, genomic research could benefit from unexpected correlations, and terrorist plots could be uncovered through the analysis of suspect telephone calls. Another technology would work in a manner similar to Google News, but return only headlines from non-English publications and provide a rough English translation upon request.
reg required : http://www.nytimes.com/2004/12/26/business/yourmoney/26techno.html