The large volumes of text contained in big data is data prepared by people in order to communicate information or their intentions to other people, and as such, it is a valuable source of information. NEC is developing technologies to extract customer opinions and reputation reports from large volumes of text data, to use in marketing, corporate risk management, client management, and other such applications.
In particular, NEC Central Research Laboratories is developing technology in the four areas of: (1) language analysis technology that can handle free expressions such as conversational phrases; (2) synonym and connotation judgment technology that can determine whether words or phrases have the same meaning; (3) meaning and connection extraction technology to extract meanings and connections; and 4) retrieval technology and monitoring technology that integrates diverse information sources based on extracted meaning.
NEC core technologies: Recognizing textual entailment
In text data, there are many cases where content that has the same meaning is expressed using different phrases. Technology to recognize whether different phrases have the same meaning is known as "recognizing textual entailment."
NEC Central Research Laboratories has developed technology to accurately determine whether two phrases have the same meaning, by considering the importance of words within the phrase and the structure of the phrase including the subject, predicator, etc. Our technology in this area boasts world-beating accuracy, and in 2011, at the "TAC2011 RTE-7" evaluation workshop held by the U.S. National Institute of Standards and Technology (NIST), NEC's technology was ranked No. 1.
Application example of recognizing textual entailment technology
At present, NEC is conducting R&D into using this technology in marketing and corporate risk management, and we are currently developing a range of solutions such as identifying reputations from cyber information and retrieving meanings to improve the efficiency of contact center operations.
We have recently deployed this technology in our "Compliance Enhancement solution" that can be used to accurately identify which documents require management from amount large amounts of documents such as emails and sales reports. Sumitomo Mitsui Banking Corporation is using this solution to automatically analyze and judge documents generated by several thousands of sales staff. It is proving fruitful for them in realizing enhanced governance while enabling all important information to be extracted.
November 11, 2015
NEC announces enhancement of its AI-related business
Data Mining TG supports the realization of Solutions for Society using Only 1, No.1, AI-related technologies (heterogeneous mixture learning, recognizing textual entailment, predictive robust optimization framework). Please see the press release for details.
November 18, 2014
NEC has developed an "RTE-based Clustering Technology" as a big data analysis technology for automatically grouping vast amounts of text data according to meaning then indicating the general semantic content of each group of sentences in the form of a headline. This technology clarifies the standpoint of grouping, which was formerly unclear. This enables, for example, the early discovery of products to be recalled through an analysis of complaints made to a call center, trend analysis through social media analysis, and prompt responses by local governments or police through the automatic organization of disaster information and information provided in calls for service.
Please see the press release for more details on this technology.