UBIC Obtains New Japanese Patent Related to Word Co-Occurrence Analysis for Keyword Extraction

Thursday, January 22, 2015 - 10:52

UBIC, a leading provider of international litigation support and big-data analysis services, has announced that the Japan Patent Office has recently issued a Notification of Decision of Grant to the company for a patent application related to "word co-occurrence analysis." The new statistical framework developed by the company strategically improves the process for extraction of important data for e-discovery and fraud investigation by analyzing and estimating the probabilities of joint occurrence of keywords in a text.

The necessary procedures for patent registration have already been executed, and the patent is currently pending issuance. (Patent application number: 2013-130766, patent application filing date: June 21, 2013.)

A keyword search frequently reveals that one keyword will likely carry different emphasis if it occurs together with another keyword. Consider, for example, a cartel investigation with (1) "price" occurring on its own in a document, versus (2) "price" co-occurring with another word such as "fixing" or "manipulate." The usage of the word "price" in example 2 is generally regarded as a more significant and important occurrence than in example 1. Analyzing word co-occurrences in this way will help enable users to find data with a higher degree of relevance as potential evidence.

The analytical framework covered by this patent has been implemented in all products of Lit i View(R), a platform for big data analysis.

The Lit i View(R) products have been designed not only to process vast amounts of e-discovery data automatically, but also to meet a wider range of new industry demands arising in the age of big data. For example, UBIC's innovative technology also provides state-of-the-art fraud investigation features and an automated email auditing system.