发明名称 INFORMATION EXTRACTION DEVICE, INFORMATION EXTRACTION METHOD AND INFORMATION EXTRACTION PROGRAM
摘要 <P>PROBLEM TO BE SOLVED: To enhance accuracy in a browsing condition by giving a high hidden condition degree to an important word, without giving a high hidden condition degree to an unimportant word among words of an EC site, etc. <P>SOLUTION: In an information extraction device 1, extraction words from a commodity detail page of a commodity list presented by an EC site are inputted through a commodity information input device 5. A common word extraction part 2 of the device 1 extracts words with high appearance frequency among the extraction words as common words. A specific word extraction part 3 gives a high score to words which appear more frequently in browsed commodity detail pages than in commodity detail pages which are not browsed by a user among the commodity list, and extracts the words with the high score as specific words. A hidden condition degree calculation part 4 acquires appearance frequencies of a dependency structure in which dependency destination is a specific word and dependency source is a common word, by referring to an all commodity database 7, and calculates hidden condition degrees of the specific words using the acquired appearance frequencies. The specific words with the calculated high hidden condition degrees are outputted on a hidden condition output device 6 as hidden conditions. <P>COPYRIGHT: (C)2013,JPO&INPIT
申请公布号 JP2013054558(A) 申请公布日期 2013.03.21
申请号 JP20110192790 申请日期 2011.09.05
申请人 NIPPON TELEGR & TELEPH CORP <NTT> 发明人 EZAKI KENJI;UCHIYAMA MASASHI;SUGIZAKI MASAYUKI;SEKIGUCHI YUICHIRO
分类号 G06F17/30;G06F17/21 主分类号 G06F17/30
代理机构 代理人
主权项
地址
您可能感兴趣的专利