IMI: INTELLIGENCE & MANAGEMENT OF INFORMATION INC.
发明人:
PROUZET, Eric Pierre
申请号：
WO2016IB55910
公开号：
WO2017081562(A1)
申请日:
2016.10.03
申请国别(地区):
世界知识产权组织国际局
年份:
2017
代理人:
摘要：
A method of processing a document for searching includes obtaining document text, and generating streamlined document text. The streamlined document text is generated by: (i) discarding a plurality of strings from the document text that match any of a plurality of preconfigured low-relevance strings to generate condensed document text; (ii) in the condensed document text, replacing a plurality of content strings with respective ones of a plurality of preconfigured content class identifiers. The method further includes determining respective frequency values indicating the frequency of the content class identifiers in the streamlined document text; determining a proximity value for at least one pair of the content class identifiers in the streamlined document text; and storing a subset of the frequency values and the proximity value in the memory.
