Title:
FEATURE EXTRACTION DEVICE, FEATURE EXTRACTION METHOD, AND PROGRAM
Document Type and Number:
WIPO Patent Application WO/2010/150900
Kind Code:
A1
Abstract:
A feature extraction device is provided with a searching means for searching a document tree, and sequentially detecting elements as search elements; a distance calculation means for calculating an inter-element distance between an extraction target element within a plurality of elements of the document tree and a search element; an exclusive element confirmation means for referring to an exclusive element name and generating exclusivity information indicating, for an exclusive target element, whether the search element is the exclusive element; an element feature vector calculation means for calculating, based on an inter-element distance and the exclusivity information, a weight for a word included in an element corresponding to the element, and for relating and calculating, for each search element, based on weights, an element feature vector having a plurality of dimensions and such that each dimension uniquely corresponds to a predetermined word; and a partial document feature vector calculation means for calculating, based on the element feature vector, a partial document feature vector of a partial document related to the extracted target element.
More Like This:
JP4723901 | Television display device |
WO/2018/230355 | INFORMATION PRESENTATION SYSTEM |
JP4662620 | Resident record data / family register data search device |
Inventors:
TAMANO HIROSHI (JP)
Application Number:
PCT/JP2010/060918
Publication Date:
December 29, 2010
Filing Date:
June 21, 2010
Export Citation:
Assignee:
NEC CORP (JP)
TAMANO HIROSHI (JP)
TAMANO HIROSHI (JP)
International Classes:
G06F17/30
Foreign References:
JP2007316743A | 2007-12-06 |
Attorney, Agent or Firm:
ASAI, TOSHIO (JP)
Toshio Asai (JP)
Toshio Asai (JP)
Download PDF: