Japan, July 14 -- TOSHIBA CORP has got intellectual property rights for 'DOCUMENT CLASSIFICATION APPARATUS, METHOD, AND PROGRAM.' Other related details are as follows:
Application Number: JP,2022-193031
Category (FI): G06F16/355,G06F16/35,G06F16/383
Stage: PROBLEM TO BE SOLVED: To provide a document classification apparatus having a document classification model effectively utilizing a supplementary sentence formed of a short text, and a method and a program.SOLUTION: A document classification apparatus 100 includes a document input unit, a supplementary sentence input unit, a word statistic calculation unit, a word attribute provision unit, an anchor word extraction unit, and a word embedding unit. The document input unit divides a sentence included in a target document into word units. The supplementary sentence input unit divides a sentence included in a supplementary sentence indicating a feature of a classification item, into word units. The word statistic calculation unit calculates an appearance statistic of a word obtained from the target document and the supplementary sentence. The word attribute provision unit provides, to the word, an attribute representing at least one of specialty, specificity and category independence on the basis of the appearance statistic. The anchor word extraction unit extracts an anchor word on the basis of the attribute. The word embedding unit performs target learning in accordance with the attribute of the anchor word and generates a word embedding vector.SELECTED DRAWING: Figure 1 (Grant)
Filing Date: Dec. 1, 2022
Publication Date: June 13, 2024
The original document can be viewed at: https://www.j-platpat.inpit.go.jp/p0100
Disclaimer: Curated by HT Syndication.
