Title:
DOCUMENT DATA PROCESSING METHOD AND DOCUMENT DATA PROCESSING SYSTEM
Document Type and Number:
WIPO Patent Application WO/2021/064510
Kind Code:
A1
Abstract:
The present invention makes it possible to enter natural language as query text and to search a plurality of documents, and presents, to a reader, portions highly related to the entered text. This document data processing system comprises: a document reading unit which reads a plurality of target documents; a document dividing unit which divides each of the plurality of target documents into a plurality of blocks; a first distributed representation acquisition unit which acquires distributed representations of the words in each block; a first distributed representation retention unit which stores, for each target document and for each block, the distributed representations acquired by the first distributed representation acquisition unit; a query text reading unit which reads a query text; a second distributed representation acquisition unit which extracts the words included in the query text, and acquires distributed representations of the words; a second distributed representation retention unit which stores the distributed representations acquired by the second distributed representation acquisition unit; and a similarity degree calculation unit which compares the distributed representations of the words included in the query text with the distributed representation of the words included in each block, and calculates the degrees of similarity between these blocks.
More Like This:
Inventors:
YAMAMOTO KUNITAKA (JP)
HIGASHI KAZUKI (JP)
DOZEN YOSHITAKA (JP)
HIGASHI KAZUKI (JP)
DOZEN YOSHITAKA (JP)
Application Number:
PCT/IB2020/058810
Publication Date:
April 08, 2021
Filing Date:
September 22, 2020
Export Citation:
Assignee:
SEMICONDUCTOR ENERGY LAB (JP)
International Classes:
G06F16/33
Foreign References:
| JP2019082931A | 2019-05-30 |
Other References:
HONMA, YUKINORI ET AL.: "Proposal of partial document retrieval method considering document structure", IPSJ TECHNICAL REPORT, vol. 2017 -SL, no. 26, 8 May 2017 (2017-05-08), pages 1 - 6, Retrieved from the Internet
PADIGELA, H. ET AL.: "Investigating the successes and failures of BERT for passage re-ranking", 5 May 2019 (2019-05-05), pages 1 - 5, XP081272487, Retrieved from the Internet [retrieved on 20201019]
PADIGELA, H. ET AL.: "Investigating the successes and failures of BERT for passage re-ranking", 5 May 2019 (2019-05-05), pages 1 - 5, XP081272487, Retrieved from the Internet
Download PDF: