Method and system for co-occurrence-based text conversion
US8600729B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Dec 8, 2010 |
| Grant date | Dec 3, 2013 |
| Priority date | — |
| Expiry date | Dec 19, 2031 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06F40/55
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A text conversion method and a text conversion system are provided. A term mapping table recording a term mapping relationship between a source language and a target language is provided. A tokenization process is performed on a paragraph in the source language to obtain tokenization results. The tokenization results are compared with the term mapping table to determine each source language term in the paragraph is belonging to a first type or a second type. The source language terms belonging to the first type are converted into corresponding target language terms according to the term mapping table. Regarding each source language term of the second type, one of multiple corresponding candidate target language terms is selected as the target language term according to a co-occurrence relevance of relevant terms, wherein each relevant term is constituted by one candidate target language term and words before and after that in the paragraph.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.