Extracting joint topic-sentiment models from text inputs
US11238243B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Sep 27, 2019 |
| Grant date | Feb 1, 2022 |
| Priority date | — |
| Expiry date | Aug 4, 2040 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG16H15/00
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
There is a need for solutions for more effective and efficient natural language processing systems for short texts. This need can be addressed, for example, by a system configured to obtain an initial term-topic correlation data object for a plurality of digital documents, obtain a user-defined term-topic correlation data object for the plurality of digital documents, generate a refined term-topic correlation data object and a refined document-sentiment correlation data object for the plurality of digital documents based at least in part on the initial term-topic correlation data object and the user-defined term-topic correlation data object, obtain a user-defined document-topic correlation data object for the plurality of digital documents, and generate a refined document-topic correlation object for the plurality of digital documents based at least in part on the refined term-topic correlation data object and the user-defined document-topic correlation data object.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.