Community-based reporting and analysis system and method
US11030410B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Jan 13, 2020 |
| Grant date | Jun 8, 2021 |
| Priority date | — |
| Expiry date | Jan 13, 2040 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06F40/30
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A computer-implemented method for training a neural network system to classify data items as true mentions of relevant data objects includes, based on a domain of interest, a processor controlling an input component and a data store to receive and store a document related to the domain of interest, control a natural language processor to extract data items related to the domain of interest from the document, create a first training data set of selected ones of the data items, using the first training data set, train the neural network to identify true mentions of data objects relevant to the domain of interest, apply tweaks to one or more of the data items to create a list of tweaked data items, creating a second training data set including the tweaked data items, and using the first and second training data sets, further train the neural network to identify true mentions of relevant data objects.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.