Intelligent matching and linking of organization data from different sources

The project will develop technologies for automatic matching and linking of information about legal organizations from various datasets. We will use AI technologies: machine learning, semantic modeling, data integration, logical inference and validation. This will result in new functionality added to our leading products: Ontotext Platform and Cognitive Cloud. This will improve the company's competitiveness and will increase the market of these products. Ontotext Platform is a technology for cognitive analysis of data bases and information. The platform enables the integration of structured information into huge Knowledge Graphs comprising information about millions of concepts, entities (people, organizations, places, products) and the relations between them. Entity descriptions augment each other and can be analyzed using cognitive methods, similar to those used by the neural net of the human brain.Theplatform calculates, for example, measures of importance and similarity between entities and concepts and recognizes implicit relations between companies.Knowledge Graphs include rich context and enable "understanding" of the concepts, which allows their precise recognition and disambiguation from similar concepts in text. Such knowledge enables the Ontotext Platform to "read" text, perform semantic indexing, and to extract new knowledge.The platform is already in use by a number of world leaders in the publishing and information business, such as Financial Times, Standard and Poor’s, Nikkei. Its wider-scale use is limited by the large effort required for creating Knowledge Graphs in specific areas, which increases the time and cost required for deployment. The goal of this project is to enrich the platform with a technology for automating the creation of Knowledge Graphs through the intelligent matching and linking of company data from several sources. This will enable the widespread deployment of the Ontotext Platform in many more application domains.

Comments