Our signal detection and fact extraction subsystem is part of our Answer bot and Conversational engine platform and can be used as a standalone system as well. This system has the ability to identify signals in vast amounts of unstructured data and extract facts associated with such signals. As an example, if the goal were to extract all M&A related information from millions of documents, the AI system can learn to identify which documents mention related information on M&A (the signal) and once these documents are identified, all facts associated with this signal are extracted. Facts like acquirer(s), acquired company and the price of acquisition. Our neural network based classification engine can classify millions of documents based on topics covered in the document.
The system can be trained to identify any user-defined signals and extract facts associated with the signal as defined by the user. Unlike systems that do entity and relationship extraction and tagging, our approach can detect complex signals and facts in large amount of unstructured data. We help enterprise customers to define the signals they are interested in tracking. This capability is being used to monitor events and information related to companies, investments, and so on, as well as tracking analyst recommendations, tracking relevant signals in earnings reports or CEO communication. Classification helps in personalizing content to both end consumers and other large enterprise content consumers.