Abstract: An analytics platform for the extraction of structured observations from largely narrative sources using a hybrid approach of user configuration and machine learning is provided. The analytics platform collects and normalizes data from public and private sources and applies extractions to the data to create a world view of objects, traits, and relationships of interest and maintains that world view as data and/or extractions are updated. The platform is further configured to apply queries to the extracted world view for a variety of purposes including scoring objects for prioritized attention, generating notifications when specific conditions are met, providing data sets for exploratory analysis, and triggering the automatic collection of enhancing data from external sources.

Abstract: An analytics platform for the extraction of structured observations from largely narrative sources using a hybrid approach of user configuration and machine learning is provided. The analytics platform collects and normalizes data from public and private sources and applies extractions to the data to create a world view of objects, traits, and relationships of interest and maintains that world view as data and/or extractions are updated. The platform is further configured to apply queries to the extracted world view for a variety of purposes including scoring objects for prioritized attention, generating notifications when specific conditions are met, providing data sets for exploratory analysis, and triggering the automatic collection of enhancing data from external sources.

Abstract: Fraud losses in a communication network are substantially reduced by automatically generating fraud management recommendations in response to suspected fraud and by deriving the recommendations as a function of selected attributes of the fraudulent activity. More specifically, a programmable rules engine automatically generates recommendations based on call-by-call fraud scoring so that the recommendations correspond directly to the type and amount of suspected fraudulent activity. Using telecommunications fraud as an example, an automated fraud management system receives call detail records that have been previously scored to identify potentially fraudulent calls. Fraud scoring estimates the probability of fraud for each call based on the learned behavior of an individual subscriber as well as that of fraud perpetrators. Scoring also provides an indication of the contribution of various elements of the call detail record to the fraud score for that call.

Abstract: Fraud losses in a communication network are substantially reduced by automatically and selectively invoking one or more authentication measures based on a fraud score that indicates the likelihood of fraud for that particular call or previously scored calls. By selectively invoking authentication on only those calls that are suspected or confirmed to be fraudulent, fraud prevention can be achieved in a way that both reduces fraud losses and minimizes disruptions to legitimate subscribers. Using telecommunication fraud as an example, a subscriber is registered in a system by collecting data on that subscriber based on the particular authentication method being used, such as shared knowledge (e.g., passwords), biometric validation (e.g., voice verification), and the like. Once registered, the authentication function for the subscriber's account is activated and subsequent calls are then scored for the likelihood of fraud during the call setup request phase.