Abstract: Systems and techniques are described for applying machine learning techniques to dynamically identify potentially anomalous activity of entities. In some implementations, peer group data is obtained. The peer group data indicates multiple entities classified as belonging to a particular peer group, and a set of attributes associated with the multiple entities. Transaction data for the multiple entities is obtained from one or more data sources. One or more transaction models are selected. The transaction models that are each trained to apply a particular set of evidence factors corresponding to the set of attributes associated with the multiple entities, and identify transaction patterns representing potentially anomalous activity. The transaction data is processed using the one or more transaction models to identify potentially anomalous activity within the transaction data for the multiple entities. A prioritization indicator is computed for each entity included in the multiple entities.

Abstract: Systems and techniques are described for applying machine learning techniques to dynamically identify potentially anomalous activity of entities. In some implementations, peer group data is obtained. The peer group data indicates multiple entities classified as belonging to a particular peer group, and a set of attributes associated with the multiple entities. Transaction data for the multiple entities is obtained from one or more data sources. One or more transaction models are selected. The transaction models that are each trained to apply a particular set of evidence factors corresponding to the set of attributes associated with the multiple entities, and identify transaction patterns representing potentially anomalous activity. The transaction data is processed using the one or more transaction models to identify potentially anomalous activity within the transaction data for the multiple entities. A prioritization indicator is computed for each entity included in the multiple entities.

Abstract: According to various embodiments, a machine learning based method, system, and non-transitory computer-readable medium for identifying content on social media related to one or more coordinated influence efforts are disclosed. The method includes generating one or more datasets of post-uniform resource locator (URL) pairs produced from one or more known coordinated influence efforts on one or more social media platforms. The method further includes generating one or more datasets of post-URL pairs produced from one or more random users on one or more social media platforms. The method additionally includes extracting a plurality of content-based features from the post-URL pairs from known coordinated influence efforts and random users.

Abstract: The invention includes delivering and monitoring digital content distributed to correctional facility inmates, giving supervisory authorities the ability to screen the incoming digital content. Digital content can include email, and stored and steamed video content, and can be scanned for keywords by supervisory authorities before delivery to an inmate. A computer kiosk can be used by inmates to view and record digital video content. A portable player is provided to inmates which can be used to play, and in some embodiments record, digital content. The player is issued to a particular inmate, and can only be used with respect to that particular inmate's digital content. The kiosk, and in some embodiments, the player, can be used to shop for items available at a store, for example a commissary.

Abstract: Systems and techniques are described for orchestrating iterative updates to machine learning models (e.g., transaction models) deployed to multiple end-user devices. In some implementations, output data generated by a first transaction model deployed at a first end-user device is obtained. The first transaction model is trained to apply a set of evidence factors to identify potentially anomalous activity associated with a first target entity. An adjustment for a second transaction model deployed at a second end-user device is determined. The second transaction model is trained to apply the set of evidence factors to identify potentially anomalous activity associated with a second target entity determined to be similar to the first target entity. A model update for the second transaction model is generated. The model update specifies a change to the second transaction model. The model update is provided for output to the second end-user device.

Abstract: Systems and techniques are described for applying machine learning techniques to dynamically identify potentially anomalous activity of entities. In some implementations, peer group data is obtained. The peer group data indicates multiple entities classified as belonging to a particular peer group, and a set of attributes associated with the multiple entities. Transaction data for the multiple entities is obtained from one or more data sources. One or more transaction models are selected. The transaction models that are each trained to apply a particular set of evidence factors corresponding to the set of attributes associated with the multiple entities, and identify transaction patterns representing potentially anomalous activity. The transaction data is processed using the one or more transaction models to identify potentially anomalous activity within the transaction data for the multiple entities. A prioritization indicator is computed for each entity included in the multiple entities.

Abstract: The invention includes delivering and monitoring electronic letters to correction facility inmates while giving supervisory authorities the ability to screen the incoming mail. This may be achieved by providing a database having an entry for each inmate and having a plurality of fields, and by scanning an original letter as an electronic letter and storing each electronic letter sent to a specific inmate in a relational database management system (RDMS) table. Another aspect of the invention involves providing a computer-operated kiosk that may be used by individuals (e.g., inmates) in a restrained environment/restricted-access location (e.g., a prison) to browse through a catalog of available digital media or content, such as music, that may be purchase with credits earned based on work performed by the inmate or bought through some other means, for example by family members of the inmate.

Abstract: The search engine provides a method and apparatus for receiving long queries, assigning a weight to each relevant word of the query, allowing a user to reformulate the query before and/or after search on the basis of the weight of each word computed by the algorithm. The search engine further provides methods for decomposing a long query into several short queries based on the importance of terms computed by the algorithm. These generated queries are submitted to existing search engine(s) producing several ranked outputs, and the obtained ranked outputs are merged into one final ranked output.