Abstract: Systems and methods are described for adjusting decision reliability scores (e.g., weights) in weighted consensus algorithms for blockchains include accessing broadcast events in a blockchain, where the broadcast events correspond to a historic period of time. A broadcast decision provided by a participant in the blockchain for each of the broadcast events may be identified. A consensus decision for each of the broadcast events may be identified where the consensus decision indicates whether there is a consensus among participants in the blockchain that a corresponding broadcast event should be added as a new node to the blockchain. A success rate may be calculated for the participant based the identified broadcast decision provided by the participant for each of the broadcast events and the identified consensus decision for each of the broadcast events. Based on the success rate, the decision reliability score for the participant may be increased, decreased, or maintained.

Abstract: Systems and methods are described for adjusting decision reliability scores (e.g., weights) in weighted consensus algorithms for blockchains include accessing broadcast events in a blockchain, where the broadcast events correspond to a historic period of time. A broadcast decision provided by a participant in the blockchain for each of the broadcast events may be identified. A consensus decision for each of the broadcast events may be identified where the consensus decision indicates whether there is a consensus among participants in the blockchain that a corresponding broadcast event should be added as a new node to the blockchain. A success rate may be calculated for the participant based the identified broadcast decision provided by the participant for each of the broadcast events and the identified consensus decision for each of the broadcast events. Based on the success rate, the decision reliability score for the participant may be increased, decreased, or maintained.

Abstract: Machine learning based techniques are described for identifying sanctionable persons via monitoring a plurality of electronic content sources. This may allow for more rapid identification of prohibited or restricted transactions. A trained sentiment analysis classifier may classify a particular electronic content item as containing sanctionable conduct. An electronic textual analysis of the electronic content item may be performed to identify one or more individual names within the particular electronic content item. An indication as to whether the one or more individual names have been identified as individuals who may be subject to one or more sanction requirements that prohibit one or more online actions may be electronically stored in a data table. Various operations may be performed to block or otherwise restrict online accounts associated with the individual from performing online activities.

Abstract: Systems and methods are described for adjusting decision reliability scores (e.g., weights) in weighted consensus algorithms for blockchains include accessing broadcast events in a blockchain, where the broadcast events correspond to a historic period of time. A broadcast decision provided by a participant in the blockchain for each of the broadcast events may be identified. A consensus decision for each of the broadcast events may be identified where the consensus decision indicates whether there is a consensus among participants in the blockchain that a corresponding broadcast event should be added as a new node to the blockchain. A success rate may be calculated for the participant based the identified broadcast decision provided by the participant for each of the broadcast events and the identified consensus decision for each of the broadcast events. Based on the success rate, the decision reliability score for the participant may be increased, decreased, or maintained.

Abstract: Machine learning based techniques are described for identifying sanctionable persons via monitoring a plurality of electronic content sources. This may allow for more rapid identification of prohibited or restricted transactions. A trained sentiment analysis classifier may classify a particular electronic content item as containing sanctionable conduct. An electronic textual analysis of the electronic content item may be performed to identify one or more individual names within the particular electronic content item. An indication as to whether the one or more individual names have been identified as individuals who may be subject to one or more sanction requirements that prohibit one or more online actions may be electronically stored in a data table. Various operations may be performed to block or otherwise restrict online accounts associated with the individual from performing online activities.