Automated prevention of sending objectionable content through electronic communications

A processing system, comprising:

• a central processing unit (CPU);
• a graphics processing unit (GPU);
• a bus electrically connected to and in electronic communication with the CPU and the GPU;
• and
• a neural network system including:
• a normalizer configured to normalize stored electronic communications from a communication history, pre-existing electronic communications of a message thread, and/or a newly created electronic message;
• a content extractor configured to extract content from the electronic communications and the electronic message using key term identification and/or Latent Dirichlet Allocation (LDA);
• a cluster generator configured to generate clusters from the content of the electronic communications and the electronic message;
• and
• a message filterer configured to filter the electronic communications and the electronic message with or without attachments using one or more filter matrices;
• and
• wherein the neural network system is a positively and negatively trained neural network system, including a positive outcome matrix, MP, to be applied to a newly created electronic message and a negative outcome matrix, MN, to be applied to the newly created electronic message, implemented on a communications platform, wherein the positively and negatively trained neural network system analyzes the newly created electronic message being sent using a messaging application of the communications platform to generate a confidence score that indicates whether the newly created electronic message contains objectionable subject matter for recipients identified for the newly created electronic message, and pre-emptively blocks the newly created electronic message found to contain material predicted to elicit a negative response from at least one recipient of the newly created electronic message.

A non-transitory computer readable storage medium comprising a computer readable program for pre-emptively blocking an electronic communication, wherein the computer readable program when executed on a computer causes the computer to perform the steps of:

• inputting an electronic communication history, wherein the electronic communication history includes a plurality of electronic communications and a set of corresponding recipients for each electronic communication;
• normalizing the plurality of electronic communications;
• extracting a topic from each of the plurality of electronic communications;
• clustering the plurality of electronic communications according to the extracted topics;
• digesting the plurality of electronic communications to form a positive learning data set, based on positive responses from the set of corresponding recipients, and a negative learning data set, based on negative responses from at least one recipient in the set of corresponding recipients, to train a neural network;
• training the neural network on the positive learning data set and the negative learning data set;
• preparing a positive neural network model including a positive outcome matrix, MP, to be applied to a newly created electronic message, and a negative neural network model including a negative outcome matrix, MN, to be applied to the newly created electronic message;
• analyzing a newly created electronic message being sent using a communications platform using the positively and negatively trained neural network, wherein the analysis generates a confidence score that indicates whether the newly created electronic message contains objectionable subject matter for recipients identified for the newly created electronic message;
• and
• pre-emptively blocking the newly created electronic message identified as containing material predicted to elicit a negative response from at least one recipient of the newly created electronic message.