Contextual interestingness ranking of documents for due diligence in the banking industry with topicality grouping

A method of ranking documents according to contextual interestingness comprising:

• receiving a document to be analyzed in computer-readable form, by executing first instructions in a computer system;
• receiving a subject of interest and a category of interest which is independent of the subject of interest, by executing second instructions in the computer system;
• classifying the document to determine that it belongs to the category of interest, by executing third instructions in the computer system, wherein said classifying is carried out by a machine learning service which utilizes multiple natural language text classifiers each of which provides a decision as to a most likely category out of a plurality of available categories for a given passage, the category of interest is a mode of the decisions from all of the natural language text classifiers, and at least one of the natural language text classifiers is a naïve Bayes classifier with feature vectors generated using Word2Vec modeling;
• identifying at least one entity in the document that corresponds to the subject of interest, by executing fourth instructions in the computer system;
• grouping passages from the document that relate to the entity, by executing fifth instructions in the computer system;
• analyzing the passages for sentiment to generate a sentiment score for each passage, by executing sixth instructions in the computer system;
• and
• ranking the document based on the sentiment scores, by executing seventh instructions in the computer system.