System and method for sentiment-based text classification and relevancy ranking

A computer implemented method of assessing human sentiment from a group of documents, each document in the group of documents having a plurality of terms and being digitally represented in a computer, the method comprising:

• receiving, by the computer, a group of documents, each document in the group of documents comprising a context of a plurality of terms and all documents in the group of documents representative of a particular topic;
• comprising:
• receiving a sentiment term meaning model for defining sentimental meanings of terms;
• and
• sentiment scoring the sentiment of each document in the group of documents based on the sentiment term meaning model and the plurality of terms in the respective document in the group of documents;
• constructing, by the computer, a document sentiment vector space from the group of documents, wherein construction of the document sentiment vector space comprises:
• assessing sentimentality of each document in the group of documents toward the topic, comprising:
• deriving a publication date for each document in the group of documents;
• electing an extrinsic metric for the particular topic for assessing the sentimentality toward the topic, the extrinsic metric being related to an affirmative and intentional human action with a value of the extrinsic metric being indicative of the human action;
• receiving extrinsic metric historical data for each document in the group of documents proximate to the respective publication date for each document;
• and
• examining the extrinsic metric historical data for each document proximate to the respective publication date for each document over a timeframe of influence for changes in the value of the extrinsic metric, wherein the timeframe of influence is a predetermined time period in which a context of a document influences humans to undertake an affirmative and intentional human action resulting in a change in the value of the extrinsic metric;
• identifying, by the computer, sentimentally significant documents in the group of documents with heightened sentimentality toward the particular topic, wherein sentimentality represents human emotion toward the topic, comprising:
• receiving an action sentiment score indicative of sentimental significance;
• comparing the sentiment score of each document in the group of documents to the action sentiment score;
• receiving a sentiment value for the change in the extrinsic metric historical data, the sentiment value being indicative of sentimental significance;
• and
• comparing the sentiment value to the changes in the value of the extrinsic metric over the timeframe of influence for each document in the group;
• labeling, by the computer, the identified sentimentally significant documents by including a unique sentiment binding term in the context of the plurality of terms;
• representing, by the computer, each document in the group of documents in the document sentiment vector space;
• defining, by the computer, a region of sentimental significance in the document sentiment vector space based on an occurrence of document representations for the identified sentimentally significant documents with the unique sentiment binding term;
• receiving, by the computer, a query string;
• and
• assessing, by the computer, the sentimentality of the query string by comparing a representation of the query string for semantic similarity to the region of sentimental significance in the document sentiment vector space.