Method and system for accelerating judgments of documents by clustering arguments and supporting evidence based on credibility distribution

摘要

As a technical solution to respond to social problems such as fake news, research and development have been conducted on methods and systems for automatic collection of related arguments and related supporting evidence for given information. The automatic collection and provision method and system that have been proposed so far provides (1) simply arranging related thesis and supporting evidence for the given information, or (2) providing the information by dividing it into information corresponding to affirmative/opposite opinions. Because of this, there is a problem that (1) it takes too long to review related thesis and supporting evidence, or (2) there is a problem of disabling relevant information that is important in determining trustworthiness, although there is neither an affirmative opinion nor a dissenting opinion. come. In order to solve the above-described problem, the present invention provides a method and system for providing output to a user by clustering and summarizing related thesis and supporting evidence more effectively as a judgment aid for accelerating the determination of reliability of specific information. It is about. The present invention utilizes the concept of reliability distribution to predict the reliability of an automated distribution of information for a given piece of information, and to link automated arguments and supporting evidence, and reliability values of multiple data used to justify the uncertain group of readers exhibiting certain reliability. According to the clustering, the predicted reliability distribution and the plurality of data clustered to correspond to the output are provided to the user, so that the process of reviewing the plurality of data necessary for determining the reliability of the given information can be accelerated.