What's in a message? Exploring dimensions of trust in reported information

Text-based information from human sources contains not just content, but also clues to where the information originated from as well as how strongly the writer believes in that information. That is, the actual content of HUMINT may be from direct observation by the writer or derived through inference by the writer, or the writer may be passing on information which has come from another source. Along with hints as to the derivation of the content, writers often embed clues as to how strongly they support the content being conveyed, i.e., a writer may indicate confidence, doubt, skepticism, conviction, etc., about the information in the communication. All of these factors contribute to the reliability and credibility of the content. The reliability of the source relies on how trustworthy the analyst believes someone is, while “self-confidence” is how strongly the human believes what he is reporting. Beyond sincerity, undisclosed intentions are also a feature of self-confidence. We propose a model that assigns trust to reported information by combining source reliability and self-confidence and describes the effects of those individual criteria on the quality of information, as perceived by the analyst.