A method for identifying relevant information sufficient to answer situation dependent queries
Shan Lu · 2022
The progressing growth of the amount of information available to decision makers makes it more and more difficult to process all of the information in a decision making process. Thus, a method is desirable for selecting only the information that is relevant to a decision, while filtering out the information that is not relevant. A similar problem arises when one agent (human or artificial) needs to share a description of a situation the agent is in. In this dissertation, we present a conceptual framework of information relevance, in which the relevance is formalized using Situation Theory developed by Barwise and others. Based on this framework, we present an inference-based relevance reasoning process to automatically identify the information that is sufficient to characterize a given situation. This information relevance reasoning process can be applicable in many decision making scenarios in which situation-specific information needs to be exchanged. Our method is evaluated on a cyber security scenario in which analysts need to answer queries based on the information available to them. The results show that our method significantly reduces the amount of time that is needed to infer an answer to queries related to the situation. We also verify that by using only the relevant information we can get the same answers to the queries as if using the whole knowledge base. Moreover, it is possible to use the proposed method on a limited number of training queries and reuse these relevant facts to answer new queries.--Author's abstract