Belief relevance in the forgetting process of goal-oriented intelligent agents

Bill F.L.; Morveli-Espinoza M.; Tacla C.A. · Acta Scientiarum - Technology (2026) · DOI: 10.4025/actascitechnol.v48i1.76371

Resumen

An intelligent agent’s belief base grows due to perceptions, communication, and deductions; however, not all of these beliefs remain relevant over time. Forgetting is an important epistemic process for removing information from the agent’s beliefs base with the aim to improve the reasoning computational efficiency. This is particularly important in goals reasoning because beliefs support the progress of goals status during the reasoning cycle. In this article, we endow intelligent agents with the capacity to forget beliefs considered irrelevant. For the forgetting process, we base on the Data-oriented Belief Revision (DBR) model and develop algorithms for the storing, retrieval, and forgetting operations that make part of this process. These operations consider the beliefs relevance; hence, for this calculation we propose to consider a factor that reflects the notion of how long it is left until a goal is achieved. We compare this form of calculation with the most used in literature. To evaluate our proposal and the impact of forgetting on goals processing, we carried out an experiment with two in beliefs-desires-intention agents (BDI) and belief-based goal processing agents (BBGP). The results of the experiment showed that our proposal leads to an improvement in processing efficiency and memory efficiency in most of the evaluated situations.

Instituciones: Universidad Nacional de San Agustín de Arequipa