Controlling Bias in User Assertions in Expert Decision Support Systems for Problem Formulation

Abstract

:Information on cause-effect relationships among variables in the problem domain is one type of knowledge required in expert decision support systems for problem formulation. This knowledge must be acquired from “expert” managers and stored in the system’s knowledge base. Unfortunately even experienced managers may be biased in their beliefs about cause-effect relationships. We present a system which uses causal modeling, path analysis, and an historical database to statistically verify asserted relationships as they are entered into the system. Since it is possible that a valid assertion is not statistically supported, the user has the option to insert a relationship into the knowledge base even though the analysis may not indicate statistical validity. Information on rejected relationships is maintained in a “rejection base” which is used later to retest assertions whose validity may have changed due to updates to the database. The intent is to provide a system which helps the user learn, in an unbiased manner, about the true nature of causal relationships in the problem domain.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 35 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.