Evaluating violations of expectations to find exceptional information
Data & Knowledge Engineering
Hierarchy as a new data type for qualitative variables
Expert Systems with Applications: An International Journal
Truth Discovery with Multiple Conflicting Information Providers on the Web
IEEE Transactions on Knowledge and Data Engineering
Obtaining the consensus and inconsistency among a set of assertions on a qualitative attribute
Expert Systems with Applications: An International Journal
Semantically-grounded construction of centroids for datasets with textual attributes
Knowledge-Based Systems
Hi-index | 12.05 |
It is clear how to compute the average of a set of numeric values; thus, handling inconsistent measurements is possible. Recently, using confusion, we showed a new way to compute the consensus (a kind of average) of a set of assertions about a non-numeric fact, such as the religion of John. This paper solves the same problem for a set of objects possessing several symbolic attributes. Suppose there is a murder, and we ask several observers about the height, sex, hair color and ethnicity of the killer. They report divergent observations. What is the most likely portrayal of the assassin? Given a bag of assertions about an object described by qualitative features, this paper tells how to assess the most plausible or ''consensus'' object description. It is the most likely description to be true, given the available information. It is the ''centroid'' of the bag. We also compute the inconsistency of the bag: how far apart the testimonies in the bag are. All observers are equally credible, so differences arise from perception errors, and from the limited accuracy of the individual findings.