Paper Info
Title
Introducing validity in fuzzy probability for judicial decision-making.
Abstract
Since the Age of Enlightenment, most philosophers have associated reasoning with the rules of probability and logic. This association has been enhanced over the years and now incorporates the theory of fuzzy logic as a complement to the probability theory, leading to the concept of fuzzy probability. Our insight, here, is integrating the concept of validity into the notion of fuzzy probability within an extended fuzzy logic (FLe) framework keeping with the notion of collective intelligence. In this regard, we propose a novel framework of possibility–probability–validity distribution (PPVD). The proposed distribution is applied to a real world setting of actual judicial cases to examine the role of validity measures in automated judicial decision-making within a fuzzy probabilistic framework. We compute valid fuzzy probability of conviction and acquittal based on different factors. This determines a possible overall hypothesis for the decision of a case, which is valid only to a degree. Validity is computed by aggregating validities of all the involved factors that are obtained from a factor vocabulary based on the empirical data. We then map the combined validity based on the Jaccard similarity measure into linguistic forms, so that a human can understand the results. Then PPVDs that are obtained based on the relevant factors in the given case yield the final valid fuzzy probabilities for conviction and acquittal. Finally, the judge has to make a decision; we therefore provide a numerical measure. Our approach supports the proposed hypothesis within the three-dimensional contexts of probability, possibility, and validity to improve the ability to solve problems with incomplete, unreliable, or ambiguous information to deliver a more reliable decision.
Year
DOI
Venue
2014
10.1016/j.ijar.2013.12.003
International Journal of Approximate Reasoning
Keywords
Field
DocType
Decision-making,f-Constraints,Judicial cases,Possibility,Probability,Validity
Defuzzification,Validity,Fuzzy measure theory,Fuzzy logic,Probability measure,Artificial intelligence,Type-2 fuzzy sets and systems,Fuzzy number,Membership function,Mathematics,Machine learning
Journal
Volume
Issue
ISSN
55
6
0888-613X
Citations 
PageRank 
References 
7
0.43
16
Authors
2
Name
Order
Citations
PageRank
Farnaz Sabahi1202.42
Mohammad R. Akbarzadeh-Totonchi212518.26