Proposing a new model to aggregate ratings in multi-source feedback approach based on the evidence theory

Researchers and practitioners in multi-source feedback (MSF) context generally use the average-based methods to aggregate ratings. Because of the uncertainties in the raters’ opinions, it is believed that the use of conventional averaging methods is not appropriate for aggregating MSF data. So, in M...

Full description

Saved in:
Bibliographic Details
Published in:Soft computing (Berlin, Germany) Germany), 2020-07, Vol.24 (13), p.9479-9494
Main Authors: Nahid Titkanloo, Hossein, Keramati, Abbas, Fekri, Roxana
Format: Article
Language:English
Subjects:
Artificial Intelligence
Basic converters
Computational Intelligence
Configurations
Control
Data collection
Decision making
Engineering
Feedback
Fuzzy sets
Human resources
Mathematical Logic and Foundations
Mechatronics
Methodologies and Application
Methods
Performance appraisal
Performance management
Ratings
Ratings & rankings
Robotics
Uncertainty
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Researchers and practitioners in multi-source feedback (MSF) context generally use the average-based methods to aggregate ratings. Because of the uncertainties in the raters’ opinions, it is believed that the use of conventional averaging methods is not appropriate for aggregating MSF data. So, in MSF approach, there is a need to design a proper aggregation method that is capable to cope with the uncertainty in ratings. In this regard, in this paper, each rating group has been considered as a source of evidence, and a new aggregation model based on evidence theory has been proposed. In the proposed model, the collected data from each rating group by designing three different methods have been converted to the basic belief assignments and then aggregated using the Dempster rule of combination. In order to resolve the conflict between evidences, the discounting and compromise methods were used, and the output of the combination process was extracted using three different methods including the pignistic probability criterion, the plausibility transformation method and the expected value method. Finally, through a simulation study, the performance of the proposed model under various configurations was investigated. The results of the simulation study show that the proposed model, in almost all configurations, provides more accurate results than traditional aggregation method in MSF approach.
ISSN:1432-7643
1433-7479
DOI:10.1007/s00500-019-04458-6