A Trustworthiness-Aware Spatial Task Allocation using a Fuzzy-based Trust and Reputation System Approach
Unlike crowdsourcing, Spatial Crowdsourcing (SC) requires workers to travel to a specific physical location to accomplish a task. Due to its open concept, the platform accepts any interested individual as workers or task requesters, including those who may be unreliable and untrustworthy. Deploying...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Published: |
Elsevier
2023
|
| Subjects: | |
| Online Access: | http://eprints.um.edu.my/39464/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Unlike crowdsourcing, Spatial Crowdsourcing (SC) requires workers to travel to a specific physical location to accomplish a task. Due to its open concept, the platform accepts any interested individual as workers or task requesters, including those who may be unreliable and untrustworthy. Deploying untrustworthy workers in spatial tasks can negatively impact the quality of the completed tasks, thus threatening the sustainability of the SC platform. Recent research has been carried out to evaluate workers' trustworthiness based on the Trust and Reputation (TR) system. Current TR system approaches for evaluating workers' trustworthiness are mostly relying on a single trust or reputation factor, and the decisions are mainly binary. This binary representation of trustworthiness is considerably rigid and may cause severe repercussions like, an untrustworthy worker who could be the victim of partial ratings may end up not getting any kind of spatial tasks from the system, or a trustworthy worker who may have malicious intention may be allocated a spatial task. To address these limi-tations, we propose a novel framework that allocates every spatial task according to a workers' degree of perceived trustworthiness computed based on multi-criteria trust and reputation factors using a Mamdani fuzzy inference system. Our work considers historical ratings to calculate reputation value, applies sentiment analysis to infer trust value, implements Mamdani fuzzy inference to determine trustworthiness degree, and introduces the concept of referral to mitigate worker cold-start problems in spatial crowdsourcing. Our experimental findings on the Yelp real-world datasets demonstrate the reliability of the proposed framework to allocate every spatial task of various types to the most trustworthy workers from huge crowds of available workers. When evaluated against other baseline approaches, our approach achieves greater accuracy in allocating the right tasks to the most trustworthy workers. |
|---|