Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form
D-Scholarship @ Pitt Banner and Links to Homepage

DATA-DRIVEN PREDICTIVE MODELING AND MULTI STAKEHOLDER RECOMMENDER SYSTEMS FOR THE PUBLIC GOOD

Arabghalizi, Tahereh (2023) DATA-DRIVEN PREDICTIVE MODELING AND MULTI STAKEHOLDER RECOMMENDER SYSTEMS FOR THE PUBLIC GOOD. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

[img] PDF
Restricted to University of Pittsburgh users only until 14 August 2024.
Download (4MB) | Request a Copy

Abstract

Public transit is a key aspect of “smart cities”. As such, many technologies, applications, and infrastructure have been deployed to improve public transportation. For example, a commuter may receive a notification while waiting at the bus stop, alerting them to the next bus being full and offering a discount (e.g., $2 off) at a nearby coffee shop if they take a later bus.
In the first part of this thesis, we address the issue of bus fullness to increase the utilization and quality of public transportation. Specifically, we propose and develop multiple predictive models and evaluate their accuracy using data from the Pittsburgh region. Our models consistently outperform the baselines.
The second part of this thesis aims to propose new approaches for recommending items (e.g., coupons) while also considering the preferences of all stakeholders. Traditional recommender
systems focus on the needs and preferences of the user, but overlook the preferences of other parties involved (e.g., “product suppliers” or “service providers”). The goal of this work is to propose solutions for “multi-stakeholder recommender systems” in an online environment where the number of stakeholders, their importance level, the number of offers,
and the availability of items can change.
In particular, we propose and develop several recommendation solutions using Multi-armed Bandits to provide a reasonable level of satisfaction for all stakeholders in a multi-stakeholder platform in the long term. Our extensive experimental results on a real-world dataset show that our proposed approaches outperform the baseline methods and provide a
good balance between the satisfaction of different stakeholders over time.

Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Unpublished
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Arabghalizi, Taherehtaa80@pitt.edutaa80
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairLabrinidis, Alexandroslabrinid@cs.pitt.edu
Thesis AdvisorLabrinidis, Alexandroslabrinid@cs.pitt.edu
Committee MemberChrysanthis, Panospanos@cs.pitt.edu
Committee MemberJia, XiaoweiXIAOWEI@pitt.edu
Committee MemberPelechrinis, Konstantinoskpele@pitt.edu
Date: 14 August 2023
Date Type: Publication
Defense Date: 3 March 2023
Approval Date: 14 August 2023
Submission Date: 11 April 2023
Access Restriction: 1 year -- Restrict access to University of Pittsburgh for a period of 1 year.
Number of Pages: 158
Institution: University of Pittsburgh
Schools and Programs: School of Computing and Information > Computer Science
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Recommender Systems, Multi-stakeholder Recommender Systems, Multi-armed Bandits, Smart City
Date Deposited: 14 Aug 2023 19:03
Last Modified: 14 Aug 2023 19:03
URI: http://d-scholarship.pitt.edu/id/eprint/44514

Metrics

Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item