Goal 1
Optimize Digital Advertising Efficiency
Analyze large volumes of user data to predict ad performance and optimize real-time bidding, so ads reach the most relevant audiences, conversion rates rise, and wasted ad spend falls.
Auburn University · Computer Science & Software Engineering
Data Mining for Real-Time Bidding
- Collect
- Mine
- Predict
- Bid
Auburn University computer scientists and Wright Media software engineers are developing data mining techniques for enhanced real-time bidding in Alabama's online display advertising industry: raising click-through rates, cutting cost per click, and matching job seekers with the right opportunities.
- 3Project GoalsEfficiency · growth · experience
- 4Design ObjectivesFramework to training
- 3Research QuestionsCTR · CPC · production
- 2Marketing Products EnhancedEmail · social media
Why it matters
Data-Driven Advertising for Alabama Businesses
The project bridges the gap between Alabama's data science workforce and the region's digital marketing needs, applying leading-edge data mining to make online advertising more efficient through an industry partnership with Wright Media LLC.
Goal 2
Help Alabama's Small and Midsize Businesses Grow
Give local businesses access to data mining techniques and tools for highly targeted, efficient campaigns, helping them compete with larger companies in the digital marketplace.
Goal 3
Enhance the Client and Customer Experience
Understand user preferences, behaviors, and trends to deliver personalized, relevant ad content that feels less intrusive and builds lasting customer relationships.
Expected benefits
- Improved targeting
- Higher return on investment
- Optimized ad spending
- Real-time insights
- Personalized advertising
- Better market understanding
- Improved analytics & reporting
- Enhanced user experience
- More accurate job matching
Design objectives
From a Data Mining Framework to a Trained Workforce
We build a data mining infrastructure, apply it to two digital marketing products developed by Wright Media, and train the engineers who will carry it forward.
-
Create a Data Mining Framework
Collect · process · analyze
A robust, scalable system that collects, processes, and analyzes large volumes of user data from web analytics, social media, and customer databases, predicting user behavior and ad performance to optimize real-time bidding strategies.
-
Integrate Data Mining into Email Marketing
Wright Media email marketing tool
Analyze customer data to find patterns and preferences, create highly targeted email content, and learn the best content and timing for each audience segment to raise open rates, click-through rates, and conversions.
-
Apply Data Mining to Social Media Management
Wright Media social media tool
Analyze user interactions, engagement, and content performance to identify trends, optimal posting times, audience preferences, and customer sentiment, strengthening the online presence of Alabama's local businesses.
-
Establish a Training Program
Wright Media engineers
Auburn researchers train Wright Media's software and data engineers in advanced data mining techniques for digital advertising, building lasting data science capacity in Alabama's workforce.
Research
Three Questions Guide the Work
Our team develops optimization algorithms that enhance click-through rate and cut cost per click in digital advertising campaigns, and tests whether they hold up in production.
- Which data mining algorithms can optimize click-through rate and reduce cost per click?
- We evaluate prediction and pattern mining algorithms on advertising data to find those that raise CTR and lower CPC.
- How should resources be balanced to achieve bidding optimization?
- We study how to balance budgets and bids across campaigns and audience segments as auctions happen in real time.
- Can the evaluated algorithms be readily implemented in a real-world production system?
- We integrate the most promising algorithms into Wright Media's email marketing and social media management products.
Question 1
Question 2
Question 3
CTRClick-Through RateRaise engagement with every impression
CPCCost per ClickSpend less for each result
RTBReal-Time BiddingBid smarter, auction by auction
MatchJob MatchingAlign job seekers with suitable positions
Research areas
- Recommender systems
- Recommendation algorithms
- User behavior analytics
- Machine learning for prediction
- Data mining & pattern mining
- Feature grouping & high-dimensional analytics
- Sentiment & social media analytics
- Business intelligence & user feedback analysis
Foundations
Related Publications
The project builds on our group's prior work in recommendation, pattern mining, high-dimensional analytics, and real-time behavioral analytics.
Recommender Systems & Prediction
Recommendation and association rule mining underpin ad targeting, click prediction, audience segmentation, and job recommendations.
Modeling Recommender Systems using Disease Spread Techniques
P.-X. He, L.-B. Sun, X. Gao, Y. Zhou, and X. Qin
Information, vol. 16, no. 8, pp. 687–703, 2025
GARMT: Grouping-Based Association Rule Mining to Predict Future Tables
P.-H. He, L.-B. Sun, X. Gao, Y. Zhou, and X. Qin
Computers, vol. 14, no. 6, pp. 220–235, 2025
Frequent Pattern & Feature Mining at Scale
Parallel pattern mining reveals user behavior and ad-click patterns; mutual information identifies the features that drive click-through and conversion.
MiCS-P: Parallel Mutual-Information Computation of Big Categorical Data on Spark
J.-L. Li, C.-W. Zhang, J. Zhang, X. Qin, and L.-H. Hu
Journal of Parallel and Distributed Computing, vol. 161, pp. 118–129, 2022
HBPFP-DC: A Parallel Frequent Itemset Mining Using Spark
Y.-L. Xun, J.-F. Zhang, H.-F. Yang, and X. Qin
Parallel Computing, vol. 101, 102738, 2021
FiDoop: Parallel Mining of Frequent Itemsets Using MapReduce
Y. Xun, J. Zhang, and X. Qin
IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 46, no. 3, pp. 313–325, 2016
High-Dimensional Analytics & Segmentation
Similarity, clustering, and outlier detection support customer segmentation, anomaly detection in bidding behavior, and click-quality checks.
Similarity Metrics: Chebyshev Coulomb Force and Resultant Force for High-Dimensional Data
J.-Y. Liu, C.-W. Zhang, M. Zhang, X. Qin, and J.-F. Zhang
ACM Transactions on Knowledge Discovery from Data, vol. 19, no. 3, pp. 1–25, 2025
ICE: Incremental Subspace Clustering of High-Dimensional Categorical Data
N. Pang, C.-W. Zhang, J.-F. Zhang, and X. Qin
International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, vol. 33, no. 1, pp. 87–118, 2025
A High-Dimensional Outlier Detection Approach Based on Local Coulomb Force
P.-Y. Zhu, C.-W. Zhang, X.-F. Li, J.-F. Zhang, and X. Qin
IEEE Transactions on Knowledge and Data Engineering, vol. 35, no. 6, pp. 5506–5520, 2022
Outlier Detection from Multiple Data Sources
Y. Ma, X.-J. Zhao, C.-W. Zhang, J.-F. Zhang, and X. Qin
Information Sciences, vol. 580, pp. 819–837, 2021
Real-Time Behavioral & Social Media Analytics
Real-time analytics pipelines over large-scale social and behavioral data, the same kind of data that drives ad engagement.
A Computational Approach for Real-Time Detection of Fake News
C.-W. Zhang, A. Gupta, X. Qin, and Y. Zhou
Expert Systems with Applications, vol. 221, 119656, 2023
Detecting Fake News for Reducing Misinformation Risks Using Analytics Approaches
C.-W. Zhang, A. Gupta, C. Kauten, A. V. Deokar, and X. Qin
European Journal of Operational Research, vol. 279, no. 3, pp. 1036–1052, 2019
User Feedback & Business Intelligence
User feedback and engagement analytics are closely related to measuring advertising effectiveness.
User Feedback Analysis for Business Intelligence: Semantics, Sentiment and Model Robustness
Chengfei Wang (Ph.D. dissertation; co-advised by X. Qin and A. Gupta)
Auburn University, 2022
Developing Mobile App Design from User Feedback using Deep Unsupervised Learning
C.-F. Wang, X. Qin, and A. Gupta
Proc. 2022 Pre-ICIS SIGDSA Symposium on Analytics for Digital Frontiers, Copenhagen, Denmark, 2022
Student Research on Recommender Systems
Earlier graduate and undergraduate projects in our group built parallel recommendation engines on Hadoop clusters.
Item-based Recommendation Algorithms using Hadoop
Chetan Prakash Somani (M.S. thesis)
Auburn University, 2015
Content Based Recommendation System on Hadoop Clusters
Hasitha Athota (M.S. project)
Auburn University, 2015
A User-based Collaborative Filtering Recommender Algorithm on Hadoop Clusters
Sudha Varanasi (M.S. thesis)
Auburn University, 2015
Parallel Recommendation Systems
Justin Lovell and James Sentell (REU undergraduate research)
Auburn University, 2015
See the complete publication list.
Questions? Talk to us
Project Team and Partner
Principal Investigator
Xiao Qin, Ph.D.
Professor and Director of Graduate Programs, Department of Computer Science and Software Engineering, Auburn University
xqin@auburn.eduIndustry partner
Wright Media LLC
An Alabama provider of web and mobile based recruiting solutions and a digital marketing leader for the transportation industry. Wright Media engineers work with Auburn researchers to bring data mining into production.
wrightmediacorp.comSupport
Supported by the Alabama Research and Development Enhancement Fund
This project is funded by the Alabama Research and Development Enhancement Fund (ARDEF) through the Alabama Department of Economic and Community Affairs (ADECA), with matching support from industry partner Wright Media LLC.
Funded project · 2025 – 2026
Data Mining for Enhanced Real-Time Bidding in Alabama's Online Display Advertising Industry
Alabama Research and Development Enhancement Fund, Alabama Department of Economic and Community Affairs
Project Team
- Xiao QinPrincipal Investigator
- Wright Media LLCIndustry partner