What should a new item token know?
Preserving Item Semantics for Free: Rethinking Token Initialization in LLM-Based Generative Recommendation
Questions, connections, and discoveries
Research.
My research centers on graph representation learning and generative recommendation systems, with an emphasis on interpretability, robustness, and efficiency in large-scale machine learning.
01 / Selected work
A few questions I work on.
How can LLMs preserve graph structure?
Formalizing and Mitigating Structural Distortion in LLM Attention for Zero-Shot Graph Reasoning
When should a graph model ask an LLM?
Glance for Context: Learning When to Leverage LLMs for Node-Aware GNN-LLM Fusion
02 / Explore the work
Paper atlas.
28 papers · 38 connections
Follow a question from one paper to another. Select a paper to explore its contribution and the ideas it shares with its neighbors.
- Recommendation
- Graph structure & prediction
- Graph editing & assignment
- Interpretability
- Scientific ML
- Early astronomy
Based on Louvain clustering of paper connections, weighted by title and summary similarity.
Read selected paperContinue exploring
This early astronomy work sits outside the connected machine-learning threads. Choose another paper above to continue exploring.
The full collection
All publications.
Open a year to browse its papers, or search by topic, title, and publication type. Conference and journal versions are listed once; preprints are labeled separately.
2026
9 papers
Training-Free LLM-Based Recommendation with Post-LLM Item Refinement Using Collaborative Signals
ACM International Conference on Information and Knowledge Management (CIKM), 2026 (Short Paper)
Citation
Kyungho Kim, Sunwoo Kim, Geon Lee, Shinhwan Kang, Sojeong Kim, Liam Collins, Bhuvesh Kumar, Donald Loveland, Kijung Shin. "Training-Free LLM-Based Recommendation with Post-LLM Item Refinement Using Collaborative Signals", ACM International Conference on Information and Knowledge Management (CIKM), 2026 (Short Paper).
Formalizing and Mitigating Structural Distortion in LLM Attention for Zero-Shot Graph Reasoning
ACM Conference on Knowledge Discovery and Data Mining (KDD), 2026
Citation
Donald Loveland, Puja Trivedi, Ari Weinstein, Edward W Huang, Danai Koutra. "Formalizing and Mitigating Structural Distortion in LLM Attention for Zero-Shot Graph Reasoning", ACM Conference on Knowledge Discovery and Data Mining (KDD), 2026.
Plain Transformers are Surprisingly Powerful Link Predictors
International Conference on Machine Learning (ICML), 2026
Citation
Quang Truong, Yu Song, Donald Loveland, Mingxuan Ju, Tong Zhao, Neil Shah, Jiliang Tang. "Plain Transformers are Surprisingly Powerful Link Predictors", International Conference on Machine Learning (ICML), 2026.
Glance for Context: Learning When to Leverage LLMs for Node-Aware GNN-LLM Fusion
International Conference on Learning Representations (ICLR), 2026
Paper : Glance for Context: Learning When to Leverage LLMs for Node-Aware GNN-LLM Fusion
Code : Glance for Context: Learning When to Leverage LLMs for Node-Aware GNN-LLM Fusion
Overview & citation
GLANCE studies when language models can help graph neural networks, using local graph structure to guide where their assistance is useful.
Donald Loveland, Yao-An Yang, Danai Koutra. "Glance for Context: Learning When to Leverage LLMs for Node-Aware GNN-LLM Fusion", International Conference on Learning Representations (ICLR), 2026
Edge-Level Automorphism in GNNs: A Quantitative Framework and Effective Designs For Link Prediction
arXiv:2609.34729, 2026
Citation
Chen Shao, Donald Loveland, Tobias Käfer, Danai Koutra. "Edge-Level Automorphism in GNNs: A Quantitative Framework and Effective Designs For Link Prediction", arXiv:2609.34729, 2026.
Preserving Item Semantics for Free: Rethinking Token Initialization in LLM-Based Generative Recommendation
arXiv:2608.07816, 2026
Citation
Donald Loveland, Liam Collins, Bhuvesh Kumar, Danai Koutra, Neil Shah. "Preserving Item Semantics for Free: Rethinking Token Initialization in LLM-Based Generative Recommendation", arXiv:2608.07816, 2026.
LLM-Based Generative Retrieval for Snapchat Content Recommendation
arXiv:2607.28895, 2026
Citation
Liam Collins, Jiwen Ren, Donald Loveland, Bhuvesh Kumar, Clark Mingxuan Ju, Xuan Guo, Mo Li, Alvin Hou, Yi Cui, Peng Yang, Jian Wang, Saud Afzal Shafi, Nga Than, Ruiming Lu, Wenfeng Zhuo, Dongheng Li, Lili Zhang, Mingtao Zhang, Jinchao Ye, Jiacong He, Vincent Xue, Chunhui Zhu, Neil Shah. "LLM-Based Generative Retrieval for Snapchat Content Recommendation", arXiv:2607.28895, 2026.
On the Memorization Behavior of LLMs in Generative Recommendation: Observations, Implications, and Training Strategies
arXiv:2606.17276, 2026
Citation
Sunwoo Kim, Sunkyung Lee, Clark Mingxuan Ju, Donald Loveland, Bhuvesh Kumar, Kijung Shin, Neil Shah, Liam Collins. "On the Memorization Behavior of LLMs in Generative Recommendation: Observations, Implications, and Training Strategies", arXiv:2606.17276, 2026.
Implicit Reasoning for Large Language Model-based Generative Recommendation
arXiv:2606.14142, 2026
Citation
Yinhan He, Liam Collins, Bhuvesh Kumar, Jundong Li, Neil Shah, Donald Loveland. "Implicit Reasoning for Large Language Model-based Generative Recommendation", arXiv:2606.14142, 2026.
2025
6 papers
On the Role of Weight Decay in Collaborative Filtering: A Popularity Perspective
ACM Conference on Knowledge Discovery and Data Mining (KDD), 2025
Overview & citation
An investigation of weight decay in collaborative filtering through the lens of item popularity.
Donald Loveland, Mingxuan Ju, Tong Zhao, Neil Shah, Danai Koutra. "On the Role of Weight Decay in Collaborative Filtering: A Popularity Perspective", ACM Conference on Knowledge Discovery and Data Mining (KDD), 2025
Understanding and Scaling Collaborative Filtering Optimization from the Perspective of Matrix Rank
ACM International World Wide Web Conference (WWW), 2025 (Oral Presentation, ~top 10%)
Citation
Donald Loveland, Xinyi Wu, Tong Zhao, Danai Koutra, Neil Shah, Mingxuan Ju. "Understanding and Scaling Collaborative Filtering Optimization from the Perspective of Matrix Rank", ACM International World Wide Web Conference (WWW), 2025 (Oral Presentation, ~top 10%)
Unveiling the Impact of Local Homophily on GNN Fairness: In-Depth Analysis and New Benchmarks
SIAM International Conference on Data Mining (SDM), 2025
Overview & citation
A study of how local graph structure affects prediction fairness, with theoretical analysis and new benchmarks for underrepresented neighborhoods.
Donald Loveland, Danai Koutra. "Unveiling the Impact of Local Homophily on GNN Fairness: In-Depth Analysis and New Benchmarks", SIAM International Conference on Data Mining (SDM), 2025
MAGNET: A Multi-Agent Graph Neural Network for Efficient Bipartite Task Assignment
International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), 2025
Citation
Donald Loveland, James Usevitch, Zachary Serlin, Danai Koutra, Rajmonda Caceres. "MAGNET: A Multi-Agent Graph Neural Network for Efficient Bipartite Task Assignment", International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), 2025
GraphHash: Graph Clustering Enables Parameter Efficiency in Recommender Systems
ACM International World Wide Web Conference (WWW), 2025 (Oral Presentation, ~top 10%)
Citation
Xinyi Wu, Donald Loveland, Runjin Chen, Yozen Liu, Xin Chen, Leonardo Neves, Ali Jadbabaie, Clark Mingxuan Ju, Neil Shah, Tong Zhao. "GraphHash: Graph Clustering Enables Parameter Efficiency in Recommender Systems", ACM International World Wide Web Conference (WWW), 2025 (Oral Presentation, ~top 10%)
Understanding ID-Text Complementarity in Sequential Recommendation
arXiv:2512.17820, 2025
Paper : Understanding ID-Text Complementarity in Sequential Recommendation
Code : Understanding ID-Text Complementarity in Sequential Recommendation
Citation
Liam Collins, Bhuvesh Kumar, Clark Mingxuan Ju, Tong Zhao, Donald Loveland, Leonardo Neves, Neil Shah. "Understanding ID-Text Complementarity in Sequential Recommendation", arXiv:2512.17820, 2025.
2023
2 papers
On Performance Discrepancies Across Local Homophily Levels in Graph Neural Networks
Learning on Graphs (LOG), 2023, (Spotlight Oral Presentation, ~top 5%)
Citation
Donald Loveland, Jiong Zhu, Mark Heimann, Ben Fish, Michael T Schaub, Danai Koutra. "On Performance Discrepancies Across Local Homophily Levels in Graph Neural Networks", Learning on Graphs (LOG), 2023, (Spotlight Oral Presentation, ~top 5%)
Network Design through Graph Neural Networks: Identifying Challenges and Improving Performance
International Conference on Complex Networks and Their Applications, 2023
Citation
Donald Loveland, Rajmonda Caceres. "Network Design through Graph Neural Networks: Identifying Challenges and Improving Performance.", International Conference on Complex Networks and Their Applications, 2023
2022
4 papers
How does Heterophily Impact the Robustness of Graph Neural Networks? Theoretical Connections and Practical Implications
ACM Conference on Knowledge Discovery and Data Mining (KDD), 2022
Citation
Jiong Zhu, Junchen Jin, Donald Loveland, Michael T Schaub, and Danai Koutra. "How does Heterophily Impact the Robustness of Graph Neural Networks? Theoretical Connections and Practical Implications.", ACM Conference on Knowledge Discovery and Data Mining (KDD), 2022
On Graph Neural Network Fairness in the Presence of Heterophilous Neighborhoods
KDD Workshop on Deep Learning on Graphs, 2022
Citation
Donald Loveland, Jiong Zhu, Mark Heimann, Ben Fish, Michael T Schaub, Danai Koutra. "On Graph Neural Network Fairness in the Presence of Heterophilous Neighborhoods", KDD Workshop on Deep Learning on Graphs, 2022
Zeroth-Order SciML: Non-intrusive Integration of Scientific Software with Deep Learning
arXiv:2206.02785, 2022
Citation
Ioannis Tsaknakis, Bhavya Kailkhura, Sijia Liu, Donald Loveland, James Diffenderfer, Anna Maria Hiszpanski, Mingyi Hong. "Zeroth-Order SciML: Non-intrusive Integration of Scientific Software with Deep Learning", arXiv:2206.02785, 2022.
FairEdit: Preserving Fairness in Graph Neural Networks through Greedy Graph Editing
arXiv:2201.03681, 2022
Citation
Donald Loveland, Jiayi Pan, Aaresh Farrokh Bhathena, Yiyang Lu. "FairEdit: Preserving Fairness in Graph Neural Networks through Greedy Graph Editing", arXiv:2201.03681, 2022.
2021
2 papers
Predicting Energetics Materials’ Crystalline Density from Chemical Structure by Machine Learning
Journal of Chemical Information and Modeling, 2021
Citation
Phan Nguyen*, Donald Loveland*, Joanne T Kim, Piyush Karande, Anna M Hiszpanski, Yong Han. "Predicting Energetics Materials’ Crystalline Density from Chemical Structure by Machine Learning", Journal of Chemical Information and Modeling, 2021
Reliable Graph Neural Network Explanations Through Adversarial Training
ICML Workshop on Theoretic Foundation, Criticism, and Application Trend of Explainable AI, 2021
Citation
Donald Loveland, Shusen Liu, Bhavya Kailkhura, Anna Hiszpanski, Yong Han. "Reliable Graph Neural Network Explanations Through Adversarial Training", ICML Workshop on Theoretic Foundation, Criticism, and Application Trend of Explainable AI, 2021
2020
3 papers
Automated Identification of Molecular Crystals’ Packing Motifs
Journal of Chemical Information and Modeling, 2020
Citation
Donald Loveland, Bhavya Kailkhura, Piyush Karande, Anna M Hiszpanski, Yong Han. "Automated Identification of Molecular Crystals’ Packing Motifs", Journal of Chemical Information and Modeling, 2020
Predicting Compressive Strength of Consolidated Molecular Solids using Computer Vision and Deep Learning
Materials & Design, 2020
Citation
Brian Gallagher, Matthew Rever, Donald Loveland, T Nathan Mundhenk, Brock Beauchamp, Emily Robertson, Golam G Jaman, Anna M. Hiszpanski, Yong Han. "Predicting Compressive Strength of Consolidated Molecular Solids using Computer Vision and Deep Learning", Materials & Design, 2020
Actionable attribution maps for scientific machine learning
ICML Workshop on ML Interpretability for Scientific Discovery, 2020
Paper : Actionable attribution maps for scientific machine learning
Related preprint : Explainable Deep Learning for Uncovering Actionable Scientific Insights for Materials Discovery and Design
Citation
Shusen Liu, Bhavya Kailkhura, Jize Zhang, Anna M Hiszpanski, Emily Robertson, Donald Loveland, Yong Han. "Actionable attribution maps for scientific machine learning", ICML Workshop on ML Interpretability for Scientific Discovery, 2020
2019
1 paper
Generative Counterfactual Introspection for Explainable Deep Learning
IEEE Global Conference on Signal and Information Processing (GlobalSIP), 2019
Citation
Shusen Liu, Bhavya Kailkhura, Donald Loveland, Yong Han. "Generative Counterfactual Introspection for Explainable Deep Learning", IEEE Global Conference on Signal and Information Processing (GlobalSIP), 2019
2018
1 paper
Studying the [OIII]λ 5007Å emission-line width in a sample of ∼ 80 local active galaxies: a surrogate for σ⋆?
Monthly Notices of the Royal Astronomical Society, 2018
Citation
Vardha N Bennert, Donald Loveland, Edward Donohue, Maren Cosens, Sean Lewis, S Komossa, Tommaso Treu, Matthew A Malkan, Nathan Milgram, Kelsi Flatland, Matthew W Auger, Daeseong Park, Mariana S Lazarova, "Studying the [OIII]λ 5007Å emission-line width in a sample of ∼ 80 local active galaxies: a surrogate for σ⋆?", Monthly Notices of the Royal Astronomical Society, 2018
No papers found.
Try a different phrase, topic, year, or publication type.
* Denotes equal contribution. Years follow the conference or journal publication; preprints use their first release year.