Runze Wu
Research Subjects
Runze Wu has a total of 211 co-authors. Here are the top co-authors by number of shared publications:
- Changjie Fan
NetEase Fuxi AI Lab, Hangzhou, China
24 shared publications - Tangjie Lv
NetEase Fuxi AI Lab, Hangzhou, China
21 shared publications - Shiwei Zhao
Fuxi AI Lab, NetEase Inc., Hangzhou, China
17 shared publications - Zhipeng Hu
Fuxi AI Lab, NetEase Inc., Hangzhou, China
10 shared publications - Haoyu Liu
NetEase Fuxi AI Lab, Hangzhou, China
9 shared publications - Xudong Shen
NetEase Fuxi AI Lab, Hangzhou, China
9 shared publications
Latest Publications
- Open Access
- 6 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 8515-8522
LASER: Label-Anchored Semantic Encoding with Multimodal-Collaborative Signals Harmonization for Generative Recommendation
- 6 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 8515-8522
Open AccessLASER: Label-Anchored Semantic Encoding with Multimodal-Collaborative Signals Harmonization for Generative Recommendation
- 15 October 2026
- Published by Springer Nature
- p. 121-136
NashLens: Game-Native Interpretable RL via CFR-Distilled Nash Anchoring
- 15 October 2026
- Published by Springer Nature
- p. 121-136
NashLens: Game-Native Interpretable RL via CFR-Distilled Nash Anchoring
- Open Access
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 51-62
The Power of Penalties: Negativity-Aware Incentives for High-Quality Crowdsourced Data Labeling
High-quality data labeling is essential for training robust machine learning models; however, existing methods often ignore fraud or assume non-negative worker utility, failing to penalize harmful contributions without discouraging participation. To address this, we propose the Negativity-Aware Incentive (NAI) mechanism which introduces two novel components. First, the Ability-Result Characteristic Function (AR-CF) adapts and extends Shapley value theory through signed valuation to explicitly capture both positive and negative contributions, by combining workers' abilities with real-time task results to define contribution values. Second, a dynamic stake pool mechanism employs pre-commitment economics with adaptive dual-control parameters to balance fairness and operational efficiency. Through extensive experiments on multimodal datasets (images, text, audio, video), NAI outperforms state-of-the-art baselines: it improves video labeling accuracy by 16.6%, and reduces fraudulent behaviors by 33.9%. Furthermore, our deployment on the NetEase Youling crowdsourcing platform, serving 430,000 registered workers with 80,000 monthly active workers, validates NAI's real-world viability. Real-time A/B testing shows a 59.6% improvement in labeling quality for beginner tasks and a consistent reduction in fraud rates (14.8%-33.9%) across difficulty levels. This work establishes a paradigm shift in crowdsourcing system design, demonstrating that explicit negative modeling can enhance data quality, optimize costs, and foster participation at scale.
- 12 April 2026
- Published by Association for Computing Machinery (ACM)
- p. 51-62
Open AccessThe Power of Penalties: Negativity-Aware Incentives for High-Quality Crowdsourced Data Labeling
High-quality data labeling is essential for training robust machine learning models; however, existing methods often ignore fraud or assume non-negative worker utility, failing to penalize harmful contributions without discouraging participation. To address this, we propose the Negativity-Aware Incentive (NAI) mechanism which introduces two novel components. First, the Ability-Result Characteristic Function (AR-CF) adapts and extends Shapley value theory through signed valuation to explicitly capture both positive and negative contributions, by combining workers' abilities with real-time task results to define contribution values. Second, a dynamic stake pool mechanism employs pre-commitment economics with adaptive dual-control parameters to balance fairness and operational efficiency. Through extensive experiments on multimodal datasets (images, text, audio, video), NAI outperforms state-of-the-art baselines: it improves video labeling accuracy by 16.6%, and reduces fraudulent behaviors by 33.9%. Furthermore, our deployment on the NetEase Youling crowdsourcing platform, serving 430,000 registered workers with 80,000 monthly active workers, validates NAI's real-world viability. Real-time A/B testing shows a 59.6% improvement in labeling quality for beginner tasks and a consistent reduction in fraud rates (14.8%-33.9%) across difficulty levels. This work establishes a paradigm shift in crowdsourcing system design, demonstrating that explicit negative modeling can enhance data quality, optimize costs, and foster participation at scale.
- 1 April 2026
- Published by Springer Nature
- p. 173-182
AI as Game Designer: Achieving Optimal Level Balance Through LLM-Based System
- 1 April 2026
- Published by Springer Nature
- p. 173-182
AI as Game Designer: Achieving Optimal Level Balance Through LLM-Based System
- Open Access
- 21 February 2026
- Published by Association for Computing Machinery (ACM)
- p. 607-617
Fast-DataShapley: Neural Modeling for Training Data Valuation
- 21 February 2026
- Published by Association for Computing Machinery (ACM)
- p. 607-617
Open AccessFast-DataShapley: Neural Modeling for Training Data Valuation
- 17 February 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in INFORMS Journal on Computing
Rational or Emotional? Next-Item Recommendations in Virtual Games via Disentangling Players’ Needs
Virtual games have become a prominent cultural, social, and economic phenomenon, captivating millions of players globally, which stimulates intelligent services, like recommendations, with strong real evidence of enabling platforms to optimize operational management, enhance user satisfaction, and drive revenue growth. Even though traditional recommender systems that rely on modeling user preferences have achieved great success in commerce, they commonly model users’ purchase behavior with a unified preference representation, which has limitations for capturing users’ diverse intrinsic purchase motivations. In this paper, we therefore conduct a study that focuses on designing a recommender system for virtual games. We argue that purchases in virtual games are mainly motivated by both rational needs and emotional needs, with the former reflecting a player’s practical need to improve their ability for a specific game campaign (e.g., attack, defense) and the latter being more of a psychological preference (e.g., color, style). We design a framework called RERec that discriminately learns representations of the two types of motivations via distinct architectures and features a better-supervised optimization that is oriented toward the entire process rather than specific items. For its rational needs module, RERec employs a unique time-gating mechanism to perceive the temporal impacts of campaigns on players’ volatile rational needs. A heterogeneous item-taxonomy graph is also used as prior knowledge of similarity among items to accelerate convergence. On the other hand, because a player’s emotional preferences are relatively stable over a short period, RERec employs a hierarchical attention mechanism to capture the comprehensiveness and focus of the player’s emotional needs. RERec also incorporates player attributes to enhance the personalized emotional need representations. Extensive experiments and analyses on large-scale real-world data sets from both a well-known game company and the public benchmarks fully demonstrate RERec’s superiority in specific respects as well as its effectiveness. History: Accepted by Ram Ramesh, Area Editor for Data Science & Machine Learning. Funding: This study was partially funded by the National Natural Science Foundation of China [Grants 72101176 and 72471165] and Emerging Frontiers Cultivation Program of Tianjin University Interdisciplinary Center. Supplemental Material: The online appendix is available at https://doi.org/10.1287/ijoc.2025.1179 . The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2025.1179 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2025.1179 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .
- 17 February 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in INFORMS Journal on Computing
Rational or Emotional? Next-Item Recommendations in Virtual Games via Disentangling Players’ Needs
Virtual games have become a prominent cultural, social, and economic phenomenon, captivating millions of players globally, which stimulates intelligent services, like recommendations, with strong real evidence of enabling platforms to optimize operational management, enhance user satisfaction, and drive revenue growth. Even though traditional recommender systems that rely on modeling user preferences have achieved great success in commerce, they commonly model users’ purchase behavior with a unified preference representation, which has limitations for capturing users’ diverse intrinsic purchase motivations. In this paper, we therefore conduct a study that focuses on designing a recommender system for virtual games. We argue that purchases in virtual games are mainly motivated by both rational needs and emotional needs, with the former reflecting a player’s practical need to improve their ability for a specific game campaign (e.g., attack, defense) and the latter being more of a psychological preference (e.g., color, style). We design a framework called RERec that discriminately learns representations of the two types of motivations via distinct architectures and features a better-supervised optimization that is oriented toward the entire process rather than specific items. For its rational needs module, RERec employs a unique time-gating mechanism to perceive the temporal impacts of campaigns on players’ volatile rational needs. A heterogeneous item-taxonomy graph is also used as prior knowledge of similarity among items to accelerate convergence. On the other hand, because a player’s emotional preferences are relatively stable over a short period, RERec employs a hierarchical attention mechanism to capture the comprehensiveness and focus of the player’s emotional needs. RERec also incorporates player attributes to enhance the personalized emotional need representations. Extensive experiments and analyses on large-scale real-world data sets from both a well-known game company and the public benchmarks fully demonstrate RERec’s superiority in specific respects as well as its effectiveness. History: Accepted by Ram Ramesh, Area Editor for Data Science & Machine Learning. Funding: This study was partially funded by the National Natural Science Foundation of China [Grants 72101176 and 72471165] and Emerging Frontiers Cultivation Program of Tianjin University Interdisciplinary Center. Supplemental Material: The online appendix is available at https://doi.org/10.1287/ijoc.2025.1179 . The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2025.1179 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2025.1179 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .
- 27 January 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in INFORMS Journal on Computing
Code and Data Repository for Rational or Emotional? Next-item Recommendations in Virtual Games via Disentangling Players' Needs
We make a contribution to the game recommendation literature by decoupling purchase sequences into two heterogeneous motivations to be modeled, i.e., rational needs and emotional needs, based on data observations and consumption motivation theory.
- 27 January 2026
- Published by Institute for Operations Research and the Management Sciences (INFORMS) in INFORMS Journal on Computing
Code and Data Repository for Rational or Emotional? Next-item Recommendations in Virtual Games via Disentangling Players' Needs
We make a contribution to the game recommendation literature by decoupling purchase sequences into two heterogeneous motivations to be modeled, i.e., rational needs and emotional needs, based on data observations and consumption motivation theory.
- Open Access
- 20 January 2026
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 38 (2)
Online label aggregation with incomplete crowd responses
Crowdsourcing delivers responses that are asynchronous and incomplete, making offline aggregators that assume complete response sets impractical. Prior online methods often either require per-step completeness or repeatedly reload historical responses, which is storage- and privacy-unfriendly and susceptible to forgetting. We present OLA-Incomplete, an online label-aggregation framework designed for incomplete response streams. It integrates a variational-inference aggregator with a generative replay module that preserves historical information without reloading prior responses and explicitly models unknown worker reliability. At each update step, the generator replays cumulative responses and side information for previously observed instances to mitigate catastrophic forgetting, while the aggregator infers current truths by maximizing the evidence lower bound over a mixture of replayed and newly received labels. Across three public datasets—Duck, RTE, and PostSent—OLA-Incomplete attains final accuracies of 90.74%, 92.50%, and 95.99%, respectively, delivering at least 7.79% relative improvement over the strongest baseline. The approach further exhibits strong instantaneous online accuracy and robustness across response-chunk sizes and arrival orders, underscoring its practical utility for real-world crowdsourcing workflows.
- 20 January 2026
- Published by Springer Nature in Journal of King Saud University Computer and Information Sciences
- Vol. 38 (2)
Open AccessOnline label aggregation with incomplete crowd responses
Crowdsourcing delivers responses that are asynchronous and incomplete, making offline aggregators that assume complete response sets impractical. Prior online methods often either require per-step completeness or repeatedly reload historical responses, which is storage- and privacy-unfriendly and susceptible to forgetting. We present OLA-Incomplete, an online label-aggregation framework designed for incomplete response streams. It integrates a variational-inference aggregator with a generative replay module that preserves historical information without reloading prior responses and explicitly models unknown worker reliability. At each update step, the generator replays cumulative responses and side information for previously observed instances to mitigate catastrophic forgetting, while the aggregator infers current truths by maximizing the evidence lower bound over a mixture of replayed and newly received labels. Across three public datasets—Duck, RTE, and PostSent—OLA-Incomplete attains final accuracies of 90.74%, 92.50%, and 95.99%, respectively, delivering at least 7.79% relative improvement over the strongest baseline. The approach further exhibits strong instantaneous online accuracy and robustness across response-chunk sizes and arrival orders, underscoring its practical utility for real-world crowdsourcing workflows.
- 21 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Games
- Vol. 18 (1) , 44-55
DPBL: Denoised Player Behavior Representation Learning
The video game industry has emerged as a significant economic force, driving extensive research on optimizing the gaming environment and improving gaming experiences. Among these endeavors, player behavior representation learning has become a critical way to model valuable player properties and is beneficial for a wide range of downstream tasks. However, some common factors, such as login rewards and daily tasks, can trigger similar behaviors among different players, which are informative and noisy for learning high-quality player behavior representations. Existing methods ignore the low signal-to-noise ratio in player behavior data and waste too much modeling capacity on less informative behaviors, resulting in their learned representations being noisy. In this paper, we propose a novel model for Denoised Player Behavior representation Learning, namely DPBL, which consists of two key modules. The first module extracts various player behavior patterns and isolates them from less informative noise. The second module utilizes the extracted patterns to refine the embedding of each behavior and eliminates noise. To optimize DPBL, two contrastive learning strategies are proposed to identify the noise that should be eliminated and to learn distinguishable representations, respectively. With the above design, DPBL is capable of mitigating the impact of noise in the data and learning high-quality representations that effectively capture player characteristics. We conducted extensive experiments on two real-world datasets, and DPBL outperforms all baselines on various downstream tasks with an improvement of 1.4% ∼ 18.1%. The results also show that DPBL achieves an improvement of 5.6% ∼ 23.0% in the denoising experiments, which proves that DPBL is more robust to noisy behaviors. Code is available at https://github.com/LwbXc/DPBL.
- 21 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Games
- Vol. 18 (1) , 44-55
DPBL: Denoised Player Behavior Representation Learning
The video game industry has emerged as a significant economic force, driving extensive research on optimizing the gaming environment and improving gaming experiences. Among these endeavors, player behavior representation learning has become a critical way to model valuable player properties and is beneficial for a wide range of downstream tasks. However, some common factors, such as login rewards and daily tasks, can trigger similar behaviors among different players, which are informative and noisy for learning high-quality player behavior representations. Existing methods ignore the low signal-to-noise ratio in player behavior data and waste too much modeling capacity on less informative behaviors, resulting in their learned representations being noisy. In this paper, we propose a novel model for Denoised Player Behavior representation Learning, namely DPBL, which consists of two key modules. The first module extracts various player behavior patterns and isolates them from less informative noise. The second module utilizes the extracted patterns to refine the embedding of each behavior and eliminates noise. To optimize DPBL, two contrastive learning strategies are proposed to identify the noise that should be eliminated and to learn distinguishable representations, respectively. With the above design, DPBL is capable of mitigating the impact of noise in the data and learning high-quality representations that effectively capture player characteristics. We conducted extensive experiments on two real-world datasets, and DPBL outperforms all baselines on various downstream tasks with an improvement of 1.4% ∼ 18.1%. The results also show that DPBL achieves an improvement of 5.6% ∼ 23.0% in the denoising experiments, which proves that DPBL is more robust to noisy behaviors. Code is available at https://github.com/LwbXc/DPBL.