Watch time is widely used as a proxy for user satisfaction in video recommendation platforms. However, raw watch times are influenced by confounding factors such as video duration, popularity, and individual user behaviors, potentially distorting preference signals and resulting in biased recommendation models. We propose a novel relative advantage debiasing framework that corrects watch time by comparing it to empirically derived reference distributions conditioned on user and item groups. This approach yields a quantile-based preference signal and introduces a two-stage architecture that explicitly separates distribution estimation from preference learning. Additionally, we present distributional embeddings to efficiently parameterize watch-time quantiles without requiring online sampling or storage of historical data. Both offline and online experiments demonstrate significant improvements in recommendation accuracy and robustness compared to existing baseline methods.

Short video streaming has become a dominant paradigm in digital media, characterized by rapid swiping interactions and diverse media content. A key technical challenge is designing an effective preloading strategy that dynamically selects and prioritizes download tasks from an evolving playlist, balancing Quality of Experience (QoE) and bandwidth efficiency under practical commercial constraints. However, real world analysis reveals critical limitations of existing approaches: (1) insufficient adaptation of download task sizes to dynamic conditions, and (2) watch time prediction models that are difficult to deploy reliably at scale. In this paper, we propose DeLoad, a novel preloading framework that addresses these issues by introducing dynamic task sizing and a practical, multi dimensional watch time estimation method. Additionally, a Deep Reinforcement Learning (DRL) enhanced agent is trained to optimize the download range decisions adaptively. Extensive evaluations conducted on an offline testing platform, leveraging massive real world network data, demonstrate that DeLoad achieves significant improvements in QoE metrics (34.4% to 87.4% gain). Furthermore, after deployment on a large scale commercial short video platform, DeLoad has increased overall user watch time by 0.09% while simultaneously reducing rebuffering events and 3.76% bandwidth consumption.

Modern recommendation systems aim to increase click-through rates (CTR) for better user experience, through commonly treating ranking as a classification task focused on predicting CTR. However, there is a gap between this method and the actual objectives of businesses across different sectors. In video recommendation services, the objective of video on demand (VOD) extends beyond merely encouraging clicks, but also guiding users to discover their true interests, leading to increased watch time. And longer users watch time will leads to more revenue through increased chances of presenting online display advertisements. This research addresses the issue by redefining the problem from classification to regression, with a focus on maximizing revenue through user viewing time. Due to the lack of positive labels on recommendation, the study introduces Tweedie Loss Function, which is better suited in this scenario than the traditional mean square error loss. The paper also provides insights on how Tweedie process capture users diverse interests. Our offline simulation and online A/B test revealed that we can substantially enhance our core business objectives: user engagement in terms of viewing time and, consequently, revenue. Additionally, we provide a theoretical comparison between the Tweedie Loss and the commonly employed viewing time weighted Logloss, highlighting why Tweedie Regression stands out as an efficient solution. We further outline a framework for designing a loss function that focuses on a singular objective.

Recommenders are significantly shaping online information consumption. While effective at personalizing content, these systems increasingly face criticism for propagating irrelevant, unwanted, and even harmful recommendations. Such content degrades user satisfaction and contributes to significant societal issues, including misinformation, radicalization, and erosion of user trust. Although platforms offer mechanisms to mitigate exposure to undesired content, these mechanisms are often insufficiently effective and slow to adapt to users' feedback. This paper introduces an intuitive, model-agnostic, and distribution-free method that uses conformal risk control to provably bound unwanted content in personalized recommendations by leveraging simple binary feedback on items. We also address a limitation of traditional conformal risk control approaches, i.e., the fact that the recommender can provide a smaller set of recommended items, by leveraging implicit feedback on consumed items to expand the recommendation set while ensuring robust risk mitigation. Our experimental evaluation on data coming from a popular online video-sharing platform demonstrates that our approach ensures an effective and controllable reduction of unwanted recommendations with minimal effort. The source code is available here: https://github.com/geektoni/mitigating-harm-recsys.

Watch time prediction (WTP) has emerged as a pivotal task in short video recommendation systems, designed to encapsulate user interests. Predicting users' watch times on videos often encounters challenges, including wide value ranges and imbalanced data distributions, which can lead to significant bias when directly regressing watch time. Recent studies have tried to tackle these issues by converting the continuous watch time estimation into an ordinal classification task. While these methods are somewhat effective, they exhibit notable limitations. Inspired by language modeling, we propose a novel Generative Regression (GR) paradigm for WTP based on sequence generation. This approach employs structural discretization to enable the lossless reconstruction of original values while maintaining prediction fidelity. By formulating the prediction problem as a numerical-to-sequence mapping, and with meticulously designed vocabulary and label encodings, each watch time is transformed into a sequence of tokens. To expedite model training, we introduce the curriculum learning with an embedding mixup strategy which can mitigate training-and-inference inconsistency associated with teacher forcing. We evaluate our method against state-of-the-art approaches on four public datasets and one industrial dataset. We also perform online A/B testing on Kuaishou, a leading video app with about 400 million DAUs, to demonstrate the real-world efficacy of our method. The results conclusively show that GR outperforms existing techniques significantly. Furthermore, we successfully apply GR to another regression task in recommendation systems, i.e., Lifetime Value (LTV) prediction, which highlights its potential as a novel and effective solution to general regression challenges.

Recommendation systems (RecSys) are designed to connect users with relevant items from a vast pool of candidates while aligning with the business goals of the platform. A typical industrial RecSys is composed of two main stages, retrieval and ranking: (1) the retrieval stage aims at searching hundreds of item candidates satisfied user interests; (2) based on the retrieved items, the ranking stage aims at selecting the best dozen items by multiple targets estimation for each item candidate, including classification and regression targets. Compared with ranking model, the retrieval model absence of item candidate information during inference, therefore retrieval models are often trained by classification target only (e.g., click-through rate), but failed to incorporate regression target (e.g., the expected watch-time), which limit the effectiveness of retrieval. In this paper, we propose the Controllable Retrieval Model (CRM), which integrates regression information as conditional features into the two-tower retrieval paradigm. This modification enables the retrieval stage could fulfill the target gap with ranking model, enhancing the retrieval model ability to search item candidates satisfied the user interests and condition effectively. We validate the effectiveness of CRM through real-world A/B testing and demonstrate its successful deployment in Kuaishou short-video recommendation system, which serves over 400 million users.

Reinforcement Learning (RL) has been widely applied in recommendation systems to capture users' long-term engagement, thereby improving dwelling time and enhancing user retention. In the context of a short video & live-stream mixed recommendation scenario, the live-stream recommendation system (RS) decides whether to inject at most one live-stream into the video feed for each user request. To maximize long-term user engagement, it is crucial to determine an optimal live-stream injection policy for accurate live-stream allocation. However, traditional RL algorithms often face divergence and instability problems, and these issues are even more pronounced in our scenario. To address these challenges, we propose a novel Supervised Learning-enhanced Multi-Group Actor Critic algorithm (SL-MGAC). Specifically, we introduce a supervised learning-enhanced actor-critic framework that incorporates variance reduction techniques, where multi-task reward learning helps restrict bootstrapping error accumulation during critic learning. Additionally, we design a multi-group state decomposition module for both actor and critic networks to reduce prediction variance and improve model stability. Empirically, we evaluate the SL-MGAC algorithm using offline policy evaluation (OPE) and online A/B testing. Experimental results demonstrate that the proposed method not only outperforms baseline methods but also exhibits enhanced stability in online recommendation scenarios.

Within the domain of short video recommendation, predicting users' watch time is a critical but challenging task. Prevailing deterministic solutions obtain accurate debiased statistical models, yet they neglect the intrinsic uncertainty inherent in user environments. In our observation, we found that this uncertainty could potentially limit these methods' accuracy in watch-time prediction on our online platform, despite that we have employed numerous features and complex network architectures. Consequently, we believe that a better solution is to model the conditional distribution of this uncertain watch time. In this paper, we introduce a novel estimation technique -- Conditional Quantile Estimation (CQE), which utilizes quantile regression to capture the nuanced distribution of watch time. The learned distribution accounts for the stochastic nature of users, thereby it provides a more accurate and robust estimation. In addition, we also design several strategies to enhance the quantile prediction including conditional expectation, conservative estimation, and dynamic quantile combination. We verify the effectiveness of our method through extensive offline evaluations using public datasets as well as deployment in a real-world video application with over 300 million daily active users.

Science fiction movies like "Blade Runner" and "The Terminator" have defined the perception of artificial intelligence within popular culture. For most people, the term AI conjures up images of a dystopian future dominated by humanoid robots that have taken over the world. This common conception leads to the dismissal of the technology as impossible, or at least faroff in the future. Few people realize that we are already delving into a world dominated by AI, and it's nothing like "The Terminator." The actual risks posed by artificial intelligence have nothing to do with killer robots; they relate to the machine-learning algorithms that recommend content on the internet.

So far, most research on recommender systems focused on maintaining long-term user engagement and satisfaction, by promoting relevant and personalized content. However, it is still very challenging to evaluate the quality and the reliability of this content. In this paper, we propose FEBR (Expert-Based Recommendation Framework), an apprenticeship learning framework to assess the quality of the recommended content on online platforms. The framework exploits the demonstrated trajectories of an expert (assumed to be reliable) in a recommendation evaluation environment, to recover an unknown utility function. This function is used to learn an optimal policy describing the expert's behavior, which is then used in the framework to provide high-quality and personalized recommendations. We evaluate the performance of our solution through a user interest simulation environment (using RecSim). We simulate interactions under the aforementioned expert policy for videos recommendation, and compare its efficiency with standard recommendation methods. The results show that our approach provides a significant gain in terms of content quality, evaluated by experts and watched by users, while maintaining almost the same watch time as the baseline approaches.