A5029024536- Recommender Systems and Techniques
- Image and Video Quality Assessment
- Advanced Bandit Algorithms Research
- Consumer Market Behavior and Pricing
- Caching and Content Delivery
- Digital Marketing and Social Media
- Advanced Graph Neural Networks
- Sentiment Analysis and Opinion Mining
- Web Data Mining and Analysis
- Advanced Computing and Algorithms
Alibaba Group (China)
2019-2023
Alibaba Group (United States)
2019
Click-through rate (CTR) prediction is a critical task in online advertising systems. Existing works mainly address the single-domain CTR problem and model aspects such as feature interaction, user behavior history contextual information. Nevertheless, ads are usually displayed with natural content, which offers an opportunity for cross-domain prediction. In this paper, we leverage auxiliary data from source domain to improve performance of target domain. Our study based on UC Toutiao (a...
Click-through rate (CTR) prediction is a critical task in online advertising systems. A large body of research considers each ad independently, but ignores its relationship to other ads that may impact the CTR. In this paper, we investigate various types auxiliary for improving CTR target ad. particular, explore from two viewpoints: one spatial domain, where consider contextual shown above on same page; temporal historically clicked and unclicked user. The intuitions are together influence...
Click-through rate (CTR) prediction is one of the most central tasks in online advertising systems. Recent deep learning-based models that exploit feature embedding and high-order data nonlinearity have shown dramatic successes CTR prediction. However, these work poorly on cold-start ads with new IDs, whose embeddings are not well learned yet. In this paper, we propose Graph Meta Embedding (GME) can rapidly learn how to generate desirable initial for ad IDs based graph neural networks meta...
Click-through rate (CTR) prediction is a critical task in online advertising systems. Most existing methods mainly model the feature-CTR relationship and suffer from data sparsity issue. In this paper, we propose DeepMCP, which models other types of relationships order to learn more informative statistically reliable feature representations, consequence improve performance CTR prediction. particular, DeepMCP contains three parts: matching subnet, correlation subnet subnet. These subnets...
Click-through rate (CTR) prediction is a critical task in online advertising systems. Models like Deep Neural Networks (DNNs) are simple but stateless. They consider each target ad independently and cannot directly extract useful information contained users' historical impressions clicks. In contrast, models Recurrent (RNNs) stateful complex. model temporal dependency between sequential behaviors can achieve improved performance than DNNs. However, both the offline training process of RNNs...
Click-through rate (CTR) prediction is a critical task in online advertising systems. Existing works mainly address the single-domain CTR problem and model aspects such as feature interaction, user behavior history contextual information. Nevertheless, ads are usually displayed with natural content, which offers an opportunity for cross-domain prediction. In this paper, we leverage auxiliary data from source domain to improve performance of target domain. Our study based on UC Toutiao (a...
In real-world advertising systems, conversions have different types in nature and ads can be shown display scenarios, both of which highly impact the actual conversion rate (CVR). This results multi-type multi-scenario CVR prediction problem. A desired model for this problem should satisfy following requirements: 1) Accuracy: achieve fine-grained accuracy with respect to any type scenario. 2) Scalability: parameter size affordable. 3) Convenience: not require a large amount effort data...
Click-through rate (CTR) prediction is a critical task in online advertising systems. Most existing methods mainly model the feature-CTR relationship and suffer from data sparsity issue. In this paper, we propose DeepMCP, which models other types of relationships order to learn more informative statistically reliable feature representations, consequence improve performance CTR prediction. particular, DeepMCP contains three parts: matching subnet, correlation subnet subnet. These subnets...
Conversion rate (CVR) prediction plays an important role in advertising systems. Recently, supervised deep neural network-based models have shown promising performance CVR prediction. However, they are data hungry and require enormous amount of training data. In online systems, although there millions to billions ads, users tend click only a small set them convert on even smaller set. This sparsity issue restricts the power these models. this paper, we propose Contrastive Learning for...
Click-through rate (CTR) prediction is a critical task in online advertising systems. Models like Deep Neural Networks (DNNs) are simple but stateless. They consider each target ad independently and cannot directly extract useful information contained users' historical impressions clicks. In contrast, models Recurrent (RNNs) stateful complex. model temporal dependency between sequential behaviors can achieve improved performance than DNNs. However, both the offline training process of RNNs...