SelfGNN: Self-Supervised Graph Neural Networks for Sequential Recommendation SelfGNN is a novel framework for sequential recommendation that addresses the challenges of modeling both short-term and long-term user-item interactions. The framework encodes short-term graphs based on time intervals and uses Graph Neural Networks (GNNs) to learn short-term collaborative relationships. It captures long-term user and item representations through interval fusion and dynamic behavior modeling. A personalized self-augmented learning structure enhances model robustness by mitigating noise in short-term graphs based on long-term user interests and personal stability. Extensive experiments on four real-world datasets show that SelfGNN outperforms various state-of-the-art baselines. The model implementation codes are available at https://github.com/HKUDS/SelfGNN. The key contributions of SelfGNN include: (1) Short-term Collaborative Graphs Encoding, which captures collaborative patterns and includes short-term temporal information. (2) Multi-level Long-term Sequential Learning, which forms interest with different granularity complement each other. (3) Personalized Self-Augmented Learning for Denoising, which corrects the corresponding relationships in the short-term graphs based on the long-term user interests. SelfGNN effectively captures dynamic user interests by integrating interval-level periodical collaborative relationship learning and attentive instance-level sequential modeling. SelfGNN is built upon three key paradigms: (1) Short-term Collaborative Graphs Encoding, (2) Multi-level Long-term Sequential Learning, and (3) Personalized Self-Augmented Learning for Denoising. The framework is evaluated on four real-world datasets and shows superior performance compared to existing methods. The model's ability to handle noise and capture both short-term and long-term user preferences makes it effective for sequential recommendation tasks. The results demonstrate that SelfGNN outperforms various state-of-the-art baselines in terms of recommendation accuracy and robustness.SelfGNN: Self-Supervised Graph Neural Networks for Sequential Recommendation SelfGNN is a novel framework for sequential recommendation that addresses the challenges of modeling both short-term and long-term user-item interactions. The framework encodes short-term graphs based on time intervals and uses Graph Neural Networks (GNNs) to learn short-term collaborative relationships. It captures long-term user and item representations through interval fusion and dynamic behavior modeling. A personalized self-augmented learning structure enhances model robustness by mitigating noise in short-term graphs based on long-term user interests and personal stability. Extensive experiments on four real-world datasets show that SelfGNN outperforms various state-of-the-art baselines. The model implementation codes are available at https://github.com/HKUDS/SelfGNN. The key contributions of SelfGNN include: (1) Short-term Collaborative Graphs Encoding, which captures collaborative patterns and includes short-term temporal information. (2) Multi-level Long-term Sequential Learning, which forms interest with different granularity complement each other. (3) Personalized Self-Augmented Learning for Denoising, which corrects the corresponding relationships in the short-term graphs based on the long-term user interests. SelfGNN effectively captures dynamic user interests by integrating interval-level periodical collaborative relationship learning and attentive instance-level sequential modeling. SelfGNN is built upon three key paradigms: (1) Short-term Collaborative Graphs Encoding, (2) Multi-level Long-term Sequential Learning, and (3) Personalized Self-Augmented Learning for Denoising. The framework is evaluated on four real-world datasets and shows superior performance compared to existing methods. The model's ability to handle noise and capture both short-term and long-term user preferences makes it effective for sequential recommendation tasks. The results demonstrate that SelfGNN outperforms various state-of-the-art baselines in terms of recommendation accuracy and robustness.