The algorithmic recommendations of social media platforms may push divisive content, leading to fierce confrontation between social groups, mainly in the following ways: 1. **Personalized recommendations**: Social media platforms use users' historical behavior data (such as likes, shares, and comments) to personalize recommended content, which may expose users to similar views or positions, leading to the "echo chamber" effect and further exacerbating confrontational emotions. 2. **Emotional resonance**: Algorithms tend to promote content that can trigger strong emotions, such as anger or fear. Such content is often related to divisive topics and is likely to trigger strong reactions from users. 3. **Exposure of extreme views**: Platforms may prioritize displaying extreme views or content that are highly interactive. These contents often arouse strong debates, thereby deepening confrontations between different groups. 4. **Group belonging**: Social media often strengthens users' sense of group belonging through functions such as groups and tags. When users are immersed in a small circle, they may be hostile to the outside world, leading to estrangement between groups. 5. **Algorithmic bias**: The design of the algorithm and the training data may be biased, causing some views to be over-promoted while other views are ignored, resulting in unbalanced information dissemination. 6. **Information cocoon**: Users on social media may tend to only follow accounts and content that agree with their own views, eventually forming information cocoons, making it difficult for opposing parties to communicate and understand each other effectively. 7. **Viral spread**: Divisive content is often more likely to be shared and spread than neutral or harmonious content, and algorithms will further promote it to a wider audience based on these interactive data. By understanding these mechanisms, users and social media platforms can more consciously manage the consumption and dissemination of content, thereby reducing the risk of social confrontation.