TENCENT(HK:00700) Sou Hu Cai Jing·2025-12-11 07:12Core Viewpoint - The article discusses the challenges and advancements in advertising recommendation algorithms, particularly focusing on the "multi-modal generative recommendation" technology that aims to improve the accuracy and relevance of product recommendations on e-commerce platforms [5][11]. Group 1: Advertising Recommendation Algorithms - The current recommendation algorithms often misinterpret user interests, leading to irrelevant product suggestions, such as recommending the same type of product repeatedly [3][11]. - The advent of 5G and the explosion of video content have increased the complexity of user data, which traditional algorithms struggle to utilize effectively [8][9]. - Multi-modal generative recommendation technology integrates text, images, and videos to create personalized recommendations, enhancing user engagement and reducing ad fatigue [11][13]. Group 2: Technical Challenges - There are three significant technical challenges in implementing multi-modal generative recommendation systems: 1. Multi-modal noise and missing data, which affect the accuracy of generated product information [15]. 2. The need for a large-scale sparse ID system to handle the vast number of products and users [17]. 3. The cold start problem, where traditional algorithms rely on past user behavior, making it difficult to recommend products to new users [18]. Group 3: Tencent's Algorithm Competition - Tencent hosted the 2025 Advertising Algorithm Competition to attract talent and address the technical challenges in the advertising industry [18][26]. - The competition encouraged innovative solutions, with teams proposing methods to organize user behavior data and improve recommendation accuracy [19][21]. - Participants in the competition had opportunities for job offers, highlighting Tencent's commitment to nurturing talent in the advertising sector [25][26]. Group 4: Industry Impact and Future Directions - Tencent's initiative reflects a broader commitment to advancing the advertising industry through collaboration between academia and industry [26][30]. - The competition's outcomes are expected to lead to practical applications of the proposed technologies, enhancing the effectiveness of advertising and changing public perceptions [30].