Workflow
生成式推荐系统
icon
Search documents
Scaling Law 仍然成立,企业搜广推怎么做才能少踩“坑”?
AI前线· 2025-12-09 06:26
Core Insights - The article discusses the transformation of search, advertising, and recommendation systems through the integration of large models, emphasizing the challenges and solutions for implementing generative recommendations in practical scenarios [2][4]. Group 1: Key Changes in Search and Recommendation Systems - The most significant change brought by large models is in feature engineering, where traditional methods are being enhanced by the capabilities of large language models to extract richer features from vast amounts of data [6]. - The industry is still far from achieving a fully unified end-to-end pipeline, with most efforts focused on integrating large models into specific points of the pipeline rather than complete reconstruction [12][4]. - The scaling law remains applicable in recommendation systems, indicating that the marginal benefits of model scaling have not yet reached their limits, particularly due to the vast amount of user behavior data available [13][17]. Group 2: Challenges and Solutions in Model Implementation - A major challenge in deploying large models is the need for extensive foundational work, such as data cleaning and sample construction, which can consume significant time and resources [8]. - The transition from traditional feature engineering to a more systematic approach to data and sample construction is crucial for realizing the potential of large models [8][9]. - Balancing model size, performance, and computational costs is essential, with smaller models being preferred in low-value scenarios while larger models are pursued for high-value applications [19][20]. Group 3: Future Directions and Innovations - The future of recommendation systems may see a shift from feature engineering to knowledge engineering, where models learn directly from raw user behavior data supplemented by incremental knowledge [30]. - The development of intelligent agents capable of autonomous planning and execution of complex tasks is anticipated, moving beyond predefined workflows [30]. - The industry is encouraged to focus on maximizing the utility of existing models by improving the quality of training data and optimizing the model's effective parameters [20][38].
生成式推荐:AI时代互联网技术皇冠上的明珠
2025-10-23 15:20
Summary of Conference Call on Generative Recommendation Systems Industry Overview - The discussion centers around the generative recommendation systems within the AI era, highlighting their significance in the internet technology landscape [1][2]. Key Companies Mentioned - **Meta**: Leading in end-to-end generative recommendation systems, with advertising revenue growth surpassing Google and a reported increase in ad conversion rates by 3-5 percentage points in Q2 2025 [5][6]. - **Kuaishou (快手)**: Achieved significant benefits through the WangRank algorithm, improving user engagement metrics [1][6]. - **Alibaba, ByteDance, Xiaohongshu**: Utilizing hybrid generative recommendation architectures, showing substantial commercial value improvements [1][9]. - **Tencent and Bilibili**: Also applying generative recommendation technologies, with varying degrees of revenue growth [3][12]. Core Insights and Arguments - **Generative Recommendation Technology**: Considered the crown jewel of internet technology in the AI era due to its scalability, data enhancement capabilities, unified optimization framework, and support for diverse recommendations [2]. - **Performance Metrics**: Generative recommendation systems have shown to increase user active days by 0.3% and total usage time by 1% in experiments [8]. Kuaishou's end-to-end model yielded revenue increases between 0.54% and 3% [8]. - **Comparison with Traditional Systems**: Traditional recommendation systems have limitations in creativity and generation, while generative systems leverage user behavior history to recommend the next likely interaction item [4][3]. Important but Overlooked Content - **Market Potential**: Mid-sized companies are expected to benefit more significantly from the generative recommendation wave, with Alibaba's GMV compound growth rate reaching 22% during the deep learning technology boom [12]. - **Business Scenarios**: E-commerce platforms are particularly suited for generative recommendation models due to stable user data and low real-time feedback requirements [10][11]. - **Risks in Implementation**: Potential risks include slower-than-expected industry advancement, supply chain issues related to computational power, and macroeconomic policy changes affecting profitability [14]. Future Outlook - Companies embracing generative recommendation technology, such as Alibaba, Kuaishou, Tencent, Bilibili, Baidu, iQIYI, Applovin, and HuiLiang Technology, are expected to achieve significant commercial value over the next two to three years [13].