业内首推数据治理大模型 政企数据治理进入“3.0时代”
Zhong Guo Jing Ying Bao·2025-11-23 08:31

Core Insights - The core issue in the digital transformation of government and enterprises is data governance, with a significant amount of data becoming "sleeping assets" due to poor governance [1][2] - By 2025, it is projected that 78% of domestic enterprises will implement data governance, but less than 30% will achieve data asset operation, highlighting the challenges in the industry [1][2] - The shift from "how to manage data" to "how to utilize data" is essential in the AI era, with vertical models being key to addressing complex governance issues [1][2] Industry Evolution - Data governance has evolved through three stages: 1.0 focused on functionality, 2.0 on intelligent platforms, and the need for a 3.0 era that leverages vertical models for comprehensive intelligent empowerment [2][3] - The industry faces a "governance paradox," where high-quality data is needed for digital transformation, but obtaining it requires significant time, cost, and coordination [2] Vertical Model Advantage - The choice of vertical models over general models is due to the latter's lack of deep business understanding, which is critical for effective data governance [4][5] - The introduction of the "BS-LM" model by 百分点科技 (Percent Technology) aims to leverage accumulated project experience to create a robust data governance framework [4][5] Knowledge Management - A unique data feedback mechanism has been established to ensure high-quality training data for the models, enhancing their effectiveness [5][6] - The BS-LM model employs a "knowledge primitive" concept to break down complex governance knowledge into computable units, addressing issues like "knowledge forgetting" and "semantic drift" [6] Practical Applications - The BS-LM model has been successfully implemented in key sectors such as government and emergency management, demonstrating its practical value [7] - The focus of data governance is shifting from merely managing data to effectively utilizing it, with an emphasis on transforming industry knowledge into computable formats [7] Future Trends - The future of data governance will see the proliferation of vertical models, with competition shifting towards depth of scenarios and richness of knowledge rather than just model size [7]