基于多模态增量学习的广告投流意图优化方法及系统
By constructing a multimodal signal time-series causal chain for the advertising delivery system, identifying and utilizing precursor signals for early warning, and generating a forward-looking delivery optimization scheme, the problem of undiscovered causal relationships in existing technologies is solved, and the forward-looking optimization and incremental learning capabilities of the advertising delivery system are realized.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING LINGMANG TECH CULTURE CO LTD
- Filing Date
- 2026-05-18
- Publication Date
- 2026-07-17
AI Technical Summary
Existing advertising delivery systems fail to effectively uncover the temporal causal relationships between various performance metrics in the advertising environment. This results in an inability to intervene in advance before deterioration signals are transmitted along the causal chain to core metrics, and a lack of incremental learning capabilities, leading to a gradual decline in optimization effectiveness.
By collecting multimodal feedback signals, constructing a temporal causal chain of signals, identifying precursor signals and calculating the advance warning amount, forming a precursor feature description, detecting the triggering of precursor signals in real time, generating a forward-looking traffic optimization plan, and performing incremental knowledge updates after each round of advertising.
It enables early warning through precursor signals before downstream signals deteriorate, allowing for targeted adjustments to the ad delivery strategy and enhancing the forward-looking optimization capabilities and long-term optimization effects of the ad delivery system.
Smart Images
Figure CN122199065B_ABST