生成式搜索的配图方法、设备、存储介质及程序产品
By evaluating the value and quality of candidate images using a multimodal model, high-quality images that can effectively answer user questions are selected, solving the problem that existing technologies do not provide sufficient help to user questions and improving the user experience.
CN120892589BActive Publication Date: 2026-07-17UC MOBILE CHINA CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- UC MOBILE CHINA CO LTD
- Filing Date
- 2025-06-10
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In existing multimodal AI search image matching technologies, the output images provide limited help to users' questions and may even be misleading, resulting in a poor user experience.
Method used
Multiple candidate images are identified based on the original questions searched by users. A multimodal model is used to evaluate the value score, quality score, and image time of the candidate images, and high-quality images that can effectively answer the questions are selected.
Benefits of technology
The images accompanying search results were made more helpful to users' questions, thus improving the user experience.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN120892589B_ABST
Abstract
本申请实施例提供一种生成式搜索的配图方法、设备、存储介质及程序产品。该方法包括：基于用户搜索的原始问题确定多个候选图片；将所述原始问题和所述多个候选图片输入多模态模型，得到各候选图片的价值评分、质量评分和图片时间，其中，所述价值评分表征所述候选图片对回答所述原始问题的贡献度，所述质量评分表征所述候选图片在用于回答所述原始问题的前提下的图片质量；根据各候选图片的价值评分、质量评分和图片时间确定原始问题的搜索回答中的待展示图片，其中，原始问题的搜索回答包括待展示图片和原始问题的文本回答。该方法解决了搜索回答中的配图对回答用户问题没有帮助甚至误导用户的问题，提升了用户的搜索体验。
Need to check novelty before this filing date? Find Prior Art