大模型驱动的多模态信息生成与信息推荐

浏览次数:10
  • 分享到:

摘要:

随着人工智能技术的快速发展,大语言模型在多模态信息生成和推荐系统中扮演核心角色。介绍了大模型如何通过跨模态学习,实现文本、图像、音频和视频数据的融合,推动信息生成的自动化和多样化,显著提升内容生成质量。在推荐系统中,大模型通过嵌入匹配、token 表示和直接作为推荐引擎,提升了个性化推荐的精准度和多样性。未来的研究可以聚焦于提升多模态模型的推理能力和生成质量,同时加强数据安全和透明性,进一步拓展大语言模型在信息生成与推荐中的应用潜力。

The rapid development of artificial intelligence technology has enabled large language models(LLMs) to play a significant role in multimodal information generation and recommendation systems. This paper introduces how LLMs achieve cross-modal learning, integrating text, image, audio, and video data to drive automation and diversiication in information generation, greatly enhancing content quality, In recommendation systems, LLMs improve the accuracy and diversity of personalized recommendations through embedding matching, token representation, and functioning directly as recommendation engines.Future research should foeus on enhaneing the reasoning ability and generating quality of multimodal models, strengthening data security and transparency, and expanding the application potential of LLMs in information generation and recommendation.

作者:

吴晔,陆俊霖

WuYe,Lu Junlin

机构地区:

北京师范大学新闻传播学院;北京师范大学计算传播学研究中心

引用本文:

吴哗,陆俊霖。大模型驱动的多模态信息生成与信息推荐[J].河南师范大学学报(自然科学版),2025,53(5):145-151.( Wu Ye, Lu Junlin.Multimodal information generation and recommendation system driven by large language models[J].Journal of Henan Normal University(Natural Seience Edition),2025,53(5);145-151.DO1:10.16366/j.cnki.1000-2367.2024.12.11.0002.)

基金:

国家自然科学基金;北京社会科学基金重点项目

关键词:

大语言模型;多模态信息;个性化推荐;智能传播

large language models; multimodal information; personalized recommendation; intelligent communication

分类号:

G202


大模型驱动的多模态信息生成与信息推荐.pdf