面向互联网的信息处理  

Internet information processing

作　　者：王海峰 吴华 刘占一 

机构地区：[1]北京百度网讯科技有限公司,北京100085

出　　处：《中国科学：信息科学》2013年第12期1624-1640,共17页Scientia Sinica(Informationis)

基　　金：国家高技术研究发展计划(批准号:2011AA01A207)资助项目

摘　　要：随着互联网海量信息的快速增长,由原来单一的以文本信息为主的信息处理发展成文本、语音、图像等多模态的信息处理.同时,用户的需求也从关键词搜索为主的信息获取向着基于语义理解的自动问答、辅助决策等智能交互的方向发展.本文从互联网服务信息处理特点以及用户需求变化出发,阐述了互联网信息处理面临的挑战和发展趋势.首先介绍了基于互联网的海量信息处理特点以及基本方法,然后分别阐述了互联网文本信息处理、语音信息处理、图像信息处理、位置信息处理的挑战以及发展趋势.最后介绍了如何对互联网的用户从属性、状态、兴趣3个维度进行建模,以满足用户个性化服务和商业分析的需要.With the fast growth of internet massive information, the simple text-oriented information processing has gradually transformed into a multi-modal technology capable of processing information of text, voice and image, etc. Meanwhile, the internet users＇ requirements have changed from information acquisition via keyword retrieval into semantic-based intelligent interaction, such as question-answering and assistant decision-making. This paper elaborates the challenges and trends of internet information processing based on the characteristics of information and the users＇ requirements transformation. First, the basic ideas and methods of internet massive information processing are introduced. Second, the challenges and trends of internet information processing on text, voice, image and location are addressed. Finally, we present how to model internet users from the perspectives of user property, status and interest, so as to satisfy the users＇ personalized requirements and business analysis needs.

关 键 词：互联网信息处理 语义理解 智能交互 用户建模 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]