大数据环境下基于多源数据协同的个性化服务关键技术研究

项目名称： 大数据环境下基于多源数据协同的个性化服务关键技术研究

项目编号： No.61462022

项目类型： 地区科学基金项目

立项/批准年度： 2015

项目学科： 其他

项目作者： 黄梦醒

作者单位： 海南大学

项目金额： 45万元

中文摘要： 个性化精确服务技术正成为解决大数据环境下日益严重的信息过载和资源迷向问题的重要途径。目前，基于多数据源的个性化服务技术研究还很少。本项目拟在我们前期对跨域协同机制、个性化服务技术等研究的基础上，应用供应链协同理论，提出多源数据有效协同的机制；开发根据各数据源领域特征及项目特征自主构建数据抽取方法的自适应ETL工具，并应用Scribe和多分类器融合算法实现用户信息实时处理与有效融合，建立一种基于动态偏好逻辑的动态更新用户模型；提出一种多源数据协同环境下的协作过滤算法，有效解决当前协作过滤算法数据稀疏性和冷启动问题；将各数据源领域情景信息融入多维语义模型中，提出基于多维语义向量空间模型的个性化推荐算法；提出一种具有语义智能，能够实现自动融合的服务融合技术，为用户提供一站式综合服务。最后，设计并实现一个基于多源数据协同的个性化智能推荐原型系统。该项目研究是对个性化精确服务理论与应用的突破。

中文关键词： 大数据；个性化服务；多源数据协同；协作过滤；服务协同

英文摘要： It is the important method of personalized precise-service to solve the problems of information overload and resources lost in big data environment. Until now, the research for personalized service technologies based on multi data sources is also rarely. On the basis of our previous researches about cross-domain coordination mechanisms and personalized service technologies, the project proposes effective coordination mechanism of multi-source data based on supply chain coordination theory. An adaptive ETL tool is developed which can self-build the data extraction method according to the domain characteristics and item characteristics of each data source, and a dynamically updated user model based on dynamic preference logic is created by using of Scribe and multi-classifier fusion algorithms to realize the user information real-time processing and effective fusion. A novel collaborative filtering recommendation algorithm under multi-sources data coordination environment is presented which can effectively solve the cold start problem and data sparse problem of the current collaborative filtering algorithms. A personalized recommendation algorithm of multi-dimensional semantic vector space model is proposed by integrating the field scenario of each data source to multi-dimensional semantic model. A semantic intelligent and automatic integrated service merging technologies is proposed, which can provide one-stop personalized service for users. Finally, a personalized intelligent recommendation prototype system is design and implement based on multi-source data coordination. The research will be the new development of personalized precise-service theories and applications.

英文关键词： big data;personalized service;multi-source data coordination;collaborative filtering;service merging