Algorithm and systems engineering for high-performance visual text analytics on big data

大数据高性能可视化文本分析的算法和系统工程

• 批准号: 499949-2016
• 负责人: Zeh, Norbert
• 金额: $ 9.57万
• 依托单位: Dalhousie University
• 依托单位国家: 加拿大
• 项目类别: Collaborative Research and Development Grants
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=618855
• 关键词: Algorithm; systems; engineering; performance; visual

Big Data is a popular term used to describe the exponential growth and availability of data and the technical opportunities and challenges it presents. In many diverse applications, the Big Data Challenge is how to infer important trends and causal relationships from extremely large data collections. Examples include the analysis of aviation incident and accident reports to link component failures to operational conditions that cause them to occur, analysis of textual feedback provided in product reviews with the goal to improve the products or the analysis of trending topics in Facebook and Twitter posts. The common theme of these examples is that at least a large part of the information is stored in unstructured textual form. Recent research has been successful in developing machine learning techniques that support the analysis of collections of text documents by grouping documents by topic, extracting the most important keywords from documents for easy review by the user, and many more. Most of these techniques have a high computation cost, which limits their applicability to very large text collections. Substantial performance improvements are possible through the development, implementation, performance evaluation, and tuning of improved algorithms for the basic analysis tasks underlying these machine learning techniques and through the use of parallel computing and cloud technologies. This will enable the application of these techniques to significantly larger text collections and is the focus of this research.

大数据是一个流行的术语，用来描述数据的指数增长和可用性，以及它所带来的技术机遇和挑战。在许多不同的应用中，大数据挑战是如何从极其庞大的数据集合中推断出重要的趋势和因果关系。例子包括分析航空事故和事故报告，将部件故障与导致故障发生的操作条件联系起来，分析产品评论中提供的文本反馈，目的是改进产品，或者分析Facebook和Twitter帖子中的热门话题。这些示例的共同主题是，至少有很大一部分信息以非结构化文本形式存储。最近的研究已经成功地开发了机器学习技术，通过按主题分组文档来支持文本文档集合的分析，从文档中提取最重要的关键字以方便用户查看，等等。这些技术中的大多数都有很高的计算成本，这限制了它们对非常大的文本集合的适用性。通过开发、实现、性能评估和优化这些机器学习技术基础分析任务的改进算法，以及通过使用并行计算和云技术，可以实现实质性的性能改进。这将使这些技术应用于更大的文本集合，是本研究的重点。