III: Small: Collaborative Research: Building a Large Multilingual Semantic Network for Text Processing Applications

该项目致力于通过应用专门针对维基百科语料库的新的语义分析技术来建立一个大型的多语言语义网络。该项目的驱动假设是，维基百科的结构可以有效地用于创建一个高度结构化的世界知识图表，其中节点与维基百科中描述的实体和概念相对应，而边缘则捕捉本体论关系，如自上而下和近义词。特别强调利用维基百科中可用的多语言信息，以提高每个语义分析工具的性能。因此，大量的研究工作旨在开发词义消除歧义、指称消解和提取本体关系的工具，这些工具使用多语言强化以及维基百科的一致结构和重点内容来准确地解决这些任务。另一个研究挑战是有效地整合来自多篇维基百科文章的固有噪音证据，以增加全球维基百科图表中编码的整体知识的可靠性。为添加到网络中的每一条结构信息计算概率置信值是这种集成中的重要一步，它也意味着为下游应用提供更多的效用。提出的高度结构化的语义网络是对现有语义资源的补充，预计将在需要大规模世界知识的广泛的自然语言处理应用中产生积极的影响。有关更多信息，请参阅项目website:http://lit.csci.unt.edu/index.php/Mu.Se.Net

This project is devoted to building a large multilingual semantic networkthrough the application of novel techniques for semantic analysisspecifically targeted at the Wikipedia corpus. The driving hypothesis ofthe project is that the structure of Wikipedia can be effectively used tocreate a highly structured graph of world knowledge in which nodescorrespond to entities and concepts described in Wikipedia, while edgescapture ontological relations such as hypernymy and meronymy. Specialemphasis is given to exploiting the multilingual information available inWikipedia in order to improve the performance of each semantic analysistool. Significant research effort is therefore aimed at developing toolsfor word sense disambiguation, reference resolution and the extraction ofontological relations that use multilingual reinforcement and theconsistent structure and focused content of Wikipedia to solve these tasksaccurately. An additional research challenge is the effective integrationof inherently noisy evidence from multiple Wikipedia articles in order toincrease the reliability of the overall knowledge encoded in the globalWikipedia graph. Computing probabilistic confidence values for every pieceof structural information added to the network is an important step inthis integration, and it is also meant to provide increased utility fordownstream applications. The proposed highly structured semantic networkcomplements existing semantic resources and is expected to have a broadimpact on a wide range of natural language processing applications in needof large scale world knowledge.For further information, please see the project website:http://lit.csci.unt.edu/index.php/Mu.Se.Net