ITR: Information Discovery in Digital Government: Self-extending Topic Maps and Ontologies (GrowOnto)

EIA-0205111 Eduard Hovy Information Sciences Institute，USCThis grant will explore automated construction of domain ontologies across governmental datasets.政府有许多“数据孤岛”，即收集的数据不能相互比较或查询。在向消费者提供政府信息时，这是一个主要问题，因为政府信息是根据消费者的兴趣而不是根据政府机构的组织结构提供的。虽然这样的自动化工具不能完全构建准确的本体，但它们可以从这样的工作中去除大量的劳动密集性，特别是在术语或变量的数量如此之大（数千）以至于阻止手动工作的情况下。与美国环境保护局合作，该机构拥有许多公众感兴趣的数据集，研究团队将开发一个半自动化的工具包，以帮助构建本体。

EIA- 0205111Eduard Hovy Information Sciences Institute, USCThis grant will explore the automated construction of domain ontologies across governmental datasets. There are many "data silos" in government; that is, collections of data which cannot be cross-compared or queried. This is a major problem in presenting government information to consumers which is tailored according to their interests, rather than according to the organizational structure of government agencies. While such automated tools cannot completely construct accurate ontologies, they can take much of the labor intensiveness out of such efforts, particularly where the number of terms or variables is so large (in the thousands) as to prevent manual efforts. Working with the US Environmental Protection Agency, which has many datasets of interest to the public, the research team will develop a semi-automated toolkit to aid in the construction of ontologies.