INFORMATION EXTRACTION AND SYNTHESIS LAB FOR TOPIC MODELING

美国国立卫生研究院（NIH）下属的国家神经疾病和中风研究所（NINDS）最近资助了一项研究分类工具，该工具使用主题建模对NIH拨款进行基于文本的分类，并使用大规模绘图算法在地图上按空间组织成最合适的类别显示这些拨款。公众可以在http://www.nihmaps.org/上获得这个互动工具。为了加强这一工具并进一步发展主题建模方法，该研究所已获得马萨诸塞大学信息提取与合成实验室（IESL）的服务，以执行NIH拨款的主题建模。IESL作为主题建模的创新中心而闻名：研究贡献包括可扩展性方面的进步，例如第一个在超过10亿单词的语料库上运行的模型，以及建模方面的进步，例如将主题建模方法与时间序列分析工具和回归模型相结合的模型的引入。该项目的这一部分的具体目标是开发一个基于java的web应用程序，以支持专家对主题的评估。

The National Institute of Neurological Disorders and Stroke (NINDS), an Institute within the National Institutes of Health (NIH), recently funded a research categorization tool that uses Topic Modeling to perform text-based categorization of NIH grants, and a large scale graphing algorithm to display these them on a map organized spatially into their best-fit categories. This interactive tool is available to the public at http://www.nihmaps.org/. The Institute has enlisted the services of the Information Extraction and Synthesis Laboratory (IESL) at the University of Massachusetts In order to enhance this tool and further develop the topic modeling method, the to perform topic modeling of NIH grants. IESL is well-known as a center of innovation for topic modeling: research contributions include both advances in scalability, such as the first reported model to run on a corpus of more than one billion words, and modeling advances, such as the introduction of models that combine topic modeling approaches with time series analysis tools and regression models. The specific goal of this portion of the project is to develop a Java-based web application to support evaluation of topics by experts.