SBIR Phase I: Units-based numeric data extraction with knowledge of scientific context

这个小型企业创新研究第一阶段项目的重点是一种开发基于单位的数字索引和搜索工具的新方法。目标是从技术文献中提取数值，用相应的物理单位标识每个数值，并进一步将这些与设备属性等其他标识的语义实体相关联。与试图使用人工智能学习算法识别歧义结构的通用语义信息提取策略不同，纠缠技术？解决方案充分利用了单位的标准化和通用性。在知识丰富的环境中，通过科学启发式方法可以准确地识别物理量，因此数字搜索可能比关键字搜索更有效。研究目标是设计和优化这些跨越各种物理单元的科学启发式方法，以智能地提取数字数据。同时，纠缠将开发详细的科学本体论，用于识别索引的数字-单位对的上下文。该项目的成功演示为快速和准确地挖掘技术和科学规范提供了可能性。这将在工业和科学研究中有广泛的应用。Enanglement Technologies预计将从访问这种搜索技术中获得许可收入，目标客户是参与高科技投资的金融机构和提供科学搜索能力的学术图书馆。能够定义全面而客观的启发式方法来对数字进行上下文分析，这在很大程度上消除了用户识别搜索引擎--S检索--正确性的责任。这一固有特征为非专家提供了搜索、汇总和分析目前仅由专家处理的技术数据的能力。此外，基于单位的数字搜索提供了自动数字提取和聚合的潜力。Entanglement将通过将其搜索功能集成到一个前端用户友好包中来利用这一能力，使客户不仅可以从搜索中受益，还可以从简化的图形和报告生成中受益。如果成功，这种自动化的潜力将降低目前由技术咨询公司提供的此类服务的成本。该奖项由2009年美国复苏和再投资法案(公法111-5)资助。

This Small Business Innovation Research Phase I project focuses on a novel approach to develop units-based numeric indexing and search tools. The goal is to extract numeric quantities from technical literature, identify each with a corresponding physical unit, and further relate these to other identified semantic entities such as device properties. Unlike generic semantic information extraction strategies, which attempt to identify ambiguous structure with AI learning algorithms, Entanglement Technologies? solution capitalizes on the standardization and universality of units. Physical quantities are identified accurately with scientific heuristics in knowledge-rich contexts, and thus numeric search can be more efficient than keyword searches. The research objectives are to design and optimize these scientific heuristics across a wide array of physical units to intelligently extract numeric data. Concurrently, Entanglement will develop detailed scientific ontologies for identifying the context of an indexed number-unit pair.Successful demonstration of this project offers the potential for rapid and accurate data mining for technical and scientific specifications. This will have broad applications in industrial and scientific research. Entanglement Technologies anticipates generating licensing revenue from access to this search technology, targeting financial institutions involved in high-tech investments and academic libraries providing scientific search capabilities. The ability to define comprehensive, yet objective, heuristics for contextualizing a number removes much of the user's responsibility for identifying the correctness of the search engine?s retrieval. This inherent feature provides a non-expert with the capability for searching, aggregating and analyzing technical data currently only processed by experts. Furthermore, units-based numeric search offers the potential for automated number extraction and aggregation. Entanglement will utilize this capability by integrating its search functionality into a front-end user-friendly package, allowing a customer to benefit not only from the search but also from streamlined graph and report generation. If successful, this potential for automation will reduce the cost of such services currently provided by technical consulting firms.This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).