Patent Cartography: Improving the Process of Searching Through the Patent Thicket

该项目将通过结合密歇根大学信息学院开发的用于测量专利空间的方法和IBM阿尔马登研究中心开发的用于浏览和探索大型文档集合中的主题和概念的方法，改进专利检索过程。这种称为专利制图的组合方法将利用多种分类法、相关术语、网络分析、可视化和用户交互来导航、探索和绘制专利空间。随着专利率的提高，专利丛林的问题也在增加，或者是重叠知识产权的密集网络，一个组织必须通过它才能将新技术商业化。在某些以累积创新和多重封锁专利为特征的行业，这种密集集中的专利权的存在会产生抑制创新而非鼓励创新的反常效果。美国联邦贸易委员会（Federal Trade Commission）最近的一份报告指出，在某些行业，大量已发布的专利使得几乎不可能搜索所有潜在的相关专利，审查每项专利中包含的权利要求，并评估侵权风险或对许可的需求。对于许多公司来说，面对这种无意的、有时不可避免的专利侵权问题，唯一切实可行的办法就是每年申请数百项专利，以便在交叉授权谈判中有所交易。换句话说，面对某一特定领域的大量专利，唯一合理的反应可能是为之做出贡献。鉴于美国每年发布20万项专利，每周都有新的专利发布，任何分析和理解专利空间动态拓扑和相互联系的尝试几乎肯定必须基于信息技术。然而，即使出现了可检索的专利数据库，专利检索的过程也没有取得很大进展。虽然是自动化的，但专利检索过程本身并没有被重新设计，因此在功能上仍然类似于19世纪开发的专利检索过程。虽然最终用户没有必要的工具来进行详尽的专利检索，但由于对知识产权重要性的认识日益提高，专业专利检索人员的工作量不断增加，使他们不堪重负。对有效的终端用户专利检索功能的需求只会随着时间的推移而增加，然而传统的检索过程必须重新设计以满足这一需求。使用现有专业生成的专利检索集，该项目将比较使用现有工具进行的最终用户搜索与使用新方法进行的其他搜索。已经确定了几个不同类别的最终用户，他们以前对专利制度的知识和经验水平不同，本项目将把每一类用户产生的搜索结果与专业专利搜索结果进行比较。在研究的最后，我们不仅将更好地了解专利制图和可视化技术在专利检索过程中的适用性，而且还将了解对专利制度的熟悉和经验对利用专利制图潜力的影响。该项目将扩展文档检索的边界，超越简单的关键字和类别搜索，进入将混合主动数据和文本挖掘与可视化相结合的多维分析。该项目还将结合内容和网络分析，并基于接近性和依赖性数据评估可视化效果。搜索过程和搜索策略也将在大型文件集合的背景下进行检查。该项目还将开发基于任务的信息搜索模型，这方面的探索通常不足。最后，该项目将通过观察专利检索专家来发展专利检索过程的民族志。这个项目将扩展搜索、分析和可视化大型文档集合的研究范围。专利是大型数字图书馆存储库的一种示例形式，其中包含可以应用这些技术的结构化和非结构化数据。其他来源包括Medline、研究摘要、万维网存储库和其他企业存储库。这种搜索过程可以用于医疗保健、生命科学、制药、市场研究、竞争情报、知识发现以及跟踪技术成熟和创新等领域。

This project will improve the process of patent search by combining methodologies for measuring the patent space developed at the University of Michigan School of Information with methodologies developed at IBM's Almaden Research Center for browsing and exploring topics and concepts within a large document collection. This combined methodology, called Patent Cartography, will leverage multiple taxonomies, related terms, network analytics, visualization, and user interaction to navigate, explore, and map the patent space. As the rate of patenting has increased, so too has the problem of patent thickets, or dense webs of overlapping intellectual property rights that an organization must hack its way through in order to commercialize new technology. In certain industries characterized by cumulative innovations and multiple blocking patents, the existence of such densely concentrated patent rights can have the perverse effect of stifling innovation rather than encouraging it. A recent Federal Trade Commission report notes that in certain industries, the large number of issued patents makes it virtually impossible to search all the potentially relevant patents, review the claims contained in each of those patents, and evaluate the infringement risk or the need for a license. For many firms the only practical response to this problem of unintentional and sometimes unavoidable patent infringement is to file hundreds of patents each year so as to have something to trade during cross-licensing negotiations. In other words, the only rational response to the large number of patents in a given field may be to contribute to it. Given that 200,000 US patents are issued each year, with new patents issuing each week, any attempt to analyze and comprehend the dynamic topology and interconnectedness of patent space will almost certainly have to be based on information technology. The process of patent search, however, has not progressed very much even with the advent of searchable patent databases. Although automated, the patent search process itself has not been reengineered and thus remains functionally similar to the patent search process developed in the nineteenth century. While end-users do not have the necessary tools to conduct exhaustive patent searches, professional patent searchers are overwhelmed with an ever increasing workload driven by an increased awareness of the importance of intellectual property. The need for effective end-user patent search capabilities will only increase over time, yet the traditional search process must be reengineered for that need to be met. Using sets of existing professionally-generated patent searches, this project will compare end-user searches conducted with existing tools with other searches conducted with the new methodologies. Several different classes of end-users have been identified, with different levels of previous knowledge and experience with the patent system, and this project will compare the search results produced by each class of user with the professional patent search results. At the conclusion of the study, we will have not only a better understanding of the applicability of Patent Cartography and visualization techniques to the process of patent search, but also an understanding of the impact of familiarity and experience with the patent system in harnessing the potential of Patent Cartography. This project will expand the boundaries of document retrieval beyond simple keyword and category search into multidimensional analysis combining mixed initiative data and text mining with visualizations. The project will also combine content and network analysis and evaluate visualizations based on both proximity and dependency data. Search processes and search strategies will also be examined in the context of large document collections. The project will also develop models of task-based information searching, which has been generally under-explored. Finally, the project will develop an ethnographic account of the patent search process by observing patent search experts. This project will extend the scope of research on searching, analyzing, and visualizing large document collections. Patents are one exemplary form of large digital library repositories that contain both structured and unstructured data to which these techniques can be applied. Other sources include Medline, Research Abstracts, repositories of the World Wide Web, and other enterprise repositories. This search process could be leveraged in health care, life sciences, pharmaceuticals, market research, competitive intelligence, knowledge discovery, and tracking technology maturity and innovation to name just a few.