SBIR Phase I: Semi-Automatically Constructing Wrappers to Access Internet-Based Information Sources

这个来自Dynamic Domain的小型企业创新研究第一阶段项目将使计算机用户能够创建包装器，以便可以像查询数据库一样查询网站。该方案的关键创新是一种从示例半自动生成包装器的方法。该方法结合了监督和非监督学习方法，最小化了生成包装器所需的用户信息量。因此，没有经验的用户可以快速创建他们自己的包装器。这项研究有可能使访问和集成数据变得非常容易，从而创建大规模的虚拟数据库，因为它使信息用户可以创建虚拟数据库，而不是依赖信息提供商来执行这项任务。动态域技术通过赋予普通用户集成和查询信息源的能力，提供了改变信息集成经济性的潜力。虽然这项技术现有的商业上可行的应用包括开发购物代理和专门的搜索引擎的工具，但互联网自动化的新可能性比比皆是。例如，假设一家制造商想要在一个非常多变的市场中将其生产计划与业务伙伴整合在一起，在这个市场中，他的合作伙伴经常更换。他不需要他的合作伙伴使用与他相同的企业资源规划软件，也不需要他们开发专门的界面，而是可以轻松地导入他们通过网络以他们选择的任何格式提供的数据。

This Small Business Innovation Research Phase I project from Dynamic Domain will enable computer users to create wrappers so that web sites can be queried as if they were databases. The key innovation of this proposal is a method for semi-automatically generating wrappers from examples. The approach combines supervised and unsupervised learning methods to minimize the amount of information required from users to generate wrappers. As a result, unsophisticated users can rapidly create their own wrappers. The research has the potential to make it vastly easier to access and integrate data to create large scale, virtual databases because it makes it possible for information users to create virtual databases, rather than relying on information providers to carry out this task.Dynamic Domain proffers technology with the potential to transform the economics of information integration by giving ordinary users the power to integrate and query information sources. While existing, commercially viable applications of this technology include tools for developing shopping agents and specialized search engines, new possibilities abound for Internet automations. For example, consider a manufacturer who wants to integrate his production schedules with his business partners in a very fluid market where his partners frequently change. Rather than requiring his partners to employ the same enterprise resource planning software that he uses or asking them to develop specialized interfaces, he can easily import data they make available via the web in whatever format they choose.