SBIR-STTR Award

This Small Business Innovation Research Phase I project aims to demonstrate the feasibility of an intelligent recommendation system based on users? personal online research annotations. A combination of three key characteristics makes this system novel. First is the use personal Web page annotations (highlights, comments in notes, etc.) to understand users? information needs. Second is social recommendation based on the personal Web page annotations of related users with shared research interests. Third is the integration of recommended content directly into users? normal reading and information gathering behavior. Taken together, these qualities of the proposed system represent a significant advance in knowledge discovery. The broader impact of this project consists of a contribution to research and education by saving time, cost and frustration for institutions and individuals seeking information online for any project. While researching online, users wish to take notes to enforce their understanding of what they read. In the absence of adequate online annotation tools, they print important Web pages to annotate them by pen. Switching from the interactive, networked Web environment to the static, disconnected medium of paper presents two limitations. First is the inability to leverage this research to identify and pull related content. Second is the inaccessibility of the researched content to others with shared research interests such as professional colleagues. The proposed research leverages a powerful web annotation system under development that allows users to directly annotate Web pages and thereby eliminates the need to print. The proposed recommendation system will analyze users annotations and suggest related Web pages, thereby saving the time, cost and frustration experienced by students, scientists, business analysts and others who aggregate online information

This Small Business Innovation Research (SBIR) Phase II project aims to develop a Recommendation System that offers users links to relevant pages as they browse the Web. As users interact with a Web page annotation platform, they use tools, such as highlighters and sticky notes, to annotate pages. The System is able to leverage these annotations to accurately model the user's information need, and to deliver high-quality recommendations. This Phase II project builds upon a prototype developed in Phase I, applying techniques from the information retrieval and natural language processing research communities to improve recommendation quality. This project encompasses primary research in document modeling, index representations and retrieval models. Further, the project proposes interesting synergies by drawing in methods from the text categorization, topic detection and tracking and collaborative filtering communities. The Broader Impact of this work lies in its potential to positively impact the task of doing research on the web. The company's nascent Web annotation Platform promises to save users time, reducing cost and frustration by providing content management and organizational structures that allow them to preserve state between web research sessions. The next step is to deploy the Recommendation System to bring users the next page they need before they even realize they need it. Individual users and businesses alike will derive value from the time savings provided by the company's Platform and its Recommendation System