Building Mashups by Example

The latest generation of WWW tools and services enables Web users to generate applications that combine content from multiple sources. This type of web application is referred to as a mashup. Many of the tools for constructing mashups reply on a widget paradigm, where users must select, customize, and connect widgets to build the desired application. While this approach does not require programming, the users must still understand programming concepts to successfully create a mashup. In this talk I describe our programming-by-demonstration approach to building mashup by example. Instead of requiring a user to select and customize a set of widgets, the user simply demonstrates the integration task by example. I will describe how this approach addresses the problems of extracting data from web sources, cleaning and modeling the extracted data, and integrating the data across sources. We implemented these ideas in a system called Karma and evaluated Karma on a set of 20 users and showed that compared to other mashup construction tools, Karma allowed more of the users to successfully build mashups and made it possible to build these mashups significantly faster compared to using a widget-based approach.

This research is joint work with Rattapoom Tuchinda and Pedro Szekely.

Speaker Details

Dr. Craig Knoblock is a Senior Project Leader at the Information Sciences Institute and a Research Professor in Computer Science at the University of Southern California (USC). He is also the Chief Scientist for both Fetch Technologies and Geosemble Technologies, which are spinoff companies from USC. He received his Ph.D. in Computer Science from Carnegie Mellon in 1991. His current research interests include information integration, automated planning, machine learning, and constraint reasoning and the application of these techniques to geospatial data integration.He has published one book and over 150 articles, book chapters, and conference papers in information integration, planning, and machine learning. He served on the Senior Program Committees of the 1997, 1998, 2000, 2004, 2006, and 2007 National Artificial Intelligence Conferences, the Senior Program Committee of the 2007 International Joint Conference on AI, and the Senior Program Committee of the 2004 International Semantic Web Conference. He is program co-chair for the 2008 AAAI track on AI and the Web and he is Conference Chair for the 2011 International Join Conference on AI (IJCAI). He is currently the President of the International Conference on Automated Planning and Scheduling, a Trustee of the International Joint Conference on Artificial Intelligence, and a Fellow of the Association for the Advancement of Artificial Intelligence (AAAI).

Date:
Speakers:
Craig Knoblock
Affiliation:
University of Southern California