250 likes | 402 Views
Building Mashups by Demonstration. ACM Transactions on the Web, Vol 5. No.3, Article 16, July 2011. Research By Rattan Tuchinda,Craig A. Knoblock, Pedro Szekely Presenter : Preeti Loomba CSU Id : 2438925. Mashup Overview.
E N D
Building Mashups by Demonstration ACM Transactions on the Web, Vol 5. No.3, Article 16, July 2011 Research By Rattan Tuchinda,Craig A. Knoblock, Pedro Szekely Presenter : Preeti Loomba CSU Id : 2438925
Mashup Overview Mashup describes a web application that integrates data from multiple web sources to provide a unique service.
Current Solutions • Widget Paradigm • Current Solutions involve selecting, customizing, and connecting widgets together • Disadvantages • As number of widgets gets large, locating the right widget becomes confusing and time consuming • Connecting widgets required understanding of programming concepts
Goal Create a mashup building framework where an average Internet user with no programming experience can build Mashups easily.
Problems • Data Retrieval • Source Modeling • Data Cleaning • Data Integration • Data Display
Mashup Categorization • One Simple Source • Combining data points from two or more separate sources • One source with a form • Combining two or more sources with a database join
Key Ideas • Focus on data, not the operation • Leverage existing databases • Consolidate rather than divide and conquer
Source Modeling • Karma compares extracted data with existing data in its repository • Automatically populates some attributes • User specifying the correct attribute • Users search existing attributes in data repository
Data Cleaning • User selects clean data tab • User specifies what data needs to be cleaned • User specifies cleaned result • Karma will try to induce the cleaning transformation
Data Integration • Karma analyzes attributes and data • Karma determines joins between data in table and data in the repository • Karma suggests existing data sources in repository that can be linked to the new data in the table
Data Retrieval • DOM tree as basic structure for extraction • Organization of HTML tags in the web page • Positioning of nodes used for data extraction
Related Works- Widget Approach • Yahoo Pipes • Microsoft’s Popfly • Marmite • IBM’s QED wiki • Bungee Labs • Proto Software
Related Works – Other approaches • Simile • Potluck • Intel’s MashMaker • Mario • Cards • Google MyMaps
Karma Advantages • An end-to-end approach • A consistent paradigm • Wide coverage
Future Works • Customizing display by Examples • Recovering from Errors • Source Quality • Support for Advanced Users • Data Cleaning Transformations