Data Aggregation Scheme for a Mashup Based Internet Application

This paper discusses a data aggregation scheme based on the query results by using a mashup technique. We propose system architecture for seamlessly integrating a portal for end-user and a data analysis application. The system architecture is applied to a mashup application for analyzing stock portfolio to integrate an Internet-based application and data analysis application. In the application, we use Protocol Buffer to define the data structure representing a message signature, Me cached to maintain the result of end-user query and NaradaBroker, pub/sub based middleware, to connect loosely with sub systems. To evaluate performance of system, we compare the size of messages written in different data formats. In the evaluation, we perform a TPC-based benchmark to evaluate caching effect.