A Web Data Aggregation and Delivery System for Voice- Enabled Application Development by Nor Adnan Yahaya, Wai-Choon Chan 1 ) Abstract This paper describes a system for performing Web data aggregation for delivery through voice-enabled portals. It comprises of a web data extraction subsystem and a data format converter as the main components, which are then being complemented by other facilities for doing simple data integration as well as programmatic flow control. The system was designed such that 1) the public users, including visually impaired community, could access the aggregated Internet information using only normal telephones, and 2) additional decision support tools can be devised and integrated into it quickly and economically. 1. Introduction Web data aggregation has become an emerging phenomenon since the late 90s. It has been predominantly been applied within the financial sector and it was reported [1] that account aggregation services began to appear in web offerings by financial institutions across the United States and this trend has created global influences to Asia-Pacific countries like Australia, South Korea, and Japan. Examples of financial account aggregators are Yodlee, VerticalOne, and CashEdge. These aggregators are currently being adopted by major financial institutions such as Chase, Citibank, and Merrill Lynch as well as non-financial institutions such as CNBC and AOL [2]. Most of the existing aggregation-based applications are geared towards delivering the data to end-users through desktop computers. While delivery to mobile users through mobile devices are also being investigated, the aggregated data are still textual in nature and not suitable for the visually impaired people. To meet this equally important requirement, we have developed a system consisting of a data format converter that complements a web data extraction tool for the purpose of aggregating web data and delivery for use by the visually impaired people. The converter does conversion on the aggregated data into VoiceXML formats that are deemed to be more appropriate for voice-enabled applications. In this paper, we describe the salient features and the utility the proposed system. 2. Current Problems & Related Solution Approaches The following two scenarios serve to illustrate the current problems and limitations in term of accessing information from the Web. 1 Malaysia University of Science & Technology (MUST), Malaysia. {noradnan, wcchan}@must.edu.my