Mobile Intelligence

20.4 ARCHITECTURE OF THE VOICE SYSTEM

The VOICE system has a multitiered architecture that is typically distributed across multiple servers corresponding to different tiers of the system's architecture, as shown in Figure 20.2. The VoiceXML service provider is the first tier of the system, which contains a VoiceXML processor and any applicable audio files (in our example, we used the service provided by BeVocal [2]). The VOICE System tier and the VOICE Application tier contain the bulk of the VOICE system and application logic, including the VoiceXML Converter, Flow Control Module, User Session State/Context Manager, XML Application Schema, Query Manager(s), Query Designer(s), and Query Parser(s). The VOICE System tier also hosts the Context Information database. The Data Provider tier contains one or more servers providing the Data Source(s) that the Data Source Adapter(s) will query. The Restaurant Search Guide described in the previous section is an application developed within the VOICE framework and will be used throughout this section to illustrate many of the concepts presented herein.

20.4.1 User Session State/Context Manager

The User Session State/Context Manager (USSCM) controls the states of the current session given its XML Application Schema, determining what states are available to transition to next, as well as maintaining the values of each variable in the current state. Given the current values for each of the variables, the USSCM determines which states are available ...