| Bor-shen Lin, Hsin-min Wang, Lin-shan Lee, "Consistent dialogue across concurrent topics based on an expert system model", European Conference on Speech Communication |
....the language models and the grammar rules are the corresponding domain dependent data. All other functional components are similarly partitioned as well. Besides, the dynamic database of the dialogue manager are also divided into the state dependent data, including the dialogue state and history [6] also enclosed by the dashed lines on the right of the figure, and the state independent data, i.e. the inference trees. The statedependent data expressed in terms of first order logic and partitioned as outside of the spoken dialogue agent (SDA) as shown in Figure 2, thus can be consulted by ....
....(P i P ) Th P j Figure 5. Facilitator (current domain i) switch to domain i result of acoustic recognition ) P max( arg i j i j # 3. 3 Concurrent Topics for Multi Agents It is important for a multi domain dialogue system to maintain consistent dialogue across concurrent topics [6]. For a distributed multi agent architecture proposed here, the control mechanism for topic switching is critical in achieving this purpose. Under such a distributed environment, the SDA which handles each topic must be stored in the dynamic database. In this way, when an active topic is closed, ....
Lin, B. S., Wang, H. M., and Lee, L. S., "Consistent Dialogue Across Concurrent Topics Based on an Expert System Model", to appear in EUROSPEECH99.
No context found.
Bor-shen Lin, Hsin-min Wang, Lin-shan Lee, "Consistent dialogue across concurrent topics based on an expert system model", European Conference on Speech Communication
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC