Loading...
Please wait, while we are loading the content...
Similar Documents
Adaptive Conversational Agents for Task-Oriented Dialogues
| Content Provider | Semantic Scholar |
|---|---|
| Author | Thompson, Cynthia A. |
| Copyright Year | 2007 |
| Abstract | This paper describes ongoing research on an adaptive conversational agent supporting task-oriented dialogues. The system will use spoken input, and spoken and screen-based output, with the addition of other modalities over time. The issues addressed by this research include adjusting dialogue management to suite an individual user's needs, viewing dialogue as the interactive construction of database queries, and e ectively managing dialogues for such interfaces. Two demonstration tasks are being developed, one for destination recommendation, speci cally restaurants, and the other for nding information and locations in a library setting. As growing numbers of users interact with the World Wide Web, more of its inadequacies as a method to e ciently access information and solve problems become apparent. The speci c interface improvements addressed by this paper are the addition of more natural human-computer interaction through human language, and system adaptation, over time, to the preferences of an individual user. The agent described here integrates natural language understanding technology with an adaptive ability to reason about and carry out dialogues. The goal is not just to allow users to engage in conversation, but to allow them to achieve some goal with their conversation. The adaptive component makes this process more e cient over time. For example, a user may want to retrieve a recommendation for a restaurant. Current Web interfaces force a keyword or other highly constrained search. An adaptive conversational interface, on the other hand, would allow the user to quickly narrow down his options in a more natural way, and would become more e cient at doing so as it learns a particular user's preferences. As this example illustrates, the types of tasks we are examining are those for which a large amount of information is available, but for which the user does not know the organization or extent of the information, but instead just some characteristics of the information, such as type of food, that she wants to retrieve. Figure 1 shows an example of a conversation that might take place in this domain. There has been some pioneering work on conversational interfaces for simple tasks in limited domains (Sene et al., 1996; Allen et al., 1995; Dowding et al., 1993). These systems have reached the point where they are fairly robust for conversations that t 1. U: Where should I eat today? 2. S: What type of cuisine would you like? 3. U: What are my choices? 4. S: You can pick from types of food like Chinese, Indian, or Mexican. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.cs.utah.edu/~cindi/papers/apa-agents99.pdf |
| Alternate Webpage(s) | http://www.cs.utah.edu/~cindi/papers/apa-agents99.ps.gz |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
