Estonian dialogues capture web application
Name
Tiit Käbin
Abstract
Using the wizard-of-Oz technology, only 95 dialogues in the Estonian language have been accumulated so far; which is obviously too small a number to make any conclusions on the structure of the dialogues and the use of language. This is why research in this field has also been complicated. The application the current study represents allows gathering new dialogues with comfort and ease, which should enliven the sphere in question in Estonia.
The application is meant for two types of users who would play one of the two possible roles. As these roles are rather different and the application is to support both, two different user interfaces have been created: the user-end interface and the Wizard-of-Oz interface. The interfaces do overlap to a certain extent but offer additional functionalities depending on the role of the user. In their turn, both interfaces consist of modules, each fulfilling a separate task, and the latter vary considerably. The connection point of the two interfaces is the database, i.e., all the actions that are performed with the use of the interfaces are saved in the database, and all the information for the functioning of the interfaces is obtained from the database. To transfer and receive the data while preventing the page from reloading all the time, the AJAX technology is employed.
As the created application contains such numerous technological innovations, as, for example, complete and semi-complete answers, the Eye of Oz and its selection and administration, the system is more comfortable and user-friendly in comparison to the system used before; which, it its turn, is supposed to revive the interest in the use of the given system.
The opportunities for the further development of the system include voice recording and playback as well as recording the web-camera image of the user. Another opportunity for further development is the complete proofreading of the texts entered into the Wizard-of-Oz to check their compliance to rules in order do minimize the probability of the user recognizing the interference of other people in the conversation.
Graduation Thesis language
Estonian
Graduation Thesis type
Bachelor - Computer Science
Supervisor(s)
prof. M. Koit
Defence year
2011