Yoga Food & Travel Rotating Header Image

Write 2 Short Dialogue About Expression Agreement

Kakavá, C. (1993). Negotiation of disagreements by Greeks in conversations and class speeches. (Dissertation), Washington, DC: Georgetown University. The corpus is the result of joint efforts of researchers in computational linguistics, pragmatics, engineering, information science and psychology. The project, which was launched in 2009, aimed at a detailed study of human-human communication in order to make an important contribution to the development of various human-computer interaction systems. From the beginning, it became clear that such a system should be multimodal, that is, it should go beyond verbal communication and also include gestures that would increase the ease of use of these systems. It was also clear that the system should be able to model two-way communication. Namely, it should be able to participate in a recursive sequence of interaction events by going beyond simply responding to a request or executing a request – it should “listen” to other reactions from the human user, evaluate them and act accordingly.

Such a system requires two simultaneously active communication channels, that of analysis and synthesis, through which actors can continuously change their roles as speakers and listeners. The model we have proposed as the basis for our corpus construction would follow exactly this requirement (cf. Hunyadi, 2011). The naturalness of two-way communication necessarily presupposes that the actors are freely involved in the given subject and that the flow of interactions allows an unrestricted expression of gestures and emotions. As a result, we designed two types of dialogues: a mainly formal dialogue – in the form of an interview with a series of predefined sentences and a second, which is an informal interview (for better data management, the latter also followed a certain orientation, but allowed individual variations). In order to better understand the possible structure of the sequences of interactions and to offer useful generalizations, the experimental scheme focused on the following elements: the management of the shooting, the variation of intentions and the generation of emotions. The video recordings were made with 111 participants as speakers (60 men and 51 women, aged 21-30) and two participants as agents (one man and one woman, aged 25 and 28 respectively), with the average duration of formal interviews being about 10 minutes and informal conversations about 20 minutes. The resulting corpus has a total duration of approximately 50 hours (for more descriptive details of the corpus, see Pápay et al., 2011; Hunyadi et al., 2012).

We can see that the frequency of elements in annotations does not necessarily correspond to their frequency in real models: Be careful, the most common Annotation Element is only classified as the 3rd most common element in a model, it is preceded in frequency by Standard Disagree and Agree. This frequency data gives us insight into the nature of recorded conversations: there were many moments of attention, as a natural component of a dialogue, but the fact that there were more moments of (standard) disagreement than agreement also suggests that the interaction was quite free. The small number of objects that are denied, dubious, and surprised suggests that the interaction doesn`t involve much direct confrontation. A series of additional notes to determine the degree of agreement revealed that there were 2331 cases of the element that were not interested – another reflection of the speaker at some point in the interaction. .

Comments are closed.