Current work, presented and discussed at the symposium, included both extensions to classical speech act theory as well as attempts at standardization of speech act labels. The extensions included accounts of dialogue phenomena other than classical illocutionary acts, such as turn-taking, feedback, problem-solving, and persuasion, as well as the importance of social phenomena such as rights, roles, and obligations. Standardization groups are working toward the purposes of both inter-agent communication (using languages such as KQML) and the creation of sharable corpora of annotated dialogues (mostly between human dialogue participants), to allow language modeling at the dialogue level. An important topic is the evaluation of the reliability of such coding efforts, across different coders, dialogues and domains. The use of a proposed general level coding scheme was used by the participants in a dialogue coding exercise.

In addition, there were a number of presentations of HCI systems engaging in communicative action using a variety of media. There was some interesting discussion about how the resources and constraints of machine communicative partners differ from human-human communication. Another important issue was the role of context, which motivated a joint session with the symposium on Context in KR and NLP.

For more information (and links to some of the papers), consult the symposium web page