SIGDIAL/SEMDIAL 2017 Joint Session on Negotiation Dialog
Most of the work to date on spoken dialog systems has focused on task-oriented dialog, where the goal of the system is to help the user complete a task, such as finding and booking a flight, a restaurant or a hotel. This type of dialog is typically modelled with an underlying frame with slots that need to be filled, in order for the system to have enough information to provide the user with the information that they seek, or to complete a transaction on behalf of the user, such as booking a flight.
However there has been less effort on dialog interactions where the dialog partners have individual, perhaps private, and potentially conflicting, task and social goals. We call dialogs with these properties negotiation dialogs. In these types of dialogs, the conversation typically revolves around abstract objects representing the conversants’ opinions, thoughts, needs or desires, rather than concrete entities that can be easily represented as frames and slots.
Negotiation dialogs show one or more of the following characteristics that makes them distinct from traditional dialogs:
- Exchange of ideas/opinions in response to a topic(s);
- Present or withhold information that may potentially lead to speaker’s task success;
- Convince the interlocutor to achieve common ground; alternatively perpetual disagreement;
- Show positive or negative attitudes towards audience during the exchange e.g., flamewars, snarky comments, insults, witty remarks at the other’s expense;
- Deliberate digression from the topic or obfuscation of one’s own or other’s proposals, claims or arguments. Recent work in this area includes models of several distinct types of dialog, from experimentally collected task-oriented dialogs where the participants must come to an agreement about how to distribute resources, or what is an appropriate plan, to work on social media dialogs identifying various aspects of the argument structure, work on understanding the main points of the discussion, and argument mining. Some of the available related corpora and resources are listed below.
Objectives of the Special Session
Our objectives in this special session are two-fold: first, to encourage discussion among researchers working on or interested in working negotiation dialog; and second, to set the stage and establish a planning committee for a 2018 shared task on negotiation dialog.
Related Corpora and Resources
This list is provided to assist potential participants in the special session. Submissions to the special session are not required to use these corpora or resources.
Corpora (we welcome your suggestions and contributions!):
- The Teams Corpus consists of audio and associated transcriptions for 62 teams of 3 and 4 speakers playing a simplified version of the cooperative board game Forbidden Island (TM). The game gives each player a different role; players need to communicate (e.g., discuss cards and strategies) to achieve team goals.
- Diane Litman, Susannah Paletz, Zahra Rahimi, Stefani Allegretti and Caitlin Rice (2016). “The Teams corpus and entrainment in multi-party spoken dialogues”. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing.
- The Internet Argument Corpus (IAC 2.0) is a database of social and political dialogs collected from social media sites such as CreateDebate and 4Forums, which has annotations for various aspects of negotiation dialog, such as agreement and disagreement, sarcasm, nastiness, factuality of arguments vs. emotionality.
- Abbott, Robert, et al. (2016). “Internet argument corpus 2.0: An SQL schema for dialogic social media and the corpora to go with it”. In Proceedings of the Language Resources and Evaluation Conference.
- The Yahoo News Annotated Comments Corpus is one of the largest annotated corpora of online human dialogues, with the most detailed set of annotations. It will be valuable for identifying Engaging Respectful Informative Conversations and other aspects of argumentation, dialogue, and discourse. ERICs are characterized by argumentative, respectful exchanges containing persuasive, informative, and/or sympathetic comments.
- Napoles, Courtney et al. (2017). “Finding good conversations online: The Yahoo News Annotated Comments Corpus”. In Proceedings of the 11th Linguistics Annotation Workshop.
- The Settlers of Catan Corpus collected by Nicholas Asher and colleagues is a collection of annotated “strategic conversations”, in which participants in a game of Settlers of Catan may negotiate or mislead in order to increase their chances of winning. Like the Teams corpus, this is a multiparty dialog corpus; unlike that corpus, it is chat-based.
- Stergos Afantenos, et al. (2012). “Developing a corpus of strategic conversation in the Settlers of Catan.” In Proceedings of SeineDial 2012-The 16th Workshop On The Semantics and Pragmatics Of Dialogue.
- Nicholas Asher, et al. (2016). “Discourse structure and dialogue acts in multiparty dialogue: the STAC corpus”. In Proceedings of the Tenth International Conference on Language Resources and Evaluation.
- The Coconut Corpus is a collection of annotated dialogs in which the conversational participants collaborate to furnish rooms in a house; each participant has their own list of furniture and prices, and may have their own preferences, but the budget is shared.
- Barbara Di Eugenio, et al. (2000). “The agreement process: An empirical investigation of human–human computer-mediated collaborative dialogs”. International Journal of Human-Computer Studies, 53(6), 1017-1076.
Other relevant references (we welcome your suggestions and contributions!):
- Candice Sidner (1994). “Negotiation in collaborative activity: A discourse analysis”. Knowledge-Based Systems, 7(4), 265-267.
- Thomas Rist and Markus Schmitt. (2008). “Applying socio-psychological concepts of cognitive consistency to negotiation dialog scenarios with embodied conversational characters”. In Animating Expressive Characters for Social Interaction, Cañamero L, Aylett R (Eds). John Benjamins Publishing Co: 213-234.
- Avi Rosenfeld, et al. (2014). “NegoChat: a chat-based negotiation agent”. In Proceedings of the International Conference on Autonomous Agents and Multi-Agent Systems.
- Nicholas Asher and Alex Lascarides (2013). “Strategic conversation”. Semantics and Pragmatics, 6(2), 1–62.
- David DeVault, Jonathan Mell, and Jonathan Gratch (2015). “Toward natural turn-taking in a virtual human negotiation agent”. In Proceedings of the AAAI Spring Symposium on Turn-taking and Coordination in Human-Machine Interaction.
Call for Papers
We welcome submissions on any aspect of negotiation dialog, including:
- The effect of negotiation on the dialog structure and pragmatic aspects e.g. entrainment;
- Systems that participate in negotiation dialog;
- Proposals for new corpora or shared tasks for negotiation dialog;
- Corpus and tool descriptions, annotation schemes;
- Statistical models of the task and the conversants in negotiation dialogs
Researchers may choose to submit:
- Regular papers – will present original research and go through the regular SIGDIAL or SEMDIAL review process. These papers will appear in the SIGDIAL or SEMDIAL proceedings.
- Position papers – need not present original research and/or could possibly propose a shared task for negotiation dialog. These will be reviewed by the special session organizers and posted on the special session website. These papers will be presented as “lightning talks” during the session.
Regular Paper Submissions
To submit a regular paper, please go to:
- SIGDIAL 2017 for SIGDIAL submissions (deadline April 18)
- SEMDIAL 2017 for SEMDIAL submissions (deadline May 8)
Position Paper Submissions
To submit a position paper, please email a 2-4 page PDF (including references), formatted using the SIGDIAL 2017 format guidelines, to firstname.lastname@example.org, by May 8 (the SEMDIAL submission deadline). You will receive an email confirmation of your submission and notification by the SEMDIAL notification date.
- Amanda Stent, Bloomberg LP
- Aasish Pappu, Yahoo Inc
- Diane Litman, University of Pittsburgh
- Marilyn Walker, University of California Santa Cruz