Blog‎ > ‎

CONVERSATIONS 2017: International workshop on chatbot research and design

posted May 25, 2017, 12:02 AM by Symeon Papadopoulos


International workshop on chatbot research and design

November 22, 2017, Thessaloniki, Greece

Call for Papers

Researchers and practitioners working on chatbot research and design are invited to submit papers to CONVERSATIONS 2017, a one-day cross-disciplinary workshop in Thessaloniki, Greece November 22. The workshop is arranged as part of INSCI 2017.

Motivation and Key Challenges

Conversational interfaces or natural language interaction is the next frontier in the development of ubiquitous data and services. This is in particular seen in the field of chatbots, that is, machine agents serving as natural language user interfaces to data and service providers. While chatbots or conversational agents have been a topic of research for decades, recent advances in artificial intelligence and machine learning has sparked renewed interest in chatbots. Chatbots powered by artificial intelligence are by some predicted to fundamentally disrupt the way humans interact with digital services.

In this context, research on the design and application of chatbots and underlying technology is critical. Current user interactions with chatbots are still quite basic, due to several limitations. Users expect a natural conversation but these are often limited to simple task delivery and execution. Challenges pertain with regard to how artificial intelligence and machine learning is leveraged in the context of chatbots, and how to benefit from chatbots and natural language interaction in the context of Internet of Things. Furthermore, natural language interfaces imply challenges related to privacy and trust, as well as safety and security.

Chatbots and natural language user interface represents a wide range of research challenges in the fields of information systems design, software engineering, and human-computer interaction.

The analysis challenge: The Internet now offers a space of unprecedented size, where discussions among humans and agents take place and are publicly observable. Monitoring, analyzing and extracting insights from such discussions is a challenge in itself, given the variety of sources where they take place, the heterogeneity of content and format in which they are available, and the difficulty in telling human agents from chatbots. However, there is potentially large value in mining these discussions both for identifying issues and bottlenecks in existing chatbot dialogues and in extracting patterns of natural interaction among human agents.

The user knowledge challenge: Chatbots, are to communicate with a varied group of people across gender, age, languages and preferences. However, new technologies too often create digital divides and biases across gender, age, and societal status [7,8], and new knowledge is needed to mitigate these. Furthermore, user models need to be developed and applied to adapt chatbot interaction to individual users' requirements.

The conversational challenge: Adapting to users' contexts, chatbots need to support interaction sequences that go beyond more than a small number of steps. For each step in the interaction sequence, a range of valid responses exist. In addition, the chatbot responses need to fit the conversational context. Failure in this regard make the system's interpretation and response to user input prone to error, with conversational breakdown as consequence. Research is needed on artificial intelligence and machine learning to effectively identify and adapt to the conversational context.

Chatbot design and operation: In design of natural language user interfaces in general, and human-chatbot interaction design in particular, we will need to move beyond the current state of the art by combining and extending the fields of information systems, software engineering and human-computer interaction (HCI). In particular, incorporating state-of-the-art knowledge and methods from the emerging field of machine learning will be important. For example, machine learning enable the development of generative interaction models, which in turn will be used to simulate interaction patterns and drive the analysis and (re-)design of natural language interactions with digital systems and services.


Open questions to be explored in the papers include but are not limited to the following:

  • How should we define and use machine learning effectively for the purpose of human-centred design when developing chatbots?

  • What are the features of existing chatbot frameworks and chatbot agents operating in the wild and what can we learn from them?

  • How to identify and harvest data of relevance from the Internet for assisting the design and development of chatbots?

  • Which analysis tools and techniques are available or needed for both the HCI-community and machine-learning community when analyzing unstructured and structured data?

  • What is the effect of different demographics and culture on the design of chatbots?

  • How to transfer analysis output into actionable insight about users?

  • How can human-centred design benefit from machine learning, sentiment analysis and topic modelling?

  • Lessons learnt from existing work on developing tools concerning chatbots?

  • How can we evaluate and compare the user experience and the quality of dialogue across different chatbots?

  • What are new applications and services that can be enabled by the next generation of chatbots?

  • How should privacy and ethical issues be considered when developing chatbots?


We invite three kinds of submissions:

  • Brief position papers or chatbot demonstrations (2-3 pages, low threshold contribution).

  • Theoretical or empirical contributions, including presentations of brave new ideas (6-12 pages, published in the Springer workshop proceedings).

  • Detailed demonstrations, e.g. of chatbot prototypes or conversational analysis approaches (6-12 pages, published in the Springer workshop proceedings).

All submissions should be prepared in the Springer LNCS format.

Submissions will be selected following peer-review, based on their originality, quality and ability to promote discussion amongst the workshop participants. Accepted theoretical or empirical contributions (6-12 pages) and demonstrations (6-12 pages) will be published in the workshop proceedings (Springer LNCS).

At least one author of each accepted paper is expected to attend the workshop for the paper to be included in the workshop proceedings.

Important dates

  • Submission deadline: August 19

  • Acceptance notification: September 15

  • Final submission deadline: October 15

  • Workshop: November 22


  • Asbjørn Følstad, SINTEF, Norway

  • Symeon Papadopoulos, Centre for Research and Technology Hellas, Greece

  • Ole-Christoffer Granmo, Uni. Agder, Norway

  • Petter Bae Brandtzæg, SINTEF, Norway