• Home
  • DTL research plan

DTL research plan

This page describes the overall goals of DTL, and collects ongoing and planned work in DTL.

1 Research strategy

Fundamental research

Research in DTL focuses on the intersection of dialogue (spoken or multimodal, human-human or human-machine) and language and communication technologies (dialogue systems, mobile phones, etc.). Some specific areas of interest are

  • Formal dialogue theory
    • Semantic coordination in dialogue
    • Issue-based
    • ...
  • Dialogue corpora and analysis
  • Dialogue technology applications
    • Cognitive load management in in-vehicle dialogue systems
    • Dialogue systems in Augmented and Alternative Communication (AAC)
    • Dialogue agents on the Web

Semantic coordination in dialogue: An important next step in our work on dialogue is to develop theories of how humans coordinate their language and learn new language through interaction. We aim to investigate the possibilities of adapting machine learning techniques (currently oriented towards training on large datasets) to incremental learning on the base of input gained from dialogic interaction.

Cognitive load management in in-vehicle dialogue systems: ...

Dialogue systems in Augmented and Alternative Communication (AAC): ...

Dialogue agents on the Web: ...

Mathematical models of dialogues: We are developing a model for dialogues between human beings and computers. The ideas are based on type theoretical expressions with placeholders which are filled in by the participants of the dialogue.

...

Research infrastructure

To enable our fundamental and applied research, we also work on research infrastructure, including

  • Information State Update approaches to Dialogue Systems
  • Tools and methods for recording and analysing mobile communication, dialogue system
    interaction, human-human dialogue, and speech
  • A simulator platform for R&D on in-vehicle dialogue systems

Applications

The main application areas that we wish to initiate or develop further in the lab are the following:

  • Dialogue systems that can adjust their language to users, and learn language from users
  • Dialogue systems which minimize user distraction and increase safety
  • ...

Current projects

Here are the main research projects (funded and unfunded) currently being actively pursued in DTL (please add if anything is missing) (this list will hopefully be replaced by a subset of the official CLT project list):

  • Semantic Coordination in Dialogue (RJ, 2008-2012): Formal semantics and pragmatics for semantic coordination
  • SAICD (VR, 2009-2011)
  • LekBot (Vinnova, in cooperation with Talkamatic AB and DART): Dialogue systems for AAC (Augmented and Alternative Communication) 
  • Open-Source Toolkit for ISU dialogue management (internally funded, in cooperation with Talkamatic AB)
  • GPCC (GU): Tools and methods for recording and analysing doctor-patient communication
  • Corpus group (FLoV/Swedish) : Tools and methods for recording and analysing mobile communication, dialogue system interaction, human-human dialogue, and speech
  • Cognitive load management in in-vehicle dialogue systems (FLoV, PhD student funding)

Project ideas

  • Cognitive load management in in-vehicle dialogue systems
  • Semantic coordination in dialogue systems
    • VIP-DiiL: FP7 application, submitted 18/1 2011
  • Mobile Communication Lab
  • Dialogue systems and Antropomorphic AI (with Claes Strannegård)
  • ISU toolkit with incrementality and machine learning
  • Accessibility: same dialogue system, different modalities for different users: in-vehicle, elderly/disabled, 3rd world
  • Rapid development of dialogue systems
  • Speech & Dialogue Corpus development (VR Forskningsinfrastruktur)

Research todo-list

Things that need to be done (apart from getting the projects listed above)

  • Build the new ISU toolkit and include it in the CLT toolkit (with Talkamatic AB)
    • Need list of requirements and ideas towards implementation - what do we want, and how do we build it?
  • Buy ASR and TTS software (Staffan)
  • Buy car simulation equipment (Staffan)
  • Continue learning about machine learning, possibly in seminar form
  • Invite guest professor for 1 month; CLT may have funding for this

Research seminar ideas

  • Corpus annotation tools, standards and schemas
  • Multimodal dialogue systems using web standards
  • Specification of next-generation ISU toolkit

 

2 Public outreach strategy

Current collaborations

(with non-academic institutions)

  • Talkamatic AB
  • Spyderbrain Technologies AB
  • DART (Sahlgrenska Universitetssjukhuset)

Collaboration goals

  • Establish DTL as a centre for in-vehicle dialogue systems research and development in the VG region
  • Establish DTL as a centre for language technology research on accessibility and interaction for persons with communicative disabilities
  • Establish and maintain contacts with local industrial partners, in particular in the area of in-vehicle interaction
  • ...

Collaboration todo-list

  • Workshop / mini-conference on in-vehicle dialogue systems, possibly as a yearly event, combining academic research and industry work; GUHAB sponsor
  • Establish an industry partner program (?), e.g. allowing DTL to list commercial partners on webpage
  • Open house in the lab, with demos, posters etc; perhaps when we have installed the car simulator.
  • DTL Newletter circulated to industry partners and anyone interested

 


 

From the CLT application: Planned research activities

The following text is an excerpt from the CLT application. It is a bit out of date and also incomplete.

Fundamental research

An important next step in our work on dialogue is to develop theories of how humans
coordinate their language and learn new language through interaction. We
aim to investigate the possibilities of adapting machine learning techniques (currently
oriented towards training on large datasets) to incremental learning on the
base of input gained from dialogic interaction.

One of the promises of this research, as well as of related research in the area of in-vehicle dialogue systems, is the development of adaptive dialogue systems which put less cognitive stress on their users. Another goal is to make a connection between our work on speaker comparison and dialogue systems, so that, for example, systems can recognize users by their speech from among a small number of users registered with the system.

A further goal is to develop methods and tools for capturing and analysing linguistic interaction involving language technologies (dialogue systems and mobile communication technologies).

Applications

The main application areas that we wish to initiate or develop further in the lab are the following:

• dialogue systems that can adjust their language to users
• dialogue systems that can learn language from users
• systems that can differentiate among speakers by examining phonetic and
possibly other properties of their language
• tools and methods for analysing mobile communication and dialogue system
interaction


 

 

To the top

Page updated: 2011-09-09 11:16

Send as email
Print page
Show as pdf

X
Loading