Industry Collaboration
Our group collaborates with researchers from a variety of companies and agencies, including:
- Large companies: Google, Microsoft, Bloomberg, Canon Medical
- Startups: MoodAgent, PolyAI, Monzo
- Agencies: BBC Scotland, U.S. NIST
If you’re interested in collaborating on research there are opportunities for industrial PhDs, visiting researchers, and others. Please contact us to discuss details.
Current & Recent Projects
TREC Conversational Assistance Track (CAsT)
We are one of the lead organizers of the TREC CAsT track, to develop reusable benchmark collections for Conversational Information Seeking. Please see the CAsT website for details.
BBC - ‘Social’ Discovery of News and Media Content for Beeb
This is a joint project with the BBC Voice and AI group in Glasgow to research content discovery via the Beeb Assistant. It studies how to blend grounded social dialogue techniques with task-based goals to assist discovering a wide range of BBC content. It will allow people to engage in conversations where they can explore topics through conversation, pick out specific parts of the topic to explore further.
Amazon - Knowledge Grounded Product Search
This project studies the task of researching product information in a voice assistant, like Alexa. We study how to inform and educate people in order to make informed product purchasing decisions.
Bloomberg - Deep multi-task models for Knowledge-centric extraction and ranking
The goal of this project is to address retrieval and extraction for complex information analysis tasks. It is exploring the use of deep multi-task learning models that can both extract and rank relevant content.
Moodagent - Conversational Music Assistant
This project studies task-based music agents to intelligently incorporate context and conversational interaction. Going deeper, the aim is to develop conversational music intelligence that allows an agent to not simply perform tasks, but to converse with an agent about music.