YOUR TEAM: You will work in the Brain, Behavior and Cognition department of Philips Research Europe.
GOAL: The goal of this thesis is to design and implement a conversational agent that can act as an automated therapeutic assistant based on best practices in human-human health communication. The agent’s goal would be to deliver automated behaviour change counselling interventions for a particular health problem such as smoking cessation to help promote the user’s health status and prevent adverse consequences of their status quo. The agent can be linked to other interaction interfaces beyond speech to enhance the intervention content and tailor it to the user’s lifestyle, feelings, preferences and context. Examples of such additional interfaces are context-acquisition sensors (wearable physiological/ activity sensors, cameras, environmental sensors)
- Acquire and pre-process conversational databases for the selected application domain(s);
- Identify key features in the user’s interaction (speech and beyond), conversation flow and common topics relevant for driving the counselling dialogue conversation in a particular direction;
- Conceptualize characteristics and their relationships in a computational model;
- Set requirements and testing of the NLP/G technology from ROs 1&2 including other state of the art techniques such as topic modelling;
- Develop a hybrid computational model & logic for conversation management;
- Develop a learning system that adapts and optimizes the conversational logic based on the conversational data with the user over time;
- Test conversational models in real application cases (with volunteers/patients);
- Investigate the dynamics between the different counselling techniques and components integrated on one hand and the success and quality of the conversation on the other.
Expected Results: The result of this ESRs project will be an intelligent agent that combines novel approaches over different AI subfields such as computational linguistics, knowledge representation and reasoning, and ML. The resulting framework requires interdisciplinary contributions including formalizing counselling know-how, NLP/G technologies and statistical-relational learning and reasoning approaches. Adopting a hybrid nature, the system will allow the integration of different paradigms (i.e. data-driven and knowledge-based) into one unified framework. Both domain knowledge and data collection play a crucial role in the design and implementation of the computational logic driving the agent.
Additional essential requirements
- Master degree in Computer Science, Information Engineering (or equivalent). A degree with distinction (cum laude) is an advantage
Principal Investigator: Rim Helaoui (PHILIPS)
Academic PhD Supervisor: Prof. Diego Reforgiato Recupero (UNICA) and Daniele Riboni (UNICA)
Industrial PhD Supervisor: Rim Helaoui (PHILIPS)
Industrial PhD co-Supervisor: Arlette van Wissen (PHILIPS)
Main contact: Rim Helaoui