Skip to content
Studentships image background

Leveraging novel knowledge base and Graph Representation Learning for Mobile Mental Health Social Networks

  • School: School of Science and Technology
  • Study mode(s): Full-time / Part-time
  • Starting: 2023
  • Funding: UK student / EU student (non-UK) / International student (non-EU) / Fully-funded


NTU's Fully-funded PhD Studentship Scheme 2023

Project ID: S&T14

There is an apparent connection between our health and our affective state. This connection has encouraged researchers to produce numerous applications and interventions to facilitate patients and therapists. In particular, the two main pillars of this proposal: natural language processing (NLP) and graph representation learning (GRL), can be used to reduce the gap between the exponentially growing amount of biomedical data and our ability to make sense of them.

In the context of user-generated mental health data, characterized by its notorious heterogeneity and complexity, NLP and GRL have begun to enrich our toolkits for making sense and using the wealth of data beyond traditional rule-based systems or regression techniques.

This proposal is based on the cusp of such a paradigm shift, sketching potential exciting works across the fields of knowledge base and affective computing through the lens of natural language processing and graph representation learning. Additionally, fine-grained classification, few-shot learning, and data augmentation will be considered and included. This study will promote the development of AI for mental health and gear up to transform the world in new and exciting ways.

Aim of the project:

This research aims to develop a series of improvement works for NLP and GRL in clinical data and other health-related data that will provide:

1. Adapt Responsible Research & Innovation (RRI) methods to start the interdisciplinary affective computing and NLP works in NTU.

2. Prototype a user-centered supervised model training process to refine definitions and expose latent knowledge in improving affective computing.

3. Identify the key factors contributing to performance in the model pipeline (data, processing, augmentation, knowledge incorporation, and deployment) by examining domain expert knowledge and user profile based on graph representation learning.

4. Assess whether the user and mental health professional knowledge base in developing a digital mental health detection tool increase performance and trust.

Research Questions:

The main research questions, which will be investigated in this project, are as follows:

(1)   Incorporate user profile information with Graph neural network and, subsequently, improve mental health-related emotion analysis.

(2)   Neural network-based Text Classification with Knowledge-based Powered attention, retrieving knowledge from external knowledge sources to enhance the semantic representation of user-generated texts.

(3)   Exploring data augmentation and contrastive learning for fine-grained mental health classification, especially for long-tailed, imbalanced user-generated data.

(4)   Developing few-short learning models for small, sampled text classification with Lightweight Word Embedding or transformer-based models.

The studentship will leverage access to large mobile social networks dataset licensed to both Alan Turing Institute and Turing, see below.

Supervisory Team:

Prof. Eiman Kanjo, Computer Science Department, Turing Network Development Award Lead

Dr. Golnaz Shahtahmassebi, Math & Statistics, Senior Lecturer

Dr. Ismahane Cheheb (ECR), Computer Science department

Dr Daria Kuss, Associate Professor, Psychology Department, Advisor

Entry qualifications

For the eligibility criteria, visit our studentship application page.

How to apply

To make an application, please visit our studentship application page.

Fees and funding

This is part of NTU's 2023 fully-funded PhD Studentship Scheme.

Guidance and support

Application guidance can be found on our studentship application page.

Still need help?

+44 (0)115 941 8418