Master's Thesis Murilo Bellatini

Last modified Jan 29

masterthesis

Enabling Personal Communication for Voice-Based Health Assistants in Geriatric Care

Motivation

Current digital assistants often lack the personal touch required in the sensitive field of geriatric care. They can appear robotic and fail to establish meaningful connections with elderly users. To address this issue, we have collaborated with our partner company to explore new ways of personalizing health assistants. By incorporating a dynamic graph-based user model, we aim to bridge the gap between technology and human connection in geriatric care.

Objective

The main objective of this research is to develop a system that provides personalized responses in geriatric care. This involves the following components:

Personal Knowledge Graph (PKG): A dynamic graph that captures user data from chat history, enabling tailored responses and enhancing engagement.
Recommendation Module: Contextualize conversations using the PKG to provide personalized suggestions, leveraging a Retrieval Augmented Generation approach.

By combining the PKG and recommendation modules, our system aims empowering digital assistants to deliver personalized and compassionate care to elderly users

Research Questions

To achieve our objectives, this research will address the following key questions:

Which concepts and entities should be included in the data model for effective personalization and user engagement?
What information extraction techniques can be employed to populate the PKG with pertinent information?
How can the extracted knowledge be integrated into the conversational framework to enable personalized responses?
What evaluation methods can be used to assess the performance and effectiveness of the developed system?

Incoming references

Files and Subpages

Name	Type	Size	Last Modification
20230606_BellatiniMurilo_ThesisExpose.pdf	File	776 KB	29.01.2024
20231208_BellatiniMurilo_MTKickOffPresentation.pdf	File	2,14 MB	29.01.2024
20231215_BellatiniMurilo_MasterThesis.pdf	File	20,30 MB	29.01.2024
20231215_BellatiniMurilo_MTFinalPresentation.pdf	File	9,09 MB	29.01.2024