- 5 views
- 5 downloads
Finding reliable resources and Chatting with Mira while considering emotions when the scenario is unscripted
-
- Author / Creator
- Gharaat, Mohamad Ali
-
The primary wellspring of information, the Internet, abounds with misinfor-
mation, particularly in the domains of mental health and psychology. This also
affects the reliability and truth of responses of chatbots counting on unverified
data.
To address this concern, the MIRA project started to create a reliable
source of information and develop a chatbot that can deliver these verified
resources of information. By respecting privacy, anonymity of user data, MIRA
seeks to support individuals looking for a safe environment to find information
and assistance in their mental health journey.
The MIRA Project consists of the MIRA resource library and the MIRA
chatbot. In this thesis we develop the MIRA resource library in which our
domain experts define, edit, and evaluate resources. This also includes a re-
source search functionality that allows the MIRA chatbot to search efficiently
among the verified resources.
The MIRA chatbot employs scripted questions to detect user intents and
entities. In this thesis we add Chatty MIRA for producing empathetic replies
for scenarios in which the user is looking for just a conversation with the MIRA
chatbot. Chatty MIRA facilitates a rule-based text generator and an emotion
detection component, proficient in identifying emotions and producing em-
pathetic replies. As users engage in casual conversations with Chatty MIRA,
Chatty MIRA can simultaneously attempt to search for related resources based
on detected intents and entities. This allows for a more engaging experience
for the end-user, where MIRA can provide empathetic responses while also
offering relevant resources to support the users with relevant resources it can
find.
We conduct a human evaluation to assess the superiority of Chatty MIRA
over its rule-based and Large Language Model (LLM) competitors. As well as
determine the gap between Chatty MIRA and advanced LLMs (i.e. GPT-3.5). -
- Subjects / Keywords
-
- Graduation date
- Spring 2024
-
- Type of Item
- Thesis
-
- Degree
- Master of Science
-
- License
- This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.