Design and Implementation of an Artificial Intelligent Counsellor-Assistant Chatbot for University Students Using Machine Learning and Natural Language Processing
- Author
- Chandiwana, Liberty A
- Title
-
Design and Implementation of an Artificial Intelligent Counsellor-Assistant Chatbot for University Students Using Machine Learning and Natural Language Processing
- Abstract
-
Background: In 2018 the cabinet minister of Higher and Tertiary Education, Science and Technology Development, Minister Amon Murwira, confirmed disturbing development of a spike in suicide cases in Zimbabwean tertiary institutions and acknowledged the need to improve the counselling systems at tertiary institutions to deal with ‘mental health issues'.
Objectives: The first research objective was to analyse the different techniques used when developing an artificial intelligent chatbot for psychotherapy. The second objective was to design and implement an artificial intelligent chatbot using machine learning and natural language processing for addressing depression in university students. The third objective was to measure the effectiveness of machine learning and natural language processing in providing relevant feedback to students’ queries.
Methods: Qualitative approach was employed due to the subjective nature of the domain particularly content analysis to determine whether the obtained data set could help achieve set objectives. The dataset is presented in CSV format and has 31 topics with more than 100 questions under depression while the least number of questions were under military issues topic (less than 10). Additionally, the researcher went on to explore the distribution of responses in the dataset. It was found that the dataset had more responses given to depression questions. 317 responses were given to depression questions with the military issues being the topic with the least number of responses
Results: The model performance is ranked on its ability to classify whether user query is depressive and the ability to give relevant feedback. The researcher thus carried out 100 tests to determine the system’s effectiveness: Tests count (total number of tests performed) = 100. Total Yes (number of feedback responses that were relevant) = Test count – Total No (number of feedback responses that were not relevant) = 100 - 6 = 94. Therefore, system accuracy is set at 94/100 = 94%.
Conclusion: Using machine learning and natural language processing have the potential of reducing depression in university students as it addresses limitations embedded in the traditional counselling set up thus the proposed model encourages students to seek counselling services and nourish their therapeutic needs.
Recommendations: Given that most universities especially in Zimbabwe are primarily using the traditional method of counselling universities can take advantage of the rapid technological advancement and adoption wave especially in young adults to encourage students to seek counselling services using artificial intelligent powered chatbots. Furthermore, education and training for practicing psychotherapists and those in authority in Higher and Tertiary Education is required so that they appreciate the value of counsellor assistant chatbot(s) to reduce uninformed resistance to change
- Date
- 2023
- Publisher
- BUSE
- Keywords
-
Design and Implementation - Machine Learning
- Natural Language Processing
- Supervisor
- Mr Mhlanganiso
- Item sets
- Department of Computer Science
