Design and Implementation of an AI Base Chatbot system for Student Support Service

Based on a research synopsis prepared by Jamin Camilus Mafos

ID NO: Ms.c/IMT/24/0415

ABSTRACT:

Artificial intelligence chatbots have strong potential as automated student support tools in
higher education. This study explores the development and implementation of a chatbot using the
RASA framework to improve student support services. The chatbot provides 24/7 access to
information and delivers instant responses to student inquiries, including details on admissions,
registration, programs, and campus facilities.

Using data collected from interviews, observations, and literature review, the system was
designed to simulate human interaction and offer accurate, real-time assistance. The findings
show that chatbot technology can enhance service delivery, streamline administrative processes,
and improve the overall student experience in higher institutions.

KEYWORDS: AI Chatbot, Student support services, Higher Education, RASA Framework,

Automated Information Service
INTROUCTION

Artificial Intelligence (AI) transforms higher education by enhancing curriculum design,

optimizing administrative operations, and facilitating personalized learning experiences. AI-
powered tools, including AI chatbots, generative AI models, and AI-driven administrative
systems, are increasingly integrated into academic institutions to support student learning and
institutional management. Generative AI models, such as ChatGPT, are capable of creating
content and offering adaptive learning environments tailored to individual student needs,
fostering personalized educational experiences (Castillo-Martínez et al., 2024; Crompton &
Burke, 2023).

Additionally, AI-driven administrative systems optimize resource allocation, automate routine

tasks, and enhance student support services, thus improving overall administrative efficiency
(Kayyali, 2025).

This study explores the implementation of a chatbot specifically tailored to support student by
leveraging the RASA framework, an open-source platform for developing conversational AI
applications. Unlike conventional methods that rely on static question-and-answer formats, the
RASA framework allows for the integration of advanced Natural Language Processing (NLP)
and machine learning techniques to better understand and respond to user inputs dynamically.
Through this research, we aim to demonstrate how a well-designed chatbot can not only increase
accessibility and efficiency in the academic process but also enhance the institution’s digital
presence and appeal to tech-savvy prospective students.

Research on chatbots in education has grown exponentially since 2019. According to Scopus
(accessed February 2024), publications on this topic increased from 53 in 2019 to 255 in 2023,
underlining their rising importance in education.

Finally, researchers address the challenges and limitations related to chatbot implementation in
education. Kuhail et al. (2023) identified both technical and pedagogical challenges. Okonkwo
and Ade-Ibijola (2021a), and Zhang et al. (2023) outlined issues such as technical constraints,
limited user acceptance, and concerns around scalability. Huang et al. (2022) highlighted not
only technical challenges but also the novelty effect and cognitive load as barriers to effective
chatbot use. Du and Daniel (2024) additionally discussed the difficulties of implementing AI
chatbots, specifically in the context of English-speaking skills development.

While we acknowledge the efforts of prior review studies in addressing different aspects of
educational chatbots, several gaps remain unaddressed. First, an in-depth analysis of the
underlying technology of chatbots in education is missing. ‘Underlying technology’ refers to the
technology, algorithm, or model on which a chatbot is built. For example, with the underlying
technology, we aim to answer questions such as whether deep learning is used or whether the
chatbot has been developed using rule-based methods. While P´erez et al. (2020) briefly discuss
frameworks and language processors, they do not delve into specific models, algorithms, or
approaches. Huang et al. (2022), Hobert and von Wolff (2019), Jeon, Lee, and Choi (2023), and
Kuhail et al. (2023) only discuss the platform on which a chatbot was developed (e.g.,
smartphone, desktop PC, or smart speaker). Understanding the underlying technology can inform
educational chatbots’ design, implementation, and optimization, potentially enhancing their
effectiveness, addressing scalability challenges, and ensuring meaningful learning experiences.

Second, another gap that emerges is the lack of a comprehensive analysis of theories
underpinning the integration of chatbots in educational settings. While Zhang et al. (2023)
suggest which theoretical frameworks chatbots could be built on, they do not conduct a literature
review to see whether chatbots are based on these theoretical frameworks. Furthermore, Hwang
and Chang (2021) primarily analyzed the learning strategies used by educational chatbots but did
not systematically examine the broader educational theories that support these strategies. Our
study addresses this gap by not only analyzing theories concerning learning strategies but also
investigating other relevant educational theories, including Social Learning Theories, Personal
Development, and Motivational Theories.

Third, it is unclear based on which criteria chatbots are being evaluated. While several literature
reviews (Hobert & von Wolff, 2019; Hwang & Chang, 2021; Jeon, Lee, & Choi, 2023; Kuhail et
al., 2023; Lai & Lee, 2024; P´erez et al., 2020) examine methods and analyses used to evaluate
chatbots, such as questionnaires, Wizard-of-Oz experiments, and ANCOVA, only Hobert (2019)
briefly discusses the specific evaluation objectives measured. For example, if a researcher seeks
to evaluate purely using technical methods, such as accuracy and BLEU, which methods are
most appropriate, and how can they be integrated? Alternatively, for user-based evaluations, such
as psychological factors, what criteria can be measured, how are these assessed, and how can
these be combined with technical evaluations? This gap highlights the need for a more
comprehensive analysis to show which criteria are measured and how these criteria are
evaluated.

Furthermore, even when a study provides a detailed analysis of which criteria are measured and
how they are assessed, the effectiveness of chatbots remains unclear without examining the
actual results. Do the findings demonstrate that chatbots improve learning outcomes? Which
chatbot designs are most effective? Addressing this gap is crucial for providing researchers and
practitioners with actionable insights into chatbot effectiveness.

Lastly, no literature review thus far has analyzed the interactions among the objectives,
technologies, theories, evaluation criteria, and effectiveness of chatbot applications in education.
Such an overview could help future research by creating an understanding of the outcome these
factors have on the success of chatbot implementations in education. This can lead to more
informed pedagogical decision-making, more effective chatbot designs, and improved user
satisfaction. Moreover, this overview can also highlight which interactions have been
underexplored or overlooked, indicating which areas need further research.

Despite the benefits, chatbot adoption in education faces challenges, particularly regarding the
accuracy of responses and user acceptance. For example, highlighted limitations in the
adaptability of RASA-based chatbots, noting that responses could be limited if user inputs
diverge significantly from the trained dataset. Reported similar findings, observing that while
chatbots using deep learning models with RNN and LSTM performed well in accuracy and
speed, they required continuous training to maintain effectiveness.

Studies on chatbots for digital customer service and AI-driven customer interaction highlight the
importance of practical, scalable implementations for institutions with high information
demands, such as universities during admission period. emphasize that RASA-based chatbots
can enhance student engagement, offering a viable solution for institutions aiming to provide
continuous, responsive support.

Moreover, recent developments in NLP and AI are set to further enhance chatbot capabilities.
Research by on classification algorithms shows that integrating machine learning techniques,
such as K-Nearest Neighbors (KNN) and Naive Bayes, could further improve chatbot response
accuracy, an approach that could be beneficial for refining responses in the admissions process.

In conclusion, chatbot technology holds significant promise for higher education, particularly in
student support services, where efficiency and responsiveness are critical. The studies reviewed
highlight various approaches to chatbot design and the potential of NLP in improving interaction
quality. However, challenges remain in areas such as data availability and user adaptation,
suggesting a need for further research and refinement to fully realize the potential of chatbots in
educational settings.

REFRENCE

i. Belisano & Sarmiento, 2025; Chen et al., 2025; Debets et al., 2025; Hadiono et al., 2024;
Mashilo & Shekgola, 2024; Milano et al., 2025; Peyton et al., 2025)
ii. Belisano & Sarmiento, 2025; Chen et al., 2025; Debets et al., 2025)
iii. Belisano & Sarmiento, 2025; Debets et al., 2025; Peyton et al., 2025)
iv. Bolambao, F. J., Pogoy, A. M., and Plaisent, M. (2024). AI Chatbots in Higher Education.
Advances in Higher Education and Professional Development Book Series, 53–74.
[Link]
v. Xiao, N., Pei, Y., Yuan, C., Bu, Y., and Cai, Z. (2025). Transforming Education with
Artificial Intelligence: A Comprehensive Review of Applications, Challenges, and Future
Directions. 2(1), 337–356. [Link]
vi. Gonsalves and S. Deshmukh, “Designing an Interactive Chatbot for Educational
Assistance using Rasa Framework,” in 2023 International Conference on Intelligent
Data Communication Technologies and Internet of Things (IDCIoT), Bengaluru, India:
IEEE, Jan. 2023, pp. 68–74. doi: 10.1109/IDCIoT56793.2023.10053457.
vii. Maphalala, M. C., and Ajani, O. A. (2025). Leveraging artificial intelligence as a learning
tool in higher education. Interdisciplinary Journal of Education Research, 7(1), a01.
[Link]
viii. Crain, C. R., Ewing, A. W. G., Billy, I., and Anush, H. (2025). The advantages and
disadvantages of AI in higher education. The Business & Management Review, 15(03).
[Link]
ix. The role of ChatGPT in higher education: Benefits, challenges, and future research
directions. (2023). Journal of Applied Learning and Teaching, 6(1).
[Link]