evidoria

← Back to browse

Good practice Imported

RMUTSB's LINE-Based AI Chatbot for University Admission and Educational Guidance

Thailand · Phra Nakhon Si Ayutthaya · See the Thailand profile

Evidence: Observational / pre–post Top 95% 20/100 · Ask Evidence Copilot about this practice

Researchers at Thailand's RMUTSB built an NLP chatbot on the popular LINE messaging app to answer prospective students' admission and tuition questions, reporting 97% technical accuracy and 92% satisfaction in a peer-reviewed 100-student usability test — a small, single-instituti

97 %
Technical accuracy (2025 study)
82 %
F1 score (2025 study)
92 %
Usability satisfaction rate (2025 study)
100 students
Usability test sample size (2025 study)

Details

Maturity
Pilot
Promoter
Rajamangala University of Technology Suvarnabhumi (RMUTSB)
Period
2025 (peer-reviewed pilot study)
Keywords
admissions, chatbot, natural language processing, higher education guidance

Context

Faced with limited admissions-office staff, researchers Nontachai Singhuang and Parinya Natho at Rajamangala University of Technology Suvarnabhumi (RMUTSB) developed an NLP/machine-learning chatbot to answer prospective students' questions about admissions, tuition fees and general educational guidance, deployed on LINE — Thailand's dominant messaging app — so that applicants could reach it through a channel they already use daily.

Activities

The team evaluated the chatbot with a mixed-methods study combining a technical accuracy assessment and a usability test with 100 students. The study was published as a peer-reviewed paper at the 2025 IEEE International Conference on Cybernetics and Innovations, framing the tool as a low-cost way to give applicants personalised, speedy answers to admissions questions without proportionally increasing administrative staff.

Results

The accuracy assessment reported 97% accuracy with an F1 score of 82%, and the usability test found a 92% satisfaction rate among the 100 students who used it.

Conclusions

This is a legitimate, peer-reviewed evaluation with real reported figures, but it should be read as a modest, single-institution research pilot: the usability sample was 100 students, and no information is available on whether the chatbot has since been rolled out university-wide across multiple admissions cycles, or on longer-term outcomes such as changes in application volumes or reductions in staff workload.

Implementation

Indicative cost
Low (< €50k) — Not disclosed; framed explicitly as a low-cost way to handle admissions queries without proportionally increasing administrative staff, built on an existing messaging platform (LINE).
Time to results
Short (< 1 year) — Developed and evaluated as a 2025 research pilot; study published at IEEE ICCI 2025 (document 10987355).
Staffing & skills
Developed by researchers Nontachai Singhuang and Parinya Natho at RMUTSB, Deployed on LINE, a messaging app applicants already use daily, minimising the need for new infrastructure or dedicated staff

Conditions for success

  • Deployment on an already-dominant messaging channel (LINE) to maximise reach without a dedicated app
  • Mixed-methods evaluation combining a technical accuracy assessment and user-facing usability testing

Common failure modes

  • No information on whether the chatbot has been rolled out university-wide across multiple admissions cycles
  • No data on longer-term outcomes such as changes in application volumes or reductions in staff workload
  • Usability sample limited to 100 students at a single institution

Where it fits

Governance type
single-university research pilot
Scale
single institution (RMUTSB), 100-student usability sample
Income level
middle-income (Thailand)

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Replication kit

Reusable artefacts from this practice — as published by their sources.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful