BUILDING AN ADMISSION QUESTION & ANSWERING SYSTEM FOR CAN THO UNIVERSITY OF TECHNOLOGY=XÂY DỰNG HỆ THỐNG HỎI ĐÁP TUYỂN SINH THUỘC TRƯỜNG ĐẠI HỌC KỸ THUẬT

BUILDING AN ADMISSION QUESTION & ANSWERING SYSTEM FOR CAN THO UNIVERSITY OF TECHNOLOGY=XÂY DỰNG HỆ THỐNG HỎI ĐÁP TUYỂN SINH THUỘC TRƯỜNG ĐẠI HỌC KỸ THUẬT - CÔNG NGHỆ CẦN THƠ

0 Người đánh giá. Xếp hạng trung bình 0

Tác giả: Binh Nguyen Diep, Ba Duy Nguyen, Phuc Tang Nguyen, Le Duy Anh Tran

Ngôn ngữ: eng

Ký hiệu phân loại:

Thông tin xuất bản: Tạp chí Khoa học và Công nghệ Cần Thơ, 2025

Mô tả vật lý: tr.109

Bộ sưu tập: Metadata

ID: 486402

Thêm vào giỏ

Phát triển hệ thống hỏi đáp là một bài toán đầy thách thức trong lĩnh vực Xử lý ngôn ngữ tự nhiên. Trong nghiên cứu này, chúng tôi xây dựng hệ thống hỏi đáp tuyển sinh dựa trên Rasa có khả năng ghi nhớ và trả lời chính xác các câu hỏi gặp phải trong giai đoạn huấn luyện. Bộ dữ liệu dùng để huấn luyện gồm 720 câu hỏi và 136 ý định được thu thập từ thông tin tuyển sinh của Trường Đại học Kỹ thuật - Công nghệ Cần Thơ, sinh viên và Internet đến năm 2024. Thông tin các câu trả lời lấy từ các chuyên gia tuyển sinh của trường. Dựa trên kết quả thu thập, chúng tôi phát triển hệ thống hỏi đáp tuyển sinh dựa trên web để trả lời các câu hỏi trực tuyến của người dùng về thông tin tuyển sinh. Kết quả thử nghiệm của hệ thống chứng minh mô hình hiểu ngôn ngữ tự nhiên đạt độ chính xác 97,4% trên bộ kiểm tra và độ chính xác được chuyên gia đánh giá là 90,1%.The problem of developing a question-and-answering system is a challenging task in the field of Natural Language Processing. Natural language is inherently ambiguous, and determining the semantic meaning of questions as well as identifying the correct answers poses significant challenges. In this research, we investigate a question-and-answering system based on the Rasa Framework. The system’s model is capable of memorizing and accurately answering questions it has encountered during the training phase. Additionally, the model can handle new questions during the testing phase by selecting an appropriate answer from the set of answers provided during training. The dataset used for training the model consists of 720 questions and 136 intents, collected from the admission information of Can Tho University of Technology, students, and the Internet up to 2024. The answers to the questions were gathered from the university's admission experts. Based on the results, we propose a question-and-answering process and develop a web-based admission question-and-answering system to replace advisors in responding to user inquiries online regarding the university's admission information. The system's experimental results demonstrate that the natural language understanding model achieves an accuracy of 97.4% on the test set and an expert-evaluated accuracy of 90.1%.

1. Đọc trực tuyến

1. Đọc trực tuyến

Tạo bộ sưu tập với mã QR