Using Conversation Agent and Extracting Relations to Develop Chinese Learning Supporting System

• Main Authors: Dang DuyNgoc, 鄧惟玉
• Other Authors: Chih-Ping Chu
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/58396300531161810523

碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 100 === Nowadays, demand of learning Chinese language is very popular in the world, especially for the Vietnamese students who have been studying in Taiwan. There are many tools and methods used to support Chinese language learning, such as reading textbooks, listening to the tapes or CDs, watching news or movies, etc. But most of them are only one-way interaction that is different from the two-ways interaction in the class room. Learning through a virtual environment using computer is a good way to enhance the performance. The purpose of this study is to develop a web application system that supports foreign students learning traditional Chinese language. The behavior of this system is the learner inputs a sentence via keyboard and then the system responds a corresponding sentence. Besides, the system can recognize the current learner who is using the system by answering the personal questions like “what is my name?”, “how old am I?”, “how many people are there in my family?”, etc. given by the user. This system is developed on basis of ALICE (Artificial Linguistic Internet Computer Entity) - an artificial intelligence software which was developed by Dr. Wallace. The operation mechanism of ALICE is pattern matching technique. To reduce quantity of patterns and the occupation of memory, we use a database of “synonym” and “is-a” to convert different sentences with the same meaning into a sentence which matches a specific pattern. In addition, in order to make the computer more friendly and intelligent, we use DIRPE (Dual Iterative Pattern Relation Expansion) - a relations extracting method - to extract the relations that are based on historical conversations between learner and computer. The data extracted will be used for answering the personal questions. This system is implemented in C# language with databases Microsoft SQL server, a AMIL file (a version of XML), and a text file to store data.