Natural Language Programming is a new direction in the field of natural language Processing. This thesis focuses on an in-depth research of Chinese Natural Language Programming. Grounded on previous work, this thesis presents some innovation methods and ideas, builds two Natural Language Programming test systems based on two different methods, and compares the experimental results.The main contributions of this thesis are summarized as followings:1. A Large number of user investigations and tests are carried. Based on these user investigations and tests, not only enough relevant corpus is collected, but also the definition of the task for our Natural Language Programming system is made clear.2. Three components, including a Word Segmenter, a Part-Of-Speech tagger and parser, are built. These components are the basis of a Chinese Natural Language Programming system. They provide lexical and syntactic information for Natural Language Programming.3. Two Natural Language Programming systems are built separately by using two different methods. One method is based on rules; another is an improvement of Transformation-based Error Driven Learning. Each system is described carefully on its modules and entire process.4. Finally, based on the comparative experiments of two systems, in-depth analysis of the advantages and disadvantages are given. Some Discussions and conclusions on the major difficulties and future prospects of Natural Language Programming are also given. |