微型汉语系统的构建及汉语文章的自动生成

In this paper, we established a mini Chinese language system that includes only a few Chinese characters in order to make the computer understanding the meanings of the sentences, the formation of Chinese words and how to automatically generate the articles by computer. We build the mini Chinese language systems based on the simple ecological system, and divided it into four layers. The four layers are the physical layer, link layer, rule layer and presentation layer. All the Chinese characters can form the connections between each other through the sentence learning at the link layer. We can also determine whether the computer understand the meaning of a sentence by the parameter of Sentence Weights. After sufficient numbers of sentences have been formed, we can get the Chinese words that composed of multiple Chinese characters from the high frequency Chinese character combinations. The mini Chinese language system can also output the long articles by using multiple rules in the rule layer to achieve the purpose of automatically generating articles by using computer. The Mini Chinese language system can identify the meaningful and meaningless sentences successfully from the results of the experiments. The Chinese words generated by the system are basically in line with the uses of modern Chinese language. We also try to use the N-gram model to solve the similar problem in this work, but we get the worse results. Although the computers abilities in automatically generating articles are relatively limited, the output format of the articles relatively simple by compared with the humans writing articles from the experiment results, but it can achieve the basic reading requirements in terms of readability. It also has a great room for improvement. The results from this work also show that there are a lot of potential to be developed in natural language processing by using the traditional computer architecture.