The methodology of Metasynthesis enlightens the future of artificial intelligence system to combine kinds of methods. Further more, besides the logical thinking, we cannot ignore other aspects of thinking, especially the image thinking. To dabble at the image thinking, in this paper, I choose the field of research as the semantic of Chinese. "Comprehension" is a layered process. However, traditional methods of language understanding only pay attention to the semantic at the layer of logic thinking. If we want to delve the cognitive behavior, we also need to look into the semantic at the layer of image thinking and take the method of multi-layered integration. In this paper, the connectionism is chosen as a tool for learning the semantics of image thinking. This method is tested by learning the semantic representations of words belong to the field of Spring Festival Couplets. During the learning course of the neural network, semantic representations are changing from random lists to ordered ones. To some extend, this process is very similar to the cognitive behavior of human. Based on the semantic representations, a parallel model for Chinese language processing is proposed, in which modules of parsing, understanding and production are combined into one system. Finally, a computer art system for Spring Festival Couplets is constructed based on the proposed model, and the result is satisfied.
修改评论