Research On The Establishment Of Song Dynasty Poetry Corpus And The Computational Methods Of Style Identification And Emotion Analysis

Computational Poetry is a new application field for computer natural language processing technology. The major work is to establish the poem corpus for the application of NLP knowledge mining in the assistant research on the poetry. To help us to know the Song Dynasty poetry, in the present paper, we set up the correlative annotated corpus and develop the primary research on computational methods of the style identification and emotion analysis. The main research outputs are as follows:The methods and experimental study introduced in this paper are all driven based on corpus database because of the needs of machine learning and digitalization of classical poetry. The corpus work involves the establishment of word list with statistical word extraction, segmentation of poetry based on the foregoing word list and rules and forms, annotation on the segmented poetry such as part-of-speech tagging and so on. Meanwhile, we also set up the correlative knowledge database. It's very import to establish the corpus well, because the whole research project bases itself on it. This part is explained from Chapters2 to Chapters5 in this thesis.Aiming at the differentiation of highly abstract artistic conceptual styles, we convert this problem into text categorization problem of Pattern Recognition. Basing on the above work, we found three categorization modules, including the module base on the"character"feature, the module base on the"word"feature, the linear combined module of fore modules which get the weight by the genetic algorithm. We compare these three modules with the KNN in this part, which is introduced in Chapters6.About the assistant research on the authors'emotions which are expressed in the poetry, we try to import the emotions and their computation to the machine understanding of poetry. The problem of the emotional meaning tagging was studied by using the multiple relaxation alternate algorithm. We designed an experimental system and obtained accurate emotion tagging matching according to the context information. It is the first step to the machine understanding of emotional meaning of poetry, thus the foundation for the future research is laid. The part of the content is covered in Chapters7 of this paper.The summaries and conclusions of the research work, as well as the suggestion for the further researches come at the end of paper.