| Ontology is an explicit specification of a conceptualization.As a common realization form of knowledge base,ontology is a formal description of concepts in a domain.As the communication between human and machine becomes more and more extensive,the structured language knowledge base plays an important role in the development of natural language.Meanwhile,the shared concept and the formal expression of knowledge and the complex relationship description function of ontology have been widely applied in the construction of knowledge base.The construction of Tibetan emotion vocabulary ontology database can not only promote the development of Tibetan text informatization,but also provide important judgment basis for text emotion analysis and emotion tendency calculation in the field of Tibetan natural language processing.Due to the lack of relatively standardized and publicly available lexical semantic ontology resources in Tibetan natural language processing,and the need to invest a lot of time and energy in ontology construction,this thesis explores and studies how to mine the information we need from unstructured textual corpus and construct affective lexical semantic resources.At the same time,based on the reuse of other excellent ontologies,Tibetan emotional words and lexical emotional information are automatically extracted from the text and the ontology structure and framework of emotional words are manually constructed.The semi-automatic method is used to realize the ontology library of emotional words.This method can not only guarantee the information accuracy of emotional word ontology,but also save time and labor,and improve the efficiency of ontology construction to a large extent.Firstly,the emotion information and semantic features of Tibetan emotion words are analyzed and studied from the perspective of lexical semantics and statistical linguistics,so as to provide accurate targets for the acquisition range of Tibetan emotion words ontology knowledge.The ontology knowledge of emotion words is acquired by combining manual collection and automatic acquisition,including emotion information such as emotion classification,polarity and intensity,and other lexical semantic knowledge.In order to reduce the workload of ontology construction and save time,a Tibetan emotion dictionary construction method based on SO-PMI is proposed,and a modern Tibetan emotion dictionary in the field of social media is constructed.Secondly,according to the semantic information and emotional information of Tibetan emotion words,Protege ontology construction tool is used to design the basic framework of Tibetan emotion words ontology.The ontology knowledge of emotion words is obtained from Tibetan emotion dictionary on the basis of reusing Chinese emotion words ontology,and the classes and attributes of Tibetan emotion words ontology as well as the hierarchical structure and relational constraints between them are defined.Emotional vocabulary ontology instances are created and corresponding attribute relationships are associated,and then the emotional vocabulary ontology is formally represented by OWL ontology description language and saved as an owl file.Finally,on the basis of the existing emotional vocabulary,the semantic similarity calculation method of word vector text of shallow neural network is used to calculate the semantic information of Tibetan emotional vocabulary,realize the semi-automatic expansion of Tibetan emotional vocabulary ontology,and compare the effects of different training methods and models,and select the optimal Tibetan word vector model for the expansion of Tibetan emotional vocabulary ontology.In order to ensure the accuracy of the ontology,the accuracy of the ontology and the construction results of the ontology are manually verified and evaluated,and the ontology construction results are continuously modified and improved in the process of verification and application.Through experimental result analysis and manual evaluation of ontology,the semi-automated ontology construction method has achieved good results in the construction and expansion of Tibetan emotional vocabulary,which verifies the feasibility of the method. |
PDF Full Text Request