Construction Of Mongolian Compound Noun Semantic Network Based On Neural Network

Faced with the increasingly diverse and complex development of natural language processing tasks,the lexical semantic web has gradually become an indispensable language resource in natural language processing tasks.At present,there are many mature lexical semantic webs with different architectures at home and abroad.in various types of natural language processing tasks.Compared with the well-developed Chinese and English Semantic Web,the research on the Mongolian Lexical Semantic Web has just started.At present,the Mongolian lexical semantic network only includes noun vocabulary,and the nouns in the existing Mongolian lexical semantic network are Mongolian simple words,and Mongolian compound words occupy a high proportion in natural language processing tasks,so in terms of Mongolian natural language processing tasks,there is currently a lack of Mongolian compound words in the Mongolian Lexical Semantic Network,which leads to the inability to obtain effective and high-quality results for related natural language tasks when completing task processing with the help of the Mongolian Lexical Semantic Network.Therefore,there is an urgent need for a Mongolian compound noun semantic network that includes Mongolian compound nouns and can be used for Mongolian-related natural language processing tasks.Based on the analysis of the existing Mongolian noun vocabulary semantic web related work,the thesis finds that the previous semantic web construction methods have problems such as time-consuming and labor-intensive artificial construction,and the construction process is strongly influenced by subjective consciousness.Although the current neural network technology is developing rapidly,the current construction of the semantic web has not introduced neural network-related technologies to reduce human effort.Therefore,the thesis proposes to apply neural network-related technologies to the construction of the Mongolian compound noun semantic web.This study has implemented the process among them,873,704 word sets of Mongolian corpus were collected,and 10 word vector evaluation sets were constructed,with a total of 8,972 pairs of evaluation words,and 80,892 Mongolian compound word vectors were constructed,realizing the extraction of compound noun synonyms from simple nouns in the Mongolian lexical semantic network Synset,It fully verifies the feasibility of constructing the semantic network of Mongolian compound nouns based on neural network.The research content of the thesis mainly includes the collection and arrangement of Mongolian corpus required by the neural network,the construction of Mongolian word vectors based on different models and parameters,and the construction of Mongolian word vector evaluation sets.Finally,based on the existing Mongolian lexical semantic web resources,the thesis will use neural network technology to realize the perfection of Mongolian lexical semantic web noun compounds and the leap from manual construction to semi-automatic construction.The innovations are mainly reflected in the internal evaluation of the static word vector Mongolian word vector,the construction of the Mongolian compound noun word vector,and the application of the word vector model in the construction of the Mongolian vocabulary semantic network.In the thesis finally realizes the supplement and improvement of the existing Mongolian lexical semantic network to automatically carry out concept synonym sets,and provides good upstream tool support for subsequent Mongolian natural language processing tasks.