Research On Key Techniques Of Syntactic Analysis Of Tibetan Interrogative Sentences

Syntactic analysis is a basic technology for natural language understanding and the cornerstone of deep language understanding,which is indispensable in many natural language processing tasks such as semantic analysis,question answering systems,search engines,information extraction and retrieval.With the continuous progress and development for information technology,the requirements for syntactic analysis technology are becoming more strict,and more and more artificial intelligence applications rely on syntactic information to process and extract the meaning in text or speech.However,due to insufficient research intensity,lack of data resources,and poor technical level,the research on Tibetan syntax analysis has not yet achieved a major breakthrough.When study Tibetan syntactic analysis,many researchers have carried out research on all Tibetan sentence patterns.However,there are obvious differences in the grammatical structure and syntactic features of different Tibetan sentence patterns,which affect overall effect of Tibetan syntax analysis.If we study syntactic analysis based on its characteristics,we can improve performance of Tibetan syntactic analysis.Interrogative sentences are a common Tibetan sentence pattern.They are also the main sentence patterns in Tibetan question answering systems,search engines,information extraction,and retrieval.Therefore,the paper aims at Tibetan interrogative sentences,and studies the related techniques of syntactic analysis for Tibetan interrogative sentences from the following aspects.(1)Construction of Tibetan Syntactic Analysis CorpusBased on researched of web crawler technology and combined with characteristics of Tibetan,a crawler algorithm for Tibetan webpage text was designed,and the Tibetan corpus was collected and preprocessed.On this basis,a corpus of 2500 Tibetan parsers was constructed through segmentation,part-of-speech tagging,sentence extraction,phrase tagging,and syntactic tagging,which laid the foundation for the identification and syntactic analysis of Tibetan interrogative sentences.(2)Recognition of Tibetan interrogative sentences.Based on classification of Tibetan interrogative sentences and structural characteristics of various interrogative sentences,a Tibetan interrogative sentence recognition algorithm based on a syntactic tree is designed.Based on the designed algorithm,a Tibetan question sentence recognition system based on syntax tree was developed.Finally,through designing different experiments,the classification and recognition effect of the algorithm were examined.Experiments show that the algorithm can achieve good classification and recognition results.The average accuracy rate,recall rate,and F value of the classification reached 96.98%,100%,and 98.39%,and the recognition accuracy rate,recall rate,and F value reached 98.21%,100.00% and 99.10%.(3)Syntactic analysis of Tibetan interrogative sentencesAccording to the classification and inductive structural features of Tibetan interrogative sentences,the model of Tibetan interrogative sentence syntactic analysis based on PCFG(Probabilistic Context-Free Grammar,PCFG)is trained,and the Tibetan interrogative sentence syntactic analysis was completed,and a PCFG-based Tibetan interrogative sentences syntax syntactic analysis system was developed.Finally,through designing different experiments,the effects of syntactic analysis on training corpora of different scales was examined.After experimental testing,the highest accuracy rate,recall rate,and F value on the open test set reached 96.0%,96.1%,and 96.1%,respectively.The effect of syntactic analysis increased by 5.40 percentage points compared with the F value of the benchmark experiment,which indicated that(This article selects Tibetan interrogative sentences)and their characteristics.Syntactic analysis on them can obtain better experimental results.