Study On Construction Of Emotional Dictionary Of Uyghur Language

Sentiment analysis refers to the process of orientation analysis of internet texts,which is an active research object for studies in natural language processing(NLP),information retrieval(IR)and AI.The Belt and Road Initiative introduced by China’s central government not only promotes development for local enterprises in Xinjiang,but spurs progress in internet technologies.Under these favorable circumstances,the Uygur language,as a popular language in Xinjiang,has seen unprecedented evolution in this field.Numerous websites in Uygur language have been set up to meet widespread local popularity.Penetration of this kind of dynamic communication platform has led to a sharp rise of the volume of texts in Uygur language.As a result,efficient and accurate classification of the language expressed in these texts will serve governments,enterprises and individuals in substantive ways.The Uygur language belongs to the Altaic family and is agglutinative,which differs from English and Chinese in term of structures.And its grammatical functions are achieved through morphologic changes.These factors combined add to the processing difficulties by computers.From another perspective,every change will abide by certain grammatical rules and exist in a certain expression.Therefore,analyzing and mastering internal lingual patterns,and basing computer technologies on them shall improve accuracy of semantic analysis.Firstly,this paper sketches the background and importance of sentiment analysis and research.Secondly,this paper illustrates the current status of research on sentiment analysis in English and Chinese as well as the construction method of sentiment dictionary.Lastly,this paper further discusses the construction method and process of sentiment dictionary in Uygur clearly.This study is conducted in two step: step one is to analyze and summarize fundamentals rules of sentimental vocabularies of the Uygur language by carrying out detailed contextual researches.These rules are derived via referencing to sentimental vocabularies in English and Chinese,which focuses on grammatical features and idiomatic uses of the Uygur language,and creating identification models for sentimental vocabularies and the rules base for analyzing polarity of Uygur vocabularies.Step two is to automatically identify sentimental vocabularies according to the characteristics model and use the rules base to judge appraisal tendencies.This process utilizes TF-IDF and machine learning approaches to improve identification rate and accuracy of polarity assessment.