Research On Scoring System Of Online Product Reviews Based On Discourse Markers Theory

The rapid development of Web technology and electronic commerce has greatly changed people's lives and their way of working.In order to enhance customers'interactive experience,most e-commerce websites allow and encourage consumers to make comments on what they have bought.Therefore,the quantity of online product reviews is also growing rapidly with the rapid development of e-commerce.In general,the reviews score can give consumers the overall value sense of the goods or services,and the text of the reviews provided a more detailed perception.Now the online reviews play a very important role in consumer purchase decisions making.However,owing to factors concerning psychology,economy,institutions etc,there is a certain gap between review content and review score,which results in score deviation,further make consumers caught in a buying decision dilemma—being vague,contradictory and hesitating.Online product review is an important information source which affects consumer purchasing decisions deeply,but the inconsistency between review content and score could bring troubles in the process of decisions-making.So the importance of solving reviews authenticity is focused on.The current researches on correction of comments deviation are mainly from the perspective of grammatical or semantic level,thus there is a great limitation on the effect and speed of correction.However,this paper attempts to introduce the Discourse Markers theory to make a correction by analyzing the content of review,achieving a breakthrough in the semantic and pragmatic level.Firstly,we build a library of discourse markers in the field of online product reviews and design a correction system including credibility,position,emotional attitude.Then,this paper get the weight of each index with improved analytic hierarchy process and propose an algorithm to calculate reviews' scores,which is more accurate.In the last,the algorithm is computerized by Python programming language and the results of example verification of Jingdong Mall show the method has better feasibility and validity.This paper will be divided into seven parts,and the specific content is as below:The first part is introduction.Research background,research purposes and significance,research content,key problems to solved,research methods and innovative works of this paper will be elaborated here.The second part is relevant researches.The types of reviews deviation,reasons of the deviation and correction method of deviation are discussed in this part.The third part is the instruction of Discourse Markers.The feasibility of Discourse Markers applied in the field of online reviews is analyzed.Also,this part introduces the types of discourse markers and builds a library of discourse markers.The fourth part propose a scoring system based on discourse markers.The content of reviews are analyzed form the perspective of credibility,position,emotional attitude,and the consistent scores are is gained which means the scores deviation is fine corrected.In the fifth part,the algorithm is computerized by Python programming language which realizes computer automation processing of content and scores at first.Secondly,online reviews of different categories of goods from well-known e-commerce site(Jingdong Mall)are downloaded,and then questionnaire and SPSS t-test for independent samples are adopted to verify the feasibility and effectiveness of the method mentioned in this paper.In the sixth part,a summary of what we have done and outlook will be discussed.