| Under the background of Web2.0 Internet,word-of-mouth launched by opinion leaders and general consumers on content-based platform is having great influence.word-of-mouth communication shows new communication features such as large spread,fast speed,and strong anonymity and non-face-to-face contact,which have brought a great impact on traditional marketing methods and business models.Many studies believe that there is a close relationship between consumers’ purchase intentions and online word-of-mouth reputation of products.At the same time,how to use the high-efficiency and highly influential social capital of products’ Internet of mouth reputation to achieve more positive marketing effects and further tap the consumer’s consumption potential is the focus of the current consumer market.With the diversified development of the Internet,the form of word-of-mouth communication of general products has become more and more abundant,covering e-commerce messages and experience sharing on word-of-mouth platforms.However,not all types of products require comprehensive and specific content of word-of-mouth communication.Based on this point of view,we propose an notion: exploratory experience products,and we then discuss in detail the characteristics of its word-of-mouth communication.Generally speaking,consumers are very willing to master all information of a product before consuming it,as a method to eliminate the additional cost caused by the information asymmetry.But we talk about the movie is a special case.Moviegoers will not try to figure out the details of the movie before consumption.For example,few people would like to know the story of the movie before going to the cinema.Therefore,in this article,we give a definition of a special experience product such as the movie: an exploratory experience product.Normal experience-based product reviews usually focus on displaying product details,while movie reviews usually focus on recommending or not recommending with the aim to prevent spoilers,so the impacts of word-of-mouth communication are more obvious.According to this feature,we define the proportion of reviews that are consistent with the final result of the movie among all the reviews of a certain type of reviewer as the word-of-mouth communication effectiveness of this type of user.The higher the effectiveness of word-of-mouth communication for particular user,the more worthy this type of user is to be trusted,and the greater the influence of word-of-mouth communication.We tried to explore whether opinion leaders who are more active and have more fans on the Douban film platform are more effective in word-of-mouth communication.At the same time,we also analyzed whether the effectiveness of different types of word-of-mouth communication exists from the perspective of positive word-of-mouth and negative word-of-mouth.difference.Therefore,our core question can be summarized as: How to identify opinion leaders on Douban movie platform? And are there any differences in the effectiveness of word-of-mouth communication between opinion leaders and non-opinion-leader users,and is there a difference in the effectiveness of positive word-of-mouth communication and negative word-of-mouth communication? Our research conclusions confirm that the heuristic products of exploratory experience products such as movies have led opinion leaders to differ significantly from non-opinion-leader users only in the effectiveness of positive word-of-mouth communication.Based on the previous description of the research background and research issues,combined with the analysis of related references,this article takes the opinion leaders and non-opinion-leader users of the movie word-of-mouth communication platform as the research object,and analyzes the effectiveness of opinion leaders and non-opinion-leader users in word-of-mouth communication from word-of-mouth communication.The differences and characteristics in the above,as well as the reasons for these differences,will ultimately provide a reference for platform users,platforms and movie distributors to achieve efficiency enhancement.The specific research content of this article includes the following four aspects:1.Based on previous research,elaborate relevant literature such as word-of-mouth,hedonic products,affect confirmation hypothesis,opinion leaders and so on.In the review of word-of-mouth,according to the different manifestations of word-of-mouth communication,the types of products are distinguished,movies are defined as exploratory experience products,and the research summarizes the characteristics of film word-of-mouth communication.Since previous research on word-of-mouth communication has focused on the research on the usefulness of reviews,and the usefulness of reviews does not have too much practical significance for the word-of-mouth communication of exploratory experience goods,so this article proposes word-of-mouth based on the actual characteristics of film word-of-mouth communication The concept of communication effectiveness is used to analyze the characteristics of word-of-mouth communication effectiveness of opinion leaders and non-opinion-leader users.2.Introduce the content of the Douban movie platform in detail,use MATLAB regularization language to extract the movie information and user information of the Douban movie platform,and use a large number of data visualization techniques to show the correlation between the characteristics of these data and each other.The opinion leader identification model establishes the foundation.3.Based on first-hand data such as movie information and user information on Douban platform,combining the characteristics and structure of these data,a model is established to identify opinion leaders.We named the model as the BLIP model: Bayesian approach to detect the opinion leaders by leveraging the user’s introduction and the degree of participation in comments.The BLIP model is an unsupervised learning model based on the Bayesian framework.Three sub-models work together to complete the identification of opinion leaders.Among them,the hybrid Poisson model classifies opinion leaders and non-opinion-leader users according to the number of user hot reviews,the hybrid Bernoulli model classifies opinion leaders and non-opinion-leader users according to whether there is promotion information on the user profile page,and the Logistic classifier is based on the user ’s previous The movies that have been watched and the average number of likes of hot reviews provide a priori distribution for the latent variables corresponding to the above two Bayesian classification models.The BLIP model comprehensively considers the characteristics of the three models and automatically gives the final probability that each user is an opinion leader.This article also shows the detailed steps of using the EM algorithm to iteratively solve the BLIP model,and the data visualization and detailed analysis of the final results.4.After using the BLIP model to obtain information about whether the user is an opinion leader,a comprehensive analysis of the effectiveness of word-of-mouth communication is conducted between opinion leaders and non-opinion-leader users,and related theories are used to explain the reasons for the differences.This article mainly has the following innovations:1.In the research process of word-of-mouth communication,this article classifies products according to the characteristics of word-of-mouth,proposes and distinguishes the concept of exploratory experience goods,and proposes the effectiveness of word-of-mouth communication based on the characteristics of word-of-mouth communication of exploratory experience goods.This new research perspective,rather than continue to study the usefulness of reviews as in previous literature.The effectiveness of word-of-mouth communication can be closely linked with the reviewers,and an analysis of the differences and characteristics of opinion leaders and non-opinion-leader users in word-of-mouth communication can be analyzed.At the same time,the effectiveness of word-of-mouth communication can better compare the influence of different types of users on word-of-mouth communication.2.Based on the data structure and characteristics of Douban movies,this paper proposes a BLIP model to identify opinion leaders.The BLIP model is an unsupervised learning model based on the Bayesian framework.The three-word model cooperates to complete the identification of opinion leaders,without the need for human intervention in the learning process.Among them,the mixed Poisson model is responsible for classifying opinion leaders and non-opinion-leader users according to the number of user hot reviews.The mixed Bernoulli model classifies opinion leaders and non-opinion-leader users according to whether there is promotion information on the user profile page.The Logistic classifier is based on the user ’s properties like the numbers of movies that have been watched and hot reviews,provide a priori distribution for the latent variables corresponding to the above two Bayesian classification models.The BLIP model comprehensively considers the performance of the three models to give the final probability that the user is an opinion leader.Compared with previous models,the BLIP model does not need to manually mark opinion leaders and can use its unsupervised learning model to automatically classify opinion leaders.The BLIP model’s Logistic classifier sub-model can identify the characteristics of opinion leaders as in the previous supervised learning model and select variables through saliency.BLIP model has the advantages of both unsupervised learning model and supervised learning model.Solved by the EM algorithm,the BLIP model can guarantee the calculation speed and accuracy.3.After identifying opinion leaders,this article uses the effective number as the number of comments belonging to the opinion leader in all comments,instead of directly setting the threshold value to determine the opinion leader’s comments the number and effective number can ensure that the number of comments of the opinion leader is relatively stable,and will not fluctuate because of the size of the sample size,making the subsequent chi-square test more stable.4.This article uses the effectiveness of word-of-mouth communication to demonstrate the rationality of the affect confirmation hypothesis from the perspectives of opinion leaders and non-opinion-leader users,and from the perspectives of positive word-of-mouth and negative word-of-mouth.5.This article uses machine learning technology and big-data methods to identify opinion leaders on the Douban film platform.Compared with previous methods such as questionnaires and interviews,this article has more objective experimental results,which can be completely repeated.Besides,the scale of the data obtained is also Larger,so can ensure the reliability and stability of the conclusion.In the current Internet era where flow is king,the influence of opinion leaders on word-of-mouth communication is becoming more and more obvious.Compared with non-opinion leaders,it is generally believed that opinion leaders rely on many fans and more professional appreciation ability,so they have a stronger influence on word-of-mouth communication.This paper uses the data of Douban film platform,which has a leading advantage in China,and establishes a model by using machine learning and empirical Bayesian framework.The conclusion is as follows: opinion leaders have a stronger influence on word-of-mouth in hedonic goods,but the advantage of opinion leaders only exists in the process of positive word-of-mouth communication,and in the spread of negative word-of-mouth,this advantage of opinion leaders has disappeared,and the emotional certainty hypothesis plays a leading role.The conclusion of this article has important practical significance for user’s word-of-mouth communication platforms and movie distributors.1.This article puts forward the concept of word-of-mouth communication effectiveness based on the characteristics of the word-of-mouth communication of exploratory experience products and combines the types of users to reveal the characteristics and differences of opinion leaders and non-opinion-leader users in word-of-mouth communication effectiveness.On the one hand,the research results can help users more accurately identify the effective information in the film word-of-mouth communication,improve the efficiency of users’ decision-making during the film consumption process,and optimize the consumer experience;Guidance.2.The BLIP model proposed in this paper uses Bayesian framework and big-data technology to efficiently and comprehensively identify the opinion leaders in the Douban film platform and does not need to manually mark the data in the recognition process to ensure the objectivity of the recognition.The identification of opinion leaders and non-opinion-leader users can help the platform to carry out various hierarchical marketing and traffic guidance and improve the efficiency and quality of user management of the platform.3.Although the quality of the film is the key to word of mouth,this paper analyzes the relevant conclusions of the characteristics and differences in the effectiveness of word-of-mouth communication between opinion leaders and non-opinion-leader users through quantitative research.Carrying out film word-of-mouth marketing to provide reference,improving the efficiency of film word-of-mouth communication,and helping related companies achieves greater profitability. |
PDF Full Text Request