| The recommendation explanation can increase the transparency of the recommendation system,so as to significantly improve the adoption and purchase intention of consumers.Therefore,how to add the appropriate explanation to the recommendation system has become a widely concerned issue.However,with the progress of technology,the recommendation interpretation is developing in the direction of precision,but the impact of recommendation explanation on perceptual diversity is ignored.Moreover,in order to increase the diversity of the recommendation system and alleviate the dilemma of accuracy and diversity faced by the recommendation system,most of the previous studies start from optimizing the underlying algorithm of the recommendation system.This thesis argues that the broad and precise recommendation interpretation will affect the perceived diversity and thus affect the recommendation effect.In addition,users with different experiences and different types of products have different demands on the diversity of recommendations,so the broad and precise degree of recommendation explanation will also produce different effects.In order to explore the above problems,this thesis conducted a preliminary experiment and two formal experiments by cooperating with a large game operation company in China to provide personalized recommendations for users through mobile phone SMS.The preliminary experiment completes the operational test of recommendation interpretation and proves that compared with broad recommendation interpretation,precise recommendation interpretation will reduce the perceived diversity of recommendation.Experiment one is used to verify the whole model.The experimental results further verified the influence of recommendation interpretation on perceptual diversity.It is also proved that,compared with accurate recommendation interpretation,the perception diversity of recommendation system using broad recommendation interpretation is higher,so broad recommendation interpretation is more effective for inexperienced users.The opposite is true for experienced users.On the other hand,it is proved that compared with the precise recommendation interpretation,the recommendation system using broad recommendation interpretation has higher perceived diversity.Therefore,when the product fit is low,broad recommendation interpretation is more effective.The opposite is true when product fit is high.Experiment 2 is similar to experiment 1,but we take user click behavior as a measurement to measure the recommendation effect,and use user information provided by the virtual world operator as sample data to further verify our conclusions and increase the external validity of the study.This thesis has great theoretical and practical significance.First of all,our study further explores the differences in users’ perception of diversity in the face of different recommendation interpretations(precise VS broad),and proposes measures and suggestions to improve recommendation interpretations and enhance the effect of recommendation interpretations.Secondly,from the two levels of products and users,this thesis also considers the differentiated demand for perceived diversity and the differentiated role of recommendation interpretation in the context of different user experience and product fit.Through our research,enterprises can understand how to choose appropriate recommendation interpretation methods when facing different target users and products,so as to maximize the effect of recommendation interpretation. |
