人格特質對於產品推薦機制之影響

Using Personality Traits to Analyze the Acceptance Degree of Product Recommendation Systems

網路世界快速的發展縮短了消費者搜尋的時間及成本，但也縮短了消費者在網站上停留與瀏覽的時間。為了能不斷的吸引消費者的注意力，且試圖讓消費者在購物時能有更快的選擇，開始從原本簡單的購物介面，逐漸增加其許多推薦機制。卻也因為推薦方式太過多樣化，消費者反而會分散其注意力，降低推薦的效果。本論文透過人格特質分析角度，探究人格特質在不同推薦方式下的接受度，期能提升其個人化推薦的程度。由消費者提供的人格特質資訊，找出最適合其推薦方式再推薦其產品，期望在實務上幫助電子商務之網站在推薦產品能更貼近消費者的需要，增加消費者之忠誠度。研究發現，人格特質對於推薦方式具有相關影響性，其中又以自我監控、A型人格，在面臨不同推薦時影響性較高，風險偏好雖在本實驗中都沒有明顯的影響性，但在多變量分析上也發現風險偏好與自我監控交互作用，也出現了明顯的干擾情況。產品的性質也是影響推薦上很重要的因素。並且不同人格對於推薦方式的不同，也有出現偏好某種推薦的情形出現。因此在管理意涵上建議，電子商務平台在顧客管理上能分其監控度高低，就能依據其特性高低進行個人化方式推薦。監控度高者多以協同推薦方式，監控度低者多以內容推薦方式作為推薦;若電子商務平台在顧客管理上是以A型人格作為基礎，對於高A型人格者可以多以內容與協同方式做為推薦，尤其內容推薦最有吸引力，但對於低A型人格則否，因此可以嘗試其他推薦如：關聯式推薦，或是其它推薦方式。

Rapid development of internet has shorten the time and cost of consumers' search online but also shorten the time consumers spent on the site and browsing. To be able to continue to attract the attention of consumers and allow consumers a faster choice in shopping, online stores has evolved from the original simple interface to many recommendation mechanisms. But also because of the diversified recommendation methods, consumers are easily be distracted and decrease the effects of the recommendation.This paper attempts to explore the extent of acceptance of personality traits toward different mechanisms of recommendation. If we have a better understanding of which affect the relationship, we will be able to enhance the degree of personalization of the recommendation. Based on the personality information provided by the subjects to identify the most suitable way to recommend, we expect to help e-commerce sites recommend the products which are closer to the needs of consumers in practice,and finally, increase the loyalty of consumers. The study find that personality traits are related to the recommended ways, that is, self-monitoring, type A personality have more influence on the different ways of recommendation, risk preference has no significant effect in this experiment. However, in multivariate analysis also find that the risk preference and self-monitoring these two variables are interacted. And the property of the product is also a very important factor in recommendation. Furthermore, different personality traits have proved to have diverse preference to distinct recommendation ways. Therefore, the results suggest that e-commerce platform can classify the customers by their degree of self-monitoring to implement personal recommendation. Collaborative-based filtering is more suitable for high Monitoring trait while content-based filtering is more suitable for low monitoring trait; If the e-commerce platform can classify their customer based on the type A personality, both collaborative-based filtering and content-based filtering are suggested to used to high type A personality (especially content-based filtering) but not for low A personality. Thus, we can try the other recommendation ways such as: relational recommendation or other recommendation ways for low type A personality.