結合序列規則及協同過濾之產品推薦方法

Abstract

客戶之購買行為會隨時間而有所差異，傳統協同式過濾推薦方法依據目標客戶之相似客戶購買行為進行推薦，並未考慮客戶在不同時期之購買行為。而序列規則推薦方法主要是分析客戶在不同時期之序列購買行為，以萃取客戶若在過去時期具有此序列購買行為，則目前時期會具備之購買行為何之序列規則。如果目標客戶過去時期購買行為符合(或相似)序列規則之過去時期購買行為，則可推論目標客戶於目前(推薦)時期可能會具備此序列規則之目前時期購買行為，並進行推薦，然而其並未考量目標客戶在目前推薦時期已有之購買行為。本研究提出一個新的混合式推薦方法，根據客戶最近購買時間，購買次數與金額進行客戶分群，並結合序列規則與協同過濾推薦方法進行推薦。所提方法針對每一客戶群，考量客戶序列購買行為進行序列規則推薦，並且考量目標客戶於目前時期之已購買行為進行相似客戶之協同過濾推薦。實驗結果顯示混合式推薦方法優於其它推薦方法。

Customers’ purchase behavior may vary over time. Traditional Collaborative Filtering (CF) methods use similar customers’ purchase behavior to provide recommendations to the target customer, without considering customers’ purchase behavior over time. The sequential rule-based recommendation method mainly analyzes customers’ purchase behavior over time to extract sequential rules with the form: purchase behavior over past periods => purchase behavior at current period. If a target customer’s purchase behavior over past periods is similar to the conditional part of the rule, then the purchase behavior of the customer at current period is predicted to be the consequent part of the rule. Although the sequential rule method considers customers’ purchase sequences over time, it does not make use of the target customer’s purchase data at current period. This work proposes a novel hybrid recommendation method that combines sequential rule and CF methods. The proposed method uses customers’ RFM (Recency, Frequency, and Monetary) values to cluster customers into groups with similar RFM values. For each group of customers, sequential rules are extracted from purchase sequences of that group to make recommendations. In addition, a KNN-based CF method is adopted to provide recommendations based on the target customer’s purchase data at current period. The results of the two methods are combined to make final recommendations. The experimental result shows that the hybrid method performs better than other methods.