Interest-Forgetting Markov Model
for Next-Basket Recommendation低收入家庭申请书
Jinghua Zhu(&),Xinxing Ma,Chenbo Yue,and Chao Wang
School of Computer Science and Technology,Heilongjiang University,
Harbin150080,Heilongjiang,China
zhujinghua@hlju.edu
Abstract.Recommendation systems provide urs with ranked items bad on思美人兮
individual’s preferences.Two types of preferences are commonly ud to
秦振华generate ranking lists:long-term preferences which are relatively stable and
short-term preferences which are constantly changeable.But short-term pref-
网推广erences have an important real-time impact on individual’s current preferences.
In order to predict personalized quential patterns,the long-term ur prefer-
ences and the short-term variations in preference need to be jointly considered
for both personalization and quential transitions.In this paper,a IFNR model
is propod to leverage long-term and short-term preferences for Next-Basket
苏菲亚的故事
recommendation.In IFNR,similarity was ud to reprent long-term prefer-
ences.Personalized Markov model was exploited to mine short-term preferences
bad on individual’s behavior quences.Personalized Markov transition
畅想的意思
matrix is generally very spar,and thus it integrated Interest-Forgetting attri-
bute,social trust relation and item similarity into personalized Markov model.
Experimental results are on two real data ts,and show that this approach can
improve the quality of recommendations compared with the existed methods.
Keywords:MarkovÁSocial trustÁNext-BasketÁRecommendationÁ
Interest-Forgetting
1Introduction我的体育
Recommendation system has become a basic function of the online application nowadays,which can recommend products or rvices that meet ur’s preferences. Recommendation system can discover the internal relationship between urs and items according to the interaction between urs and items[1].The goal of item recom-mendation is to recommend a specific list of items to each ur.Most existing methods do not consider the recent behavior of urs but recommend items bad on the whole purcha history of urs.Predicting the ur’s recent behavior requires recommen-dation bad on the time quence of the items purchad.
In order to reasonably predict the ur’s personalized preferences,it is necessary to combine the ur’s long-term preferences with short-term preferences.The long-term preferences means that the ur’s preferences are inherently have not changed with time.For example,if the ur prefers a certai
n brand of goods,the purcha behavior of the brand’s goods will not change with time.Short-term preferences means that the
花样网名©Springer Nature Singapore Pte Ltd.2019
X.Cheng et al.(Eds.):ICPCSEE2019,CCIS1058,pp.20–31,2019.
doi/10.1007/978-981-15-0118-0_2
