Hybrid Recommendation Scheme Based on Deep Learning

Algorithm 2 XGBoost classification and labeling.

	Input: user list and product list , Residual data collection Data
	Output: User description category , product description category
(1)	Init sample same training weight
(2)	for data in Data do
(3)	Use XGBOOST to iterate m times to calculate the error rate through
(4)	Update the learning rate according to
(5)	Reset the weight of the nth sample as
(6)	repeat//after multiples iterations find the wrong samples by updating the learning rate and weight coefficients
(7)	Constructs a regression tree and input the data set as the root node in the form of a label;
(8)	Build the XGBoost model according to the objective function of formulas (3)–(6)
(9)	Defined the logistics loss function
(10)	Obtain the second-order partial deivative of the loss function update the learning rate
(11)	until N;
(12)	Add label description in US and IS
(13)	Final;
(14)	Return US, IS