rDppDiversity (version 0.0.2)

learnItemEmb: Machine learning algorithm to learn item representations maximizing log likelihood under DPP assumption.

Description

Machine learning algorithm to learn item representations maximizing log likelihood under DPP assumption.

Usage

learnItemEmb(
  train_data_path,
  emb_size,
  regularization,
  learning_rate,
  neg_sample_cnt,
  epoch
)

Arguments

train_data_path

A string for text file path. Each line: item_id,item_id,item_id

emb_size

int. ColumnNum for model parameter. While RowNum = number of uniq items parsed in train_data_path

regularization

float. Default = 0.1

learning_rate

float. Generally begin with small learning_rate will train better.

neg_sample_cnt

int.

epoch

int.

Value

A list contains 1) learned item embedding matrix; 2) item names vector; 3) log likelihood on each training step vector.

Examples

# NOT RUN {
library(rDppDiversity)
data_path=system.file("extdata", "data.txt", package = "rDppDiversity")
learnItemEmb(data_path, 3, 0.1, 0.01, 0, 10)
# }