ruimtehol (version 0.3.2)

embed_pagespace: Build a Starspace model for interest-based recommendation

Description

Build a Starspace model for interest-based recommendation (pagespace). For example a user clicks on a webpage.

Usage

embed_pagespace(
  x,
  model = "pagespace.bin",
  early_stopping = 0.75,
  useBytes = FALSE,
  ...
)

Value

an object of class textspace as returned by starspace.

Arguments

x

a list where each list element contains a character vector of pages which the user was interested in

model

name of the model which will be saved, passed on to starspace

early_stopping

the percentage of the data that will be used as training data. If set to a value smaller than 1, 1-early_stopping percentage of the data which will be used as the validation set and early stopping will be executed. Defaults to 0.75.

useBytes

set to TRUE to avoid re-encoding when writing out train and/or test files. See writeLines for details

...

further arguments passed on to starspace except file, trainMode and fileFormat

Examples

data(dekamer, package = "ruimtehol")
x <- subset(dekamer, !is.na(question_theme))
x <- strsplit(x$question_theme, ",")
x <- lapply(x, FUN=unique)
str(x)
set.seed(123456789)
model <- embed_pagespace(x, dim = 5, epoch = 5, minCount = 10, label = "__THEME__")
plot(model)
predict(model, "__THEME__MARINE __THEME__DEFENSIEBELEID")

pagevectors <- as.matrix(model)

mostsimilar <- embedding_similarity(pagevectors,
                                    pagevectors["__THEME__MIGRATIEBELEID", ])
head(sort(mostsimilar[, 1], decreasing = TRUE), 3)
mostsimilar <- embedding_similarity(pagevectors,
                                    pagevectors["__THEME__DEFENSIEBELEID", ])
head(sort(mostsimilar[, 1], decreasing = TRUE), 3)