R RKEA - Not enough training instances with class labels (required: 1, provided: 0)!

Question

I'm trying to get RKEA to work in R Studio. Here's my current code:

#Imports packages
library(RKEA)
library(tm)

#Creates a corpus of training sentences
data <- c("This is a sentence",
          "I am in an office",
          "I'm working on a laptop",
          "I have a glass of water",
          "There is a wooden desk",
          "I have an apple for lunch")
data <- as.data.frame(data)
data <- Corpus(VectorSource(data$data))

#Creates a corpus of training keywords
keywords <- c("sentence",
              "office",
              "working",
              "glass",
              "wooden",
              "apple")
keywords <- as.data.frame(keywords)
keywords <- Corpus(VectorSource(keywords$keywords))

#Creates output file for created model
tmpdir <- tempfile()
dir.create(tmpdir)
model <- file.path(tmpdir, "MyModel")

#Creates RKEA model
createModel(data, keywords, model)

This is mostly modelled after the example given in the RKEA documentation. However, when I run this I get the following error message:

Error in .jcall(km, "V", "saveModel") : 
  weka.core.WekaException: weka.classifiers.bayes.NaiveBayesSimple: Not enough training instances with class labels (required: 1, provided: 0)!

Answer 1

I think your example sentences are too short as documents. The following modification (primarily to the first sample document) works without error:

data <- c("This is a longer and longer sentence.",
      "I am in an office.",
      "I'm working on a laptop.",
      "I have a glass of water.",
      "There is a wooden desk.",
      "I have an apple for lunch.")

My guess is that with really short sentences, there aren't enough words that aren't keywords to use in building the model.