Predict method for Stupid Back-off text predictor — predict.sbo

Predictive text based on Stupid Back-off N-gram model.

# S3 method for sbo_predictor
predict(object, input, ...)

Arguments

object	a `sbo_predictor` object.
input	a character vector, containing the input for next-word prediction.
...	further arguments passed to or from other methods.

Value

A character vector if length(input) == 1, otherwise a character matrix.

Details

This method returns the top L next-word predictions from a text predictor trained with Stupid Back-Off.

Trying to predict from a sbo_predtable results into an error. Instead, one should load a sbo_predictor object and use this one to predict(), as shown in the example below.

Author

Valerio Gherardi

Examples

p <- sbo_predictor(twitter_predtable)
x <- predict(p, "i love")
x
#> [1] "you" "it"  "my" 
x <- predict(p, "you love")
x
#> [1] "<EOS>" "me"    "the"  
#N.B. the top predictions here are x[1], followed by x[2] and x[3].
predict(p, c("i love", "you love")) # Behaviour with length()>1 input.
#>      [,1]    [,2] [,3] 
#> [1,] "you"   "it" "my" 
#> [2,] "<EOS>" "me" "the"