如何使用网状结构从R运行Python gensim函数

Question

I'd like to use the reticulate package to run gensim from R. I'm not sure I fully understand the syntax of reticulate because I can get this to work with the default function settings but it fails when I try to pass more arguments.

library(reticulate)
gensim <- import("gensim")

model<-gensim$models$Word2Vec$load("word2vec_gensim")

matrix(unlist(model$wv$most_similar("queen")),ncol=2,byrow=T)
      [,1]                           [,2]               
 [1,] "princess"                     "0.76466166973114" 
 [2,] "king"                         "0.728749990463257"
 [3,] "prince"                       "0.653270363807678"
 [4,] "lady"                         "0.611525416374207"
 [5,] "consort"                      "0.609499335289001"
 [6,] "duchess"                      "0.608054518699646"
 [7,] "monarch"                      "0.606827557086945"
 [8,] "lady-in-waiting"              "0.605596661567688"
 [9,] "empress"                      "0.602727890014648"
 [10,] "wiki/margrethe_ii_of_denmark" "0.59738427400589"

but...

matrix(unlist(model$wv$most_similar("queen",topn = 25)),ncol=2,byrow=T)

Error in py_call_impl(callable, dots$args, dots$keywords) : 
TypeError: Partition index must be integer

Here "word2vec_gensim" is a pre-trained model, I can't include it because it is a large file but pick your fav pre-trained model. I think my issue is in how I'm providing additional args to the python function.

EDIT:I figured it out

looks like the R to python communication doesn't handle numbers as expected.

matrix(unlist(model$wv$most_similar("queen",topn = as.integer(25))),ncol=2,byrow=T)

works

Answer 1

You are correct that reticulate doesn't handle the automatic typecasting. There is no way to programmatically determine this unless knowing specifically whether a particular Python parameter needs to be an integer or a float and then intercede, which is impossible in Python since Python is a typeless language.

You can get around it by explicitly casting your arguments as integers using as.integer(25) or you can use 25L instead of 25 .