r - R: agrep とのあいまい一致が遅い

Question

パターンのベクトルと潜在的な一致候補の大きなベクトルがあります。の各要素について、で近似一致のリストを取得するためxに使用します。問題は、コードが非常に遅いことです。 からの各要素ごとに約 2 秒かかります。agrepyx

Is there a way to speed it up? x in this example is only 6 elements, but in real project x is of length 41k. y here is approximately 103k elements, which is close to real life.

サンプルの出力を表示する必要がある場合3300はy、1

前もって感謝します！

x = c("procter & gamble; tide free & gentle", "procter & gamble; tide he turbo clean", 
      "procter & gamble; tide simply clean", "procter & gamble; tide simply clean & fresh", 
      "procter & gamble; tide simply clean & sensitive", "procter & gamble; tide total care")

y = rep(c("procter & gamble; tide", "procter & gamble; tide & downy", 
          "procter & gamble; tide actilift", "procter & gamble; tide basic", 
          "procter & gamble; tide boost", "procter & gamble; tide boost vivid white + bright", 
          "procter & gamble; tide buzz", "procter & gamble; tide cold water", 
          "procter & gamble; tide colorguard", "procter & gamble; tide compact", 
          "procter & gamble; tide febreze", "procter & gamble; tide febreze sport", 
          "procter & gamble; tide high efficiency", "procter & gamble; tide oxi", 
          "procter & gamble; tide plus", "procter & gamble; tide plus colorguard", 
          "procter & gamble; tide pods", "procter & gamble; tide pods plus febreze", 
          "procter & gamble; tide pure essentials", "procter & gamble; tide simple pleasures", 
          "procter & gamble; tide simply clean & fresh", "procter & gamble; tide simply clean & sensitive", 
          "procter & gamble; tide stain release", "procter & gamble; tide stain release free", 
          "procter & gamble; tide to go", "procter & gamble; tide total clean", 
          "procter & gamble; tide totalcare", "procter & gamble; tide ultra 2", 
          "procter & gamble; tide vivid white & bright", "procter & gamble; tide with dawn", 
          "procter & gamble; tidekick"),3300)

mapped = as.matrix("",nrow=length(x))

myMap = function() {
  for (i in 1:length(x)) {
    mapped[i] = paste(y[agrep(x[i],y,max.distance=2.9,fixed=T,useBytes=T)],collapse = "|")  
  }
  return(mapped)
}

print(microbenchmark(myMap(),times=5))

タイミング

Unit: seconds
    expr      min       lq    mean   median       uq    max neval
 myMap() 11.57354 11.61535 11.6225 11.61919 11.64641 11.658     5

で 1 回だけ繰り返したサンプル出力y:

1    
2    
3   procter & gamble; tide simple pleasures|procter & gamble; tide simply clean & fresh|procter & gamble; tide simply clean & sensitive
4   procter & gamble; tide simply clean & fresh
5   procter & gamble; tide simply clean & sensitive
6   procter & gamble; tide total clean|procter & gamble; tide totalcare