使用R对单词中的相同模式进行分类

Question

我想进行文本挖掘分析，但遇到了一些问题。 使用dput()函数，我加载了一小部分文本。

text<-structure(list(ID_C_REGCODES_CASH_VOUCHER = c(3941L, 3941L, 3941L, 
3945L, 3945L, 3945L, 3945L, 3945L, 3945L, 3945L, 3953L, 3953L, 
3953L, 3953L, 3953L, 3953L, 3960L, 3960L, 3960L, 3960L, 3960L, 
3960L, 3967L, 3967L, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA), GOODS_NAME = structure(c(19L, 
17L, 15L, 18L, 16L, 23L, 21L, 14L, 22L, 20L, 6L, 2L, 10L, 8L, 
7L, 13L, 5L, 11L, 7L, 12L, 4L, 3L, 9L, 9L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L
), .Label = c("", "* 2108609 SLOB.Mayon.OLIVK.67% 400ml", "* 3014084 D.Dym.Spikachki DEREVEN.MINI 1kg", 
"* 3398012 DD Kolb.SERV.OKHOTN in / to v / y0.35", "* 3426789 WH.The corn rav guava / yagn.d / CAT seed 85g", 
"197 Onion 1 kg", "2013077 MAKFA Makar.RAKERS 450g", "2030918 MARIA TRADITIONAL Biscuit 180g", 
"2049750 MAKFA Makar.SHIGHTS 450g", "3420159 LEBED.Mol.past.3,4-4,5% 900g", 
"3491144 LIP.NAP.ICE TEA green yellow 0.5 liter", "6788 MAKFA Makar.perya 450g", 
"809 Bananas 1kg", "FetaXa Cheese product 60% 400g (", "Lemons 55+", 
"MAKFA Macaroni feathers like. in / with", "Napkins paper color 100pcs PL", 
"Package \"Magnet\" white (Plastiktre)", "Pasta Makfa snail flow-pack 450 g.", 
"SHEBEKINSKIE Macaroni Butterfly №40", "SOFT Cotton sticks 100 PE (BELL", 
"TENDER AGE Cottage cheese 10", "TOBUS steering-wheel 0.5kg flow"
), class = "factor")), .Names = c("ID_C_REGCODES_CASH_VOUCHER", 
"GOODS_NAME"), class = "data.frame", row.names = c(NA, -61L))

(NA是意外发生。) 这段文字是关于检查产品名称的。

我想将任何相似的名称分组。

例如，这里我手动选择MAKFA makar（乌克兰名称）。我找到了7行带有"root or key word MAKFA Makar"的内容。

Pasta Makfa snail flow-pack 450 g.
MAKFA Macaroni feathers like. in / with
2013077 MAKFA Makar.RAKERS 450g
2013077 MAKFA Makar.RAKERS 450g
6788 MAKFA Makar.perya 450g
2049750 MAKFA Makar.SHIGHTS 450g
2049750 MAKFA Makar.SHIGHTS 450g

所有产品位置都有相同的根词。 MAKFA Makar不能像 MFAMKR 这样。 输出应该是：

                                                Initially                 class
1                       Pasta Makfa snail flow-pack 450 g.          MAKFA Makar.
2                  MAKFA Macaroni feathers like. in / with          MAKFA Makar.
3                          2013077 MAKFA Makar.RAKERS 450g          MAKFA Makar.
4                          2013077 MAKFA Makar.RAKERS 450g          MAKFA Makar.
5                              6788 MAKFA Makar.perya 450g          MAKFA Makar.
6                         2049750 MAKFA Makar.SHIGHTS 450g          MAKFA Makar.
7                         2049750 MAKFA Makar.SHIGHTS 450g          MAKFA Makar.
8          * 3398012 DD Kolb.SERV.OKHOTN in / to v / y0.35                  kolb
9               * 3014084 D.Dym.Spikachki DEREVEN.MINI 1kg             Spikachki
10                                         809 Bananas 1kg              Bananas 
11                                              Lemons 55+                Lemons
12                           Napkins paper color 100pcs PL        Napkins paper 
13                         SOFT Cotton sticks 100 PE (BELL         Cotton sticks
14                     SHEBEKINSKIE Macaroni Butterfly №40 SHEBEKINSKIE Macaroni
15 * 3426789 WH.The corn rav guava / yagn.d / Cat SEED 85g              CAT seed
16                        FetaXa Cheese product 60% 400g (               Cheese 
17          3491144 LIP.NAP.ICE TEA green yellow 0.5 liter                  TEA 
18                  2030918 MARIA TRADITIONAL Biscuit 180g              Biscuit 
19                                          197 Onion 1 kg                 Onion
20                          TOBUSsteering-wheel 0.5kg flow        steering-wheel
21                     Package "Magnet" white (Plastiktre) Package  (Plastiktre)
22                    * 2108609 SLOB.Mayon.OLIVK.67% 400ml                 Mayon
23                            TENDER AGE Cottage cheese 10        Cottage cheese

我该如何通过词根分类产品？（即，通过Makar.Makfa和cheese这些单词中相同的模式来区分它们）

- d-max

Answer 1

我认为你可以通过清理和聚类文本来获得你想要的结果 - 这里是一个入门级：

text <- text[1:24,]
library(quanteda)
library(tidyverse)
hc <- text %>% 
  pull(GOODS_NAME) %>% 
  as.character %>% 
  quanteda::tokens(
    remove_numbers = T,  
    remove_punct = T,
    remove_symbols = T, 
    remove_separators = T
  ) %>% 
  quanteda::tokens_tolower() %>% 
  quanteda::tokens_remove(valuetype="regex", pattern = c("^\\d.*")) %>% 
  quanteda::dfm() %>% 
  textstat_simil(method = "jaccard") %>% 
  magrittr::multiply_by(-1) %>% 
  `attr<-`("Labels", text$GOODS_NAME) %>% 
  hclust(method = "average") 

pdf(tf<-tempfile(fileext = ".pdf"), width = 20, height = 10)
plot(hc)
dev.off()
shell.exec(tf)

clusters <- cutree(hc, h = -0.1)
split(text, clusters)

- lukeA

Answer 2

这里有一种使用单词向量进行搜索的方法：

patt <- c("MAKFA Makar.", "kolb","Spikachki", "Bananas", "Lemons",
"Napkins paper", "Cotton sticks","SHEBEKINSKIE Macaroni","CAT seed","Cheese",
"TEA", "Biscuit", "Onion", "steering-wheel", "Package  (Plastiktre)",
"Mayon", "Cottage", "cheese")

lst <-lapply(patt, function(x) text[grep(x,text$GOODS_NAME), ])
do.call(rbind.data.frame, lst)

   ID_C_REGCODES_CASH_VOUCHER                                              GOODS_NAME
15                       3953                         2013077 MAKFA Makar.RAKERS 450g
19                       3960                         2013077 MAKFA Makar.RAKERS 450g
20                       3960                             6788 MAKFA Makar.perya 450g
23                       3967                        2049750 MAKFA Makar.SHIGHTS 450g
24                       3967                        2049750 MAKFA Makar.SHIGHTS 450g
22                       3960              * 3014084 D.Dym.Spikachki DEREVEN.MINI 1kg
16                       3953                                         809 Bananas 1kg
3                        3941                                              Lemons 55+
2                        3941                           Napkins paper color 100pcs PL
7                        3945                         SOFT Cotton sticks 100 PE (BELL
10                       3945                     SHEBEKINSKIE Macaroni Butterfly №40
17                       3960 * 3426789 WH.The corn rav guava / yagn.d / CAT seed 85g
8                        3945                        FetaXa Cheese product 60% 400g (
18                       3960          3491144 LIP.NAP.ICE TEA green yellow 0.5 liter
14                       3953                  2030918 MARIA TRADITIONAL Biscuit 180g
11                       3953                                          197 Onion 1 kg
6                        3945                         TOBUS steering-wheel 0.5kg flow
12                       3953                    * 2108609 SLOB.Mayon.OLIVK.67% 400ml
9                        3945                            TENDER AGE Cottage cheese 10
91                       3945                            TENDER AGE Cottage cheese 10

- Chriss Paul

---

你的方法不错，但无论如何，我必须从整个数组中获取这些根词，以便稍后可以将它们映射到单独的类。因此，如果您有选择根词的机制，那么您的方法将很好地发挥作用。在我们获得根词之后，我们将其粘贴到patt中，然后就可以开始了))) - d-max