匹配相似的字符/单词

Question

我有以下包含列X和Y的数据框架，

    X                                   Y
1   SAN DIEGO                           FOND DU LAC
2   THE RIO GRANDE                      RIO GRANDE
3   RIO GRANDE                          RIO GRANDE
4   WEST TENNESSEE                      TENNESSEE
5   EP De SAN JOAQUIN                   De SAN JOAQUIN
6   SOUTHERN VIRGINIA                   VIRGINIA
7   SOUTHERN VIRGINIA                   SOUTHWESTERN VIRGINIA
8   EN COLOMBIA                         COLOMBIA
9   THE EP De NORTHERN CALIFORNIA       De NORTHERN CALIFORNIA
10  FLORIDA                             NEW JERSY

我想要得到不匹配的行，1和10。行2-9是匹配的或接近匹配的，没有问题。我期望的数据帧是

    X                                   Y
1   SAN DIEGO                           FOND DU LAC
10  FLORIDA                             NEW JERSY

Answer 1

在R中，我们将字符串按每列中的空格拆分，检查单词之间是否有intersect，找到list的lengths，并将长度为0的数据集设置为子集

df1[!lengths(Map(intersect, strsplit(df1$X, "\\s+"), strsplit(df1$Y, "\\s+"))),]
#          X           Y
#1  SAN DIEGO FOND DU LAC
#10   FLORIDA   NEW JERSY

我们也可以循环遍历列，而不是按每列拆分，而是执行split

df1[!lengths(do.call(Map, c(intersect, unname(lapply(df1, strsplit, split="\\s+"))))),]
#      X           Y
#1  SAN DIEGO FOND DU LAC
#10   FLORIDA   NEW JERSY

或者另一种选择是stringdist

library(stringdist)
i1 <- with(df1, stringdist(X, Y, method = "qgram"))
df1[i1 %in% tail(sort(i1), 2),]
#          X           Y
#1  SAN DIEGO FOND DU LAC
#10   FLORIDA   NEW JERSY