从聊天对话中提取相关文本

Question

我有下面的聊天对话：

聊天开始:2018年4月2日，星期一，10:23:30 (+0100)聊天来源: GB -我的帐户(登录)代理Navin P(1) Navin:谢谢您联系XYZ，您正在与Navin通话。我能帮上什么忙吗？(34 S)访问者:你好，我刚刚从华晨宽带转向光纤( 39s ) Navin:嗨(42 S) Navin:早上好。(47)参观者:我发现这里的宽带太慢了(52 S)，纳文:今天我能帮你什么吗？( 1m 1秒)访问者:我们被告知有一天可以换掉(1米5秒)纳文:我会帮你的。( 1m 11)参观者:你能告诉我什么时候会发生这种事吗？

我只想使用R.从上面提取相关的文本，我本质上只希望访问者注释出现在结果中。

我想要的结果如下：

我发现宽带太慢了，我们被告知有一天会被调换，你能告诉我这会在什么时候发生吗？

我试着用gsub和strsplit来完成这个任务，但是没有结果。感谢这里的投入。

mytext <- paste(c("Agent Navin P ( 1s ) Navin: Thanks for contacting XYZ, you are talking to Navin. How can I help? ( 34s ) Visitor:", 
    "Hello , I?ve just currently switched from brillian broadband to fab fibre ( 39s ) Navin: Hi ( 42s ) Navin: Good morning. ( 47s )", 
    "Visitor: I find the brilliant broadband so slow ( 52s ) Navin: How can i help you today? ( 1m 1s ) Visitor: And we got told would be", 
    "a day to get switched over ( 1m 5s ) Navin: I'll help you with it. ( 1m 11s ) Visitor: Can you tell me when this will happen by"
), collapse = ' ')

Answer 1

我相信有很多方法可以做到这一点。我和来访者分手用潜艇移除纳文的答案。对于替换，我们需要在末尾添加[-1]，因为我们不需要在我们拆分的第一个“访问者”之前添加任何内容。

str <- "Chat Started: Monday, April 02, 2018, 10:23:30 (+0100) Chat Origin: GB - My Account (Signed In) Agent Navin P ( 1s ) Navin: Thanks for contacting XYZ, you are talking to Navin. How can I help? ( 34s ) Visitor: Hello , I?ve just currently switched from brillian broadband to fab fibre ( 39s ) Navin: Hi ( 42s ) Navin: Good morning. ( 47s ) Visitor: I find the brilliant broadband so slow ( 52s ) Navin: How can i help you today? ( 1m 1s ) Visitor: And we got told would be a day to get switched over ( 1m 5s ) Navin: I'll help you with it. ( 1m 11s ) Visitor: Can you tell me when this will happen by"
str <- strsplit(str," Visitor: ")[[1]]
sub(" \\((.*?)\\) Navin:.*","",str)[-1]

# [1] "Hello , I?ve just currently switched from brillian broadband to fab fibre"
# [2] "I find the brilliant broadband so slow"                                   
# [3] "And we got told would be a day to get switched over"                      
# [4] "Can you tell me when this will happen by"

如果你想要像你的例子那样的一行，你可以使用paste

paste(sub(" \\((.*?)\\) Navin:.*","",str)[-1],collapse = " ")
# [1] "Hello , I?ve just currently switched from brillian broadband to fab fibre I find the brilliant broadband so slow And we got told would be a day to get switched over Can you tell me when this will happen by"

如果这个人的名字不是一致的"Navin"，您可以使用\\w+来匹配sub查询中的任何人的名字来删除它。

sub(" \\((.*?)\\) \\w+:.*","",str)[-1]

Answer 2

保留更多信息的一个可能的解决办法是：

mytext <- paste(c("Agent Navin P ( 1s ) Navin: Thanks for contacting XYZ, you are talking to Navin. How can I help? ( 34s ) Visitor:", 
    "Hello , I?ve just currently switched from brillian broadband to fab fibre ( 39s ) Navin: Hi ( 42s ) Navin: Good morning. ( 47s )", 
    "Visitor: I find the brilliant broadband so slow ( 52s ) Navin: How can i help you today? ( 1m 1s ) Visitor: And we got told would be", 
    "a day to get switched over ( 1m 5s ) Navin: I'll help you with it. ( 1m 11s ) Visitor: Can you tell me when this will happen by"
), collapse = ' ')


library(dplyr); library(textshape); library(stringi)

mytext %>%
    stri_replace_all_regex('(\\( [0-9ms ]+ \\))(\\s+)', '$1<<splithere>>') %>%
    stri_split_fixed('<<splithere>>') %>%
    lapply(function(x) {
        x %>%
            split_transcript() %>%
            mutate(dialogue = ifelse(!grepl('\\(\\s*([0-9ms ]+)\\s\\)', dialogue), paste(dialogue, '( - )'), dialogue)) %>%
            extract(dialogue, c('dialogue', 'timestamp'), '(^.+)\\s\\(\\s*([0-9ms -]+)\\s\\)')
    })

## [[1]]
##                  person                                                                  dialogue timestamp
## 1  Agent Navin P ( 1s )                                                             Agent Navin P        1s
## 2                 Navin      Thanks for contacting XYZ, you are talking to Navin. How can I help?       34s
## 3               Visitor Hello , I?ve just currently switched from brillian broadband to fab fibre       39s
## 4                 Navin                                                                        Hi       42s
## 5                 Navin                                                             Good morning.       47s
## 6               Visitor                                    I find the brilliant broadband so slow       52s
## 7                 Navin                                                 How can i help you today?     1m 1s
## 8               Visitor                       And we got told would be a day to get switched over     1m 5s
## 9                 Navin                                                    I'll help you with it.    1m 11s
## 10              Visitor                                  Can you tell me when this will happen by         -

然后你就可以按人过滤等等。