使用短语中的信息查找句子中单词的索引

Posted 2023-03-12

【中文标题】使用短语中的信息查找句子中单词的索引

【英文标题】：find index of word in sentence with information from phrase

【发布时间】：2021-12-18 14:16:44

【问题描述】：

我需要sentence 中的word 的索引。但有时会出现重复的单词。 phrase 信息会很有帮助。或word 列中的上一行或下一行。

基本上，我只需要识别话语中的单词，例如如果word 是“海边”，我想知道它在句子中是哪个“海边”。我有来自phrase 的额外信息可以帮助识别。它们在数据框中的出现顺序也有帮助。

我现在拥有的是这样的：

file_id	phrase	word	sentence	word_indices
A	I am	I	I am a happy bird. I sing every day. I eat worms.	[0, 5, 9]
B	the seaside is	the	she is by the seaside. The seaside is packed.	[3, 5]
B	the seaside is	seaside	she is by the seaside. The seaside is packed.	[4, 6]
B	the seaside is	is	she is by the seaside. The seaside is packed.	[1, 7]
C	nobody knows	nobody	nobody knows what is going on. She can find nobody	[0, 9]
C	find nobody	nobody	nobody knows what is going on. She can find nobody	[0, 9]
D	it is such a sunny day	sunny	it is such a sunny day ah I am so happy when it's sunny such a sunny day is the best	[4, 13, 16]

但我想得到的是target 列中的内容。

file_id	phrase	word	sentence	word_indices	target
A	I am	I	I am a happy bird. I sing every day. I eat worms.	[0, 5, 9]	[0]
B	the seaside is	the	she is by the seaside. The seaside is packed.	[3, 5]	[5]
B	the seaside is	seaside	she is by the seaside. The seaside is packed.	[4, 6]	[6]
B	the seaside is	is	she is by the seaside. The seaside is packed.	[1, 7]	[7]
C	nobody knows	nobody	nobody knows what is going on. She can find nobody	[0, 9]	[0]
C	find nobody	nobody	nobody knows what is going on. She can find nobody	[0, 9]	[9]
D	it is such a sunny day	sunny	it is such a sunny day ah I am so happy when it's sunny such a sunny day is the best	[4, 13, 16]	[4]

我在这里发现了一个类似的问题：Find index of words in matched text 但不幸的是，这是在 java 中，我需要使用 python 来回答。

非常感谢！

【问题讨论】：

你能给出更准确的定义吗？我假设，如果word 在句子中不是唯一的，算法将查找phrase 术语并返回该短语第一次出现的单词的索引，对吗？如果phrase 出现多次会怎样？如果word多次出现但phrase没有出现怎么办？

感谢您的评论。是的，你问的问题也是我的问题。基本上，我只需要识别话语中的单词，例如如果word 是“海边”，我想知道它在句子中是哪个“海边”。我有来自phrase 的额外信息可以帮助识别。它们在数据框中出现的顺序也有帮助。

【参考方案1】：

我会将其分为两个步骤。找出导致句子中短语的单词数，然后找到短语中单词的单词索引号：如下所示：

def get_index_of_word_in_sentence(word, phrase, sentence):
    index1 = sentence.index(phrase)
    word_num1 = len(sentence[:index1].split())
    word_num2 = phrase.split().index(word)
    return word_num1 + word_num2

df["target"] = df.apply(lambda x: get_index_of_word_in_sentence(x["word"], x["phrase"], x["sentence"]), axis=1)

【讨论】：

感谢您的回答！看来这是在计算字符串的索引？我得到以下结果：[0, 23, 27, 35, 0, 44, 13]。这似乎不是我需要的......

啊啊是的，我误读了这个问题。我已经修改了我的答案。

成功了！谢谢！

@SuperDuperMario，你能告诉我你是怎么得到这个词的索引的吗？

@DanielWyatt 如果我没有它们，我该如何修改你的函数来获取单词索引（基本上，我需要填写 word_indices 列）？谢谢！！