时间戳

首页 > 解决方案 > Python NLTK：如何找到用户输入和 Excel 数据之间的相似性

python - Python NLTK：如何找到用户输入和 Excel 数据之间的相似性

问题描述

所以我正在尝试创建一个 python 聊天机器人，我有一个包含数百行的 excel 文件，如下所示：

QuestionID     Question               Answer        Document 
Q1             Where is London?       In the UK     Google
Q2             How many football      22            Google
               players on the pitch?

现在，当用户输入一个问题时，例如“伦敦在哪里？” 或“伦敦在哪里”我希望它返回该行中的所有文本。

我可以成功打印 excel 文件中的内容，但我不确定如何遍历所有行并找到与用户问题相似或匹配的行。

text = []
    
with open("dataset.csv") as csvfile:
    reader = csv.DictReader(csvfile)
    for row in reader:
        text.append((row['Question'], row['Answer'], row['Document'] ))
    
print(text)

标签： pythonnltk

解决方案

您可能不想进行完全匹配，因为这意味着它将区分大小写，并且需要精确的标点符号，没有拼写错误等。

我会考虑使用fuzzywuzzy来查找匹配分数。然后您可以返回与问题最匹配的解决方案：

例子：

from fuzzywuzzy import fuzz
import pandas as pd

lookup_table = pd.DataFrame({
        'QuestionID':['Q1','Q2','Q3'],
        'Question':['Where is London?','Where is Rome?', 'How many football players on the pitch?'],
        'Answer':['In the UK','In Italy', 22],
        'Document':['Google','Google','Google']})



question = 'how many players on a football pitch?'

lookup_table['score'] = lookup_table.apply(lambda x : fuzz.ratio(x.Question, question), axis=1)
lookup_table = lookup_table.sort_values('score', ascending=False)

结果表：

print (lookup_table.to_string())

  QuestionID                                 Question     Answer Document  score
2         Q3  How many football players on the pitch?         22   Google     71
0         Q1                         Where is London?  In the UK   Google     34
1         Q2                           Where is Rome?   In Italy   Google     27

给出最佳选择的答案：

print (lookup_table.iloc[0]['Answer'])
22

或者因为您想要该行

print (lookup_table.head(1))
  QuestionID                                 Question Answer Document  score
2         Q3  How many football players on the pitch?     22   Google     71

推荐阅读