Mehyaar commited on
Commit
83c66a3
1 Parent(s): 73ea499

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +98 -1
README.md CHANGED
@@ -25,10 +25,107 @@ The dataset consists of the following columns:
25
  | `Label` | Like count range category |
26
  | `Count` | Number of tweets in the like count range category |
27
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
  ## Usage
29
 
30
  This dataset can be used for various research purposes, including sentiment analysis, trend analysis, and event impact studies related to the Israel-Palestine conflict.
31
  For questions or feedback, please contact:
32
 
33
  - **Name:** Mehyar Mlaweh
34
- - **Email:** mehyarmlaweh0@gmail.com
 
25
  | `Label` | Like count range category |
26
  | `Count` | Number of tweets in the like count range category |
27
 
28
+
29
+
30
+
31
+ ## How to Process the Data
32
+
33
+ To process the dataset, you can use the following Python code. This code reads the CSV file, cleans the tweets, tokenizes and lemmatizes the text, and filters out non-English tweets.
34
+
35
+ ### Required Libraries
36
+
37
+ Make sure you have the following libraries installed:
38
+
39
+ ```bash
40
+ pip install pandas nltk langdetect
41
+ ```
42
+
43
+ ## Data Processing Code
44
+
45
+ Here’s the code to process the tweets:
46
+
47
+ ```python
48
+ import pandas as pd
49
+ import re
50
+ from nltk.tokenize import word_tokenize
51
+ from nltk.corpus import stopwords
52
+ from nltk.stem import WordNetLemmatizer
53
+ from langdetect import detect, LangDetectException
54
+
55
+ # Define the TweetProcessor class
56
+ class TweetProcessor:
57
+ def __init__(self, file_path):
58
+ """
59
+ Initialize the object with the path to the CSV file.
60
+ """
61
+ self.df = pd.read_csv(file_path)
62
+ # Convert 'text' column to string type
63
+ self.df['text'] = self.df['text'].astype(str)
64
+
65
+ def clean_tweet(self, tweet):
66
+ """
67
+ Clean a tweet by removing links, special characters, and extra spaces.
68
+ """
69
+ # Remove links
70
+ tweet = re.sub(r'https\S+', '', tweet, flags=re.MULTILINE)
71
+ # Remove special characters and numbers
72
+ tweet = re.sub(r'\W', ' ', tweet)
73
+ # Replace multiple spaces with a single space
74
+ tweet = re.sub(r'\s+', ' ', tweet)
75
+ # Remove leading and trailing spaces
76
+ tweet = tweet.strip()
77
+ return tweet
78
+
79
+ def tokenize_and_lemmatize(self, tweet):
80
+ """
81
+ Tokenize and lemmatize a tweet by converting to lowercase, removing stopwords, and lemmatizing.
82
+ """
83
+ # Tokenize the text
84
+ tokens = word_tokenize(tweet)
85
+ # Remove punctuation and numbers, and convert to lowercase
86
+ tokens = [word.lower() for word in tokens if word.isalpha()]
87
+ # Remove stopwords
88
+ stop_words = set(stopwords.words('english'))
89
+ tokens = [word for word in tokens if word not in stop_words]
90
+ # Lemmatize the tokens
91
+ lemmatizer = WordNetLemmatizer()
92
+ tokens = [lemmatizer.lemmatize(word) for word in tokens]
93
+ # Join tokens back into a single string
94
+ return ' '.join(tokens)
95
+
96
+ def process_tweets(self):
97
+ """
98
+ Apply cleaning and lemmatization functions to the tweets in the DataFrame.
99
+ """
100
+ def lang(x):
101
+ try:
102
+ return detect(x) == 'en'
103
+ except LangDetectException:
104
+ return False
105
+
106
+ # Filter tweets for English language
107
+ self.df = self.df[self.df['text'].apply(lang)]
108
+
109
+ # Apply cleaning function
110
+ self.df['cleaned_text'] = self.df['text'].apply(self.clean_tweet)
111
+ # Apply tokenization and lemmatization function
112
+ self.df['tokenized_and_lemmatized'] = self.df['cleaned_text'].apply(self.tokenize_and_lemmatize)
113
+
114
+ ```
115
+
116
+ Feel free to add or modify any details according to your specific requirements!
117
+
118
+ Let me know if there’s anything else you’d like to adjust or add!
119
+
120
+
121
+
122
+
123
+
124
+
125
  ## Usage
126
 
127
  This dataset can be used for various research purposes, including sentiment analysis, trend analysis, and event impact studies related to the Israel-Palestine conflict.
128
  For questions or feedback, please contact:
129
 
130
  - **Name:** Mehyar Mlaweh
131
+ - **Email:** mehyarmlaweh0@gmail.com