Spaces:
Running
Running
File size: 4,327 Bytes
8da79d7 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 |
import streamlit as st
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
import re
import pandas as pd
import googleapiclient.discovery
import plotly.express as px
# Load the BERT tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
# Set up the YouTube API service
api_service_name = "youtube"
api_version = "v3"
DEVELOPER_KEY = "AIzaSyC4Vx8G6nm3Ow9xq7NluTuCCJ1d_5w4YPE" # Replace with your actual API key
youtube = googleapiclient.discovery.build(api_service_name, api_version, developerKey=DEVELOPER_KEY)
# Function to fetch comments for a video ID
def scrape_comments(video_id):
request = youtube.commentThreads().list(
part="snippet",
videoId=video_id,
maxResults=100
)
response = request.execute()
comments = []
for item in response['items']:
comment = item['snippet']['topLevelComment']['snippet']
comments.append([
comment['textDisplay']
])
comments_df = pd.DataFrame(comments, columns=['comment'])
# df.head(10).
return comments_df
# Function to extract video ID from YouTube URL
def extract_video_id(video_url):
match = re.search(r'(?<=v=)[\w-]+', video_url)
if match:
return match.group(0)
else:
st.error("Invalid YouTube video URL")
# Function to fetch YouTube comments for a video ID
def fetch_comments(video_id):
# Example using youtube-comment-scraper-python library
comments = scrape_comments(video_id)
return comments
# Function to analyze sentiment for a single comment
def analyze_sentiment(comment):
tokens = tokenizer.encode(comment, return_tensors="pt", max_length=512, truncation=True)
# input_ids = tokens['input_ids']
# attention_mask = tokens['attention_mask']
# result = model(input_ids, attention_mask=attention_mask)
result = model(tokens)
sentiment_id = torch.argmax(result.logits) + 1
if(sentiment_id > 3):
sentiment_label = "Positive"
elif(sentiment_id < 3):
sentiment_label = "Negative"
else:
sentiment_label = "Neutral"
return sentiment_label
def main():
st.title("YouTube Comments Sentiment Analysis")
st.write("Enter a YouTube video link below:")
video_url = st.text_input("YouTube Video URL:")
if st.button("Extract Comments and Analyze"):
video_id = extract_video_id(video_url)
if video_id:
comments_df = fetch_comments(video_id)
# Comments is a dataframe of just the comments text
# st.write("Top 100 Comments extracted\n", comments_df)
comments_df['sentiment'] = comments_df['comment'].apply(lambda x: analyze_sentiment(x[:512]))
sentiment_counts = comments_df['sentiment'].value_counts()
positive_count = comments_df['sentiment'].value_counts().get('Positive', 0)
negative_count = comments_df['sentiment'].value_counts().get('Negative', 0)
neutral_count = comments_df['sentiment'].value_counts().get('Neutral', 0)
# Create pie chart in col2 with custom colors
fig_pie = px.pie(values=[positive_count, negative_count, neutral_count],
names=['Positive', 'Negative', 'Neutral'],
title='Pie chart representations',
color=sentiment_counts.index, # Use sentiment categories as colors
color_discrete_map={'Positive': 'green', 'Negative': 'red', 'Neutral': 'blue'})
st.plotly_chart(fig_pie, use_container_width=True)
# Create bar chart below the pie chart with custom colors
fig_bar = px.bar(x=sentiment_counts.index, y=sentiment_counts.values,
labels={'x': 'Sentiment', 'y': 'Count'},
title='Bar plot representations',
color=sentiment_counts.index, # Use sentiment categories as colors
color_discrete_map={'Positive': 'green', 'Negative': 'red', 'Neutral': 'blue'})
st.plotly_chart(fig_bar)
if __name__ == "__main__":
main()
|