📊 Conversight - Chat Analyzer

Conversight is a Python-based chat analysis tool to extract insights from WhatsApp, Facebook Messenger, and Telegram conversations.
It includes cleaning, statistics generation, personality summarization, sentiment analysis, TF-IDF, LDA topic modeling, and fun comparisons!

Live Demo

🚀 Features

1. Cross-Platform Support:

   • WhatsApp: Import and analyze exported .txt files from WhatsApp.
   • Telegram: Import and analyze exported .html files from Telegram.
   • Facebook: Import and analyze exported .html files from Facebook.

2. Chat Summary:

   • Get an overall chat recap, including:
     • Most Active User 🗣️
     • Chattiest Day 📅
     • Prime Talk Hour 🕒
     • Average Words per Message 📏
   • Streak Feature: View your longest uninterrupted streak of chat activity.
   • Throwback Feature: Discover the first-ever message in your chat history.

3. User Comparison:

   • Chat Duel: Compare two or more users' activity.
   • Metrics include messages, words, media, emojis, sentiment, and more.

4. Sentiment Analysis:

   • Mood Map: Visualize the overall sentiment of your messages.
   • Mood Over Time: Track mood changes throughout the chat timeline.
   • Sentiment scores and vibes are automatically assigned to each message.

5. Deep Talk Dive (NLP Analysis):

   • TF-IDF Keywords: Find the most important keywords in the chat.
   • Topic Clustering (LDA): Discover hidden topics through Latent Dirichlet Allocation (LDA).

6. User Activity:

   • Most Active and Least Active Users: View daily and monthly activity heatmaps.
   • Weekday and Monthly Activity: Track user engagement across different timeframes.

7. Words & Emojis Showdown:

   • Word Cloud: Visualize the most commonly used words in the chat.
   • Top Emojis Used: View the most frequently used emojis with a frequency breakdown.
   • Emoji Pie Chart: A pie chart displaying the distribution of emojis used.

8. Export Data:

   • Download CSV and Excel Reports: Export detailed chat and emoji analysis for further inspection.
   • Full Report: Download a comprehensive chat analysis with all metrics, sentiment, word analysis, and emoji usage.

9. Fun Facts & Interactive Elements:

   • Chat Fun Facts: Enjoy random fun facts about texting and chat behavior.
   • Engagement Reactions: Get quirky reactions from Conversight based on your data.

---

🎯 How to Use

1. Export your chat from WhatsApp (TXT), Telegram (HTML), or Facebook (HTML).
2. Pick Your Platform from the sidebar.
3. Upload Your Chat File.
4. Explore various features such as user comparison, sentiment analysis, emoji showdown, and more!

---

🛠 Installation

1. Clone the repository:

   git clone https://github.com/supriya811106/Multi-Platform-Chat-Analyzer.git
   cd Multi-Platform-Chat-Analyzer

2. Install dependencies:

   pip install -r requirements.txt

   Required libraries:

• streamlit: For creating the web interface of the chat analyzer.
• pandas: For data manipulation and analysis.
• matplotlib: For plotting data visualizations.
• seaborn: For enhanced data visualizations.
• plotly: For interactive visualizations.
• textblob: For performing sentiment analysis.
• wordcloud: For generating word clouds.
• numpy: For numerical operations.
• scikit-learn: For machine learning tools like CountVectorizer and LatentDirichletAllocation.
• urlextract: For extracting URLs from chat messages.
• emoji: For processing emoji data in messages.
• beautifulsoup4: For parsing HTML data, used for Telegram chat parsing.
• datetime: For handling date and time data.

---

📈 Output Examples

• Personality Summaries like:
  • 🎉 Emoji Queen/King
  • 🧠 Philosopher
  • 📸 Storyteller
• Fun Comments like:
  • 🧨 You’re breaking records with that message count!
  • 🤣 Emoji addict detected.
• Sentiment Mood like:
  • 😄 Lots of good vibes here!
  • 😬 More negative vibes than good ones!

---

📝 Notes

• Platform-specific cleaning is supported: WhatsApp, Facebook Messenger, and Telegram.
• Can work on both individual and group chats.
• Optimized for conversational datasets with username, message, date, day columns.

---