Customer Message Categorization and Analysis

This project processes and analyzes customer messages from CSV files, categorizing them into predefined topics using LDA (Latent Dirichlet Allocation) and keyword-based matching. It also generates visualizations to summarize the results.

Features

• Data Preprocessing:

  • Combines multiple CSV files containing customer messages (from the data/ folder).
  • Filters messages from customers and tokenizes text using both English and Chinese tokenizers (nltk and jieba).
  • Removes stopwords and non-essential characters.

• Topic Modeling:

  • Uses LDA to identify latent topics in the messages.
  • Matches messages to predefined categories using keyword-based scoring.

• Categorization:

  • Maps messages to one of the following categories:
    • 訂房 (Booking)
    • 優惠 (Promotion)
    • 訂餐 (Food Ordering)
    • 設施 (Facility)
    • 行程 (Itinerary)
    • 服務 (Service)
    • 附近 (Nearby)
    • 抱怨 (Complaint)
    • 其他 (Other)

• Visualization:

  • Generates pie charts for:
    • Distribution of customer issue categories.
    • Platform comparison based on unique customer counts.
  • Plots daily unique customer counts over time.

• Export:

  • Saves categorized messages to a CSV file (categorized_customer_messages.csv).

Usage

1. Prepare Data:

   • Place your CSV files in the data/ folder. Do not include real or sensitive company/customer data in the repository.
   • Each CSV should have the following columns:
     • Platform, Source, Customer ID, Message Content, Send TIme
   • You may use the provided example_data.csv as a template (with dummy data).

2. Install Requirements:

   • Install Python 3.7+ and the required packages:

     pip install pandas matplotlib nltk jieba gensim

3. Run the Script:

   • Execute the script:

     python Categorization.py

4. View Results:

   • Visualizations will be displayed and saved as PNG files.
   • The categorized results will be saved as categorized_customer_messages.csv.

Data Privacy

• No real company/customer data is included in this repository.
• The data/ folder is excluded via .gitignore to prevent accidental upload of sensitive files.
• Only use example or dummy data for demonstration and sharing.

Example Data

You can create a file like data/example_data.csv:

Platform,Source,Customer ID,Message Content,Send TIme
LINE,customer,12345,"I want to book a room",2025-01-01 12:00:00
WhatsApp,customer,67890,"Do you have any discounts?",2025-01-02 13:45:00

License

License Apache 2.0