Flat Finder Doc

Overview

• Project Name: Flat Finder
• Project Description: Tool for finding shared apartments using user preferences and shared-apartments descriptions.
• Data Source: Scraping wg-gesucht using Selenium.
• Approach: Unfiltered scraping, temporary storage in a JSON file, structural and deep filtering, recommendation generation.
• Tools: Selenium, MongoDB, GPT-3.5, Telegram.

Components

• Selenium scrapper
• mongoDB
• Telegram Bot
• OpenAI Agent

Use-Cases:

• Profile creation and update.
• Preference setting.
• WG recommendations (active and passive).

Workflow

• Creating User: User will be created in UI and preferences will be set.
• Scraping: Collecting data from wg-gesucht using selenium.
• Storage: Temporary storage in a JSON file.
• Filtering: Apartments data will be filtered for the user.
• Recommendation: Deep filtering with non-structural data, generating recommendation for user and an example application text.
• UI notification: Notifying the user of the recommendation and example application.

Experimentation

---

Setup

1. Download Telegram on your Smart phone (IOS) or Desktop (windows)
2. Create an account in Telegram.
3. setup your env. file to the flat finder Folder:
4. download the Google Chrome browser.
5. Install the requirements:

   pip install -r requirements.txt

6. Run the Agent:

      python main.py

7. Run the Telegram Bot:

     python src/FE/app.py

8. Start the conversation with the FlatFinderHTWBot:
   1. Start a Chat with the bot using this link.
   2. Klick on the start button or send /start.
   3. send /profile to set up your profile and preferences.

congrats you are ready to go!

---

Testing:

1. uncomment the Line 384 in FE/app.py.

   # response = agent.get_response(user_id, message)
   ```:

2. use this example user Data for creating your profile

  {
    "average_monthly_net_income": 2000,
    "date_of_birth": "2000-01-01",
    "email": "[email protected]",
    "employment_type": "Student",
    "full_name": "John Doe",
    "phone_number": "1245",
    "smoker": false
    "additional_info": ["balcony", "peaceful"],
    "address": {
      "street": "Seidelbastweg",
      "house_number": 1,
      "zip_code": 12357,
      "city": "Rudow",
      "country": "Germany"
    },
    "apartment_preferences": {
      "max_rent": 600,
      "location": "Berlin",
      "bezirk": ["Mitte", "Friedrichshain", "Rummelsburg", "Wedding", "Kreuzberg"],
      "min_size": 20,
      "ready_to_move_in": "2024-06-01",
      "preferred_roommates_sex": "gender_irrelevant",
      "preferred_roommate_age": [20, 30],
      "preferred_roommate_num": 3,
      "smoking_ok": false
    }
  }

3. this will make sure that you receive a static listing stored in the output.json file.