/products_search

Primary LanguagePython

Project Strucutre

  • be_app has the backend project
  • fe_app has the frontend project
  • I used mongo Mongodb Atlas because i had no space to install mongodb :D
  • The tests are for the backend on the tests folder

Approach

  • I used inverted index algorithm for searching.
  • I used very simple algorithm for this task:
    1- Mainly I fetch all products in the begging of the server run
    2- For every product I split the description and title to words, you can see the logic for this in split_words
    3- Then I convert each word into singular using inflect library and save it to a dictionary with the object id of the product, so it the dictionary will look like this 
    {
  'black': [ObjectID(123), ObjectID(11)],
  'hat': [ObjectID(11)]
  ...
}
    4- When i get new sentence for search i split it again using split_words then i search for each word in the dictionary.
    5- I merge all of these results and then query the DB with these object IDs

Improvments

This can be improved in several ways, for example I am not doing ranking right now, i just get all results.
also we can convert all nouns with same meaning to abstract word for example: shade and shadow there is alot more into techniques for improving search results.

How To Run

For backend:

python3 -m venv intelistyle_env
source intelistyle_env/bin/activate
cd be_app
pip install -r requirements.txt
DATABASE_ADMIN_PASSWORD=admin_user_123123 flask run --host=0.0.0.0 --port=5010

for test run

DATABASE_ADMIN_PASSWORD=admin_user_123123 pytest -s

For frontend:

cd fe_app
npm install
npm start