/search-app

A search app to look for relevant offers. User can search with Keyword based search and Semantic based search methods.

Primary LanguagePython

Offer Search App

LICENSE PYTHON PYTORCH


📃 Dependencies

Ideal setup requires a Graphic Card


🛠 Installation

  1. Clone repo

    git clone https://github.com/95anantsingh/search-app.git

  2. Create conda environment

    conda env create -f env.yml

  3. Download NLTK Data

    conda activate search
python -m nltk.downloader punkt stopwords

⚡ Run App

cd search-app
conda activate search
streamlit run 🔍_Search.py

Now search for offers with options on the left to see results.

Demo


📘 Report

Project report can be found here or you can find it in the app as well.


📁 Project Structure

Folder Name Description
.streamlit Configuration files for Streamlit
.vscode Visual Studio Code settings and files
core Core application module
├─ base_search.py Base search class
├─ bm25.py BM25 search class
├─ data_processor.py Data processing code
├─ hybrid.py Hybrid search class
├─ init.py Initialization module
├─ neural.py Neural search class
├─ offers_db.py Offers database class
├─ tfidf.py TF-IDF search class
data Data used by the application
├─ processed Processed data files
│ ├─ database.sqlite Offers SQLite database
│ ├─ syn_queries.json Synthetic queries
│ ├─ true_scores.csv True scores (CSV)
│ ├─ true_scores_gold.csv True scores (gold) (CSV)
│ └─ true_scores_syn.csv True scores (synthetic) (CSV)
└─ raw Raw data files
notebooks Jupyter Notebook files
├─ eval.ipynb Evaluation notebook
├─ queries.ipynb Query Generation notebook
└─ search_exp.ipynb Search experiment Notebook
vectors Vector Database files
├─ bm25 BM25 model files
├─ neural Neural model files
│ └─ retrieval FAISS Vector Database Files
└─ tfidf TF-IDF files
pages Application web pages
🔍_Search.py Streamlit App File
env.yml Environment configuration file
README.md Repository README file
requirements.txt Python package requirements

👩🏻‍💻 UML Diagrams

The pacakge core has the main code of this app. UML diagrams are shown below.

1. Core Package

package

2. Classes

classes


📧 Contact

If you have any question, please email anant.singh@nyu.edu