CoeuSearch

---

In Greek mythology, Coeus was the Titan-god of the inquisitive mind, his name meaning "query" or "questioning"

Neural File Search Engine

CoeuSearch is an NLP based intelligent local-file search engine that searches for relevant text documents in a specific folder, considering the semantics of the file’s name & it's content and returns the most relevant files.

Input

• Search Directory: Location of folder to be searched
• Search Query: Phrase/keywords to be searched

Output

• Relevant Files: Location of top matched files

Files Supported

• File Content: .docx, .txt, .pdf, .ppt, .csv
• File Name: ALL excluding images, audio and video files

System Design

Getting Started

Download or clone this repository on your system.

Prerequisites

- PYTHON >= 3.8.1
- DJANGO >= 1.1.2

Installing

• Install python3 on your system
• Add python to environment variables
• Navigate to 'CoeuSearch' folder in terminal
• Create an evironment using the following command -> python3 -m venv venv
• Activate the evironment using the following command

     .\venv\Scripts\activate    (For Windows)
      source venv\bin\activate  (For Ubuntu)
      source venv/bin/activate  (For MacOS) 

• Run command pip3 install -r requirements.txt
• Run command python3 download_nltk_data.py
• Install openssl ( for macOS run brew install openssl)
• Change cache_base_path in configs.py according to your machine path
• Run command python3 manage.py runserver
• Click on localhost link generated after execution of previous command

Future Work

• Extend this tool to a multi-modal search engine that supports image, audio and video data by utilising Vision Transformers and Automatic Speech Recognition.
• Enable multilingual search queries by using pre-trained language models like mBERT and XLM
• Improve inference speed by using for semantic similarity

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.