similix 0.2

No project description provided

Similix

System Similarity, a cutting-edge solution for identifying article similarities with speed, security, and 85% accuracy. The specialized Article Recommendation submodule leverages OpenAI modules, numpy, and pandas for vector generation, data cleaning, and similarity calculations.

Features

• Fast & Efficient: Quick results without compromising accuracy.
• Secure: Robust security measures for data protection.
• High Accuracy: 99% accuracy in article recommendations.

Overview

Focused on suggesting articles similar to a target article, the submodule optimizes vector generation, data cleaning, and similarity calculations.

Workflow

Vector Generation:

• OpenAI module generates a vector for the target article. Cached for future optimization.

• Data Cleaning: Numpy and pandas clean CSV data, including titles and descriptions.

• Vectorization: Converts article descriptions into numerical vectors.

• Similarity Calculation: System computes similarity between target and other article vectors.

• Recommendation: Recommends articles with the highest similarity.

Environment Variables

To run this project, you will need to add the following environment variables to your .env file

OPENAI_API_KEY Just for testing purpose you can contact me or you can get it from oepnai which is you have to paid for it.

Installation

• Make sure in your dataset contain title and description as the fields requirement

We assume that you have Python installed and have set up the entire environment

  pip install similix

Usage

   - Make sure your dataset contain 'title' and 'description' as the fields requirement 

from similix.article_recommender import ArticleRecommender
import os

os.environ['OPENAI_API_KEY'] = 'OPENAI_API_KEY'

ac = ArticleRecommender(dataset_path='cleaned_dataset.csv', embedding_model='text-embedding-3-large')

limit_articles = 2
article_target = 'Twenty six women and five children were murdered by current'

print(ac.recommend_articles(article_target, limitArticles))