Advanced Transformer Summarization Suite: Abstract & Extractive AI Models 2026
AI Text Summarization with Transformers and Rule-Based NLP 2026
Hybrid book recommender combining content-based TF-IDF with item-item collaborative filtering, weighted by each book's rating evidence to solve cold start. Built on 271k books and 433k ratings from Book-Crossing. Python, scikit-learn, Streamlit.
A photo and IT help chatbot that runs entirely in your browser. No server, no API key.
Search engine for code and documents, built from scratch in Python: inverted index, TF-IDF ranking, trie autocomplete, typo correction and phrase search, with a secure FastAPI backend and an animated web UI.
Content-based movie recommender built with Flask and scikit-learn (TF-IDF, cosine similarity) on the TMDB 5000 dataset, with an offline evaluation. PolyForm Noncommercial license.
AI Healthcare Chatbot desktop application built with Python, Tkinter, Machine Learning, and OpenAI API.
Enterprise Document Q&A Chatbot powered by local RAG and Llama 3.3 (Groq). Fast, private, and free PDF search without external embedding APIs or vector databases.
AI-Powered Bengali Spam Detection Using NLP
AI Document Intelligence Platform — ask questions over your own documents with cited answers. RAG built from scratch in pure Python (TF-IDF retrieval, extractive QA), REST API + web UI in progress.
End-to-end SMS spam classification using TF-IDF and scikit-learn, comparing Naive Bayes and Logistic Regression. Includes real UCI data, evaluation metrics, and a live Streamlit demo.
TF-IDF + Logistic Regression fake/real news classifier (~98% accuracy) and an investigation showing it learns the source's writing style, not truth: the "Reuters" shortcut, top model weights, and custom tests.
Verified AI conference companion — every answer sourced, cited & token-optimised by SigMap. Zero hallucination by construction. Next.js 15 · KendoReact · Gemini 2.5 Flash · Cloud Run.
Sentiment and aspect analysis of Flipkart electronics reviews
Chinese essay scoring research toolkit: TF-IDF/Ridge, native 0–10 scores, CPU inference and Python/CLI/HTTP APIs; weights have separate rights.
Compare lexical vs semantic search on a bundled DeFi/Web3 corpus: deterministic TF-IDF baseline, optional sentence-transformers dense tier, RRF hybrid, and a 16-query benchmark (recall@5 + MRR) — with a Streamlit playground. CPU-only, keyless.
NLP phishing email detector using TF-IDF, Naive Bayes and Logistic Regression (97.35% accuracy)
An AI-powered system designed to analyze candidate resumes against a job description and automatically rank them based on skill relevance, experience, and keyword similarity using NLP & ML.
Recommender system for low-carbon, healthy eating
To associate your repository with the tf-idf topic, visit your repo's landing page and select "manage topics."