SummarIQ

AI-Powered PDF Summarizer Application for Scientific Papers

Comment

🚀 SummarIQ – AI-Powered Scientific Paper Summarizer

Extract • Understand • Summarize • Visualize (PDF + LaTeX) SummarIQ is an AI-powered system designed to summarize scientific research papers written in PDF or LaTeX. It extracts sections, equations, metadata, and produces structured summaries using transformer-based models like T5-small and BART-large.

This project includes:

  • AI summarization models (T5, BART)
  • LaTeX equation extraction & rendering
  • FastAPI backend
  • Gradio UI
  • Docker deployment
  • Prometheus + Grafana monitoring
  • iOS UI prototypes
  • Remote GPU inference (Kaggle)

System Demo Video (Google Drive)

Since GitHub does not allow uploading large .mov files, the demo video is hosted on Google Drive: 🔗 Demo Video: https://drive.google.com/your-video-link

Project Structure

├── App design
│   ├── Purple Pink Gradient Login Page Mobile Prototype (3)
│   │   ├── 1.jpg
│   │   ├── 2.jpg
│   │   └── 3.jpg
│   ├── Screen_1.png
│   ├── Screen_2.png
│   └── Screen_3.png
├── deployment
│   ├── docker-compose.yml
│   ├── Dockerfile
│   └── gradio_app.py
├── documentation
│   ├── plot_images
│   │   ├── Architechture_diagram.png
│   │   ├── fairness_explainability.png
│   │   ├── githubrepo.png
│   │   ├── sample_equation_output.png
│   │   ├── sample_summary_output.png
│   │   └── shap_summary.png
│   ├── Poster.pdf
│   ├── Project_report.pdf
│   └── SummarIQ_project_template.pdf
├── monitoring
│   ├── grafana
│   │   └── summariq_graphana_dashboard.json
│   ├── metrices_report
│   │   ├── metrics_report_valset.csv
│   │   ├── monitoring_images
│   │   │   ├── cpu_usage.png
│   │   │   ├── Fairlear_matrices.png
│   │   │   ├── feedback_latest_rating.png
│   │   │   ├── feedback_no_comments_total.png
│   │   │   ├── feedback_rating.png
│   │   │   ├── feedback_total.png
│   │   │   ├── feedback_with_comments_total.png
│   │   │   ├── localhost_metrices_log.png
│   │   │   ├── request_count.png
│   │   │   ├── request_latency_seconds.png
│   │   │   └── shap_values.png
│   │   └── structured_summary.csv
│   └── prometheus
│       ├── alert_rules.yml
│       └── prometheus.yml
├── Notebooks
│   ├── bart_train.ipynb
│   ├── kaggle_server_run.ipynb
│   ├── risk_management.ipynb
│   ├── t5_model_test.ipynb
│   ├── test_summarize.ipynb
│   ├── test_t5_small.ipynb
│   ├── training.ipynb
│   └── valset_test.ipynb
├── README.md
├── requirements_projects.txt
├── requirements.txt
├── src
    ├── __pycache__
    │   └── app.cpython-310.pyc
    ├── app.py
    ├── data
    │   ├── latex_extracted.json
    │   └── Test_pdf
    │       ├── 2404.08534v2.pdf
    │       ├── sm.pdf
    │       ├── sm.tex
    │       ├── Tower_V3.tex
    │       └── tower.pdf
    ├── equation_renderer
    │   ├── __pycache__
    │   │   ├── app.cpython-310.pyc
    │   │   ├── eqapp.cpython-310.pyc
    │   │   ├── quicklatex_app.cpython-310.pyc
    │   │   ├── quicklatex_renderer.cpython-310.pyc
    │   │   └── renderer.cpython-310.pyc
    │   ├── eqapp.py
    │   └── renderer.py
    ├── extractor
    │   ├── __pycache__
    │   │   └── latex_extractor.cpython-310.pyc
    │   ├── extract_usingcv.py
    │   └── latex_extractor.py
    ├── models
    │   ├── __pycache__
    │   │   └── summarizer_remote.cpython-310.pyc
    │   └── summarizer_remote.py
    ├── summarizer_kaggle.py
    ├── summarizer_local.py
    └── utils
        ├── load_data.py
        └── response_server
            ├── api_response.json
            └── response_without_T5server.json

Features

AI Summarization

  • T5-small (fast, lightweight)
  • BART-large (high accuracy)
  • Section-wise summarization
  • Handles long scientific text

LaTeX Equation Handling

  • Detects inline & block equations
  • Renders equations as images
  • Ranks important equations

FastAPI Backend

  • /summarize-latex endpoint
  • PDF/LaTeX processing
  • JSON structured output

Gradio UI

  • Upload interface
  • Real-time response
  • Feedback collection

Deployment

  • Dockerfile + docker-compose
  • Separate services for API, Gradio, Prometheus
  • Works on local + cloud environments

Monitoring

  • Prometheus for metric scraping
  • Grafana dashboards
  • Custom metric: summariq_feedback_total

iOS Mobile App Prototype

  • Screens included in App design/
  • Designed for future live deployment

Installation

1. Clone the repository

git clone https://github.com/yourusername/SummarIQ.git
cd SummarIQ

2.Create a virtual environment

python3 -m venv venv
source venv/bin/activate

3. Install dependencies

This project uses two separate requirements files:

  • #### requirements_projects.txt(local development) pip install -r requirements_projects.txt
  • #### requirements.txt(Docker/production) pip install -r requirements.txt ## Run with Docker
  • #### Build containers docker-compose build
  • #### Start services docker-compose up This starts:
  • FastAPI → http://localhost:8000
  • Gradio UI → http://localhost:7860
  • Prometheus → http://localhost:9090
  • Grafana → http://localhost:3000

Run Locally (without Docker)

Start FastAPI server

uvicorn app:app --reload --port 8000

Start Gradio UI

python gradio_app.py

API Usage

Endpoint

POST /summarize-latex

Example (cURL)

curl -X POST -F "file=@sample.tex" http://localhost:8000/summarize-latex

Output Example

utils/response_server/api_response.json

Model Pipeline

  • Receive LaTeX/PDF
  • Extract sections & equations
  • Preprocess scientific text
  • Summarize using T5/BART
  • Produce structured JSON output
  • Render LaTeX equations
  • Display summary in Gradio / iOS UI

Monitoring (Prometheus + Grafana)

Metrics collected:

  • API latency (p50, p90, p95)
  • Total requests
  • Feedback count summariq_feedback_total
  • CPU/RAM usage
  • Docker container performance
  • Error rates Dashboards:
  • Grafana → http://localhost:3000
  • Prometheus → http://localhost:9090

iOS App Prototype

Located in App design/: Includes:

  • Login screen
  • Upload screen
  • Processing screen
  • Summary UI design Powered by SwiftUI (planned for Phase 2 deployment).

Roadmap

Phase 2

  • Fine-tuned summarization (domain-specific)
  • Full iOS app integration
  • Deploy backend to AWS/GCP
  • Real-time feedback storage in DB (MongoDB/PostgreSQL)
  • Advanced Grafana dashboards
  • Semantic equation understanding
  • User accounts + authentication

Built With

  • dockerfile
  • jupyter-notebook
  • python
  • tex
Share this project:

Updates