gwti

Gram-Weighted Tracing for Interpretability...

https://github.com/juancalderon/gwti

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • CITATION.cff file
    Found CITATION.cff file
  • codemeta.json file
    Found codemeta.json file
  • .zenodo.json file
    Found .zenodo.json file
  • DOI references
  • Academic publication links
  • Academic email domains
  • Institutional organization owner
  • JOSS paper metadata
  • Scientific vocabulary similarity
    Low similarity (8.0%) to scientific vocabulary
Last synced: 6 months ago · JSON representation ·

Repository

Gram-Weighted Tracing for Interpretability...

Basic Info
  • Host: GitHub
  • Owner: JuanCalderon
  • Language: Jupyter Notebook
  • Default Branch: main
  • Size: 3.08 MB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created 10 months ago · Last pushed 8 months ago
Metadata Files
Readme Citation

README.md

GWTI: Gram-Weighted Tracing for Interpretability

This repository contains the full implementation of GWTI (Gram-Weighted Tracing for Interpretability), a lightweight and pre-hoc interpretability framework designed for sparse short-text NLP tasks such as tweet classification and user profiling.

🌟 Overview

GWTI combines: - q-gram tokenization - TF-IDF vectorization - Linear classifiers - A tracing mechanism to recover token-level contributions for visualization.

It provides precise, low-cost visual explanations that are particularly effective in high-dimensional, sparse input spaces.

📁 Repository Structure

  • experiments/ — Python scripts for reproducing the experiments and figures reported in the PRL manuscript.
  • supplementary/ — Supplementary materials heatmaps, token-level visualizations, and UI snapshots used in the paper
  • CITATION.cff — Citation metadata for properly referencing this work.
  • README.md — This file.

📖 Publication

This code accompanies the manuscript:

“GWTI: A Pre-Hoc Framework for Visual Interpretability in Sparse Short-Text NLP Tasks”, submitted to Pattern Recognition Letters (PRL).

🔗 Citation

To cite this work, please see the CITATION.cff file or use the following BibTeX (coming soon).

📬 Contact

For questions or collaboration inquiries, please contact:

José J. Calderón
Email: jose.calderon@cimav.edu.mx
ORCID: 0000-0002-XXXX-XXXX


© 2025 José J. Calderón, Mario Graff, Eric S. Téllez. Licensed under MIT.

Owner

  • Name: Juan Calderon
  • Login: JuanCalderon
  • Kind: user

Citation (CITATION.cff)

cff-version: 1.2.0
title: "GWTI: A Pre-Hoc Framework for Visual Interpretability in Sparse Short-Text NLP Tasks"
authors:
  - family-names: Calderón
    given-names: José J.
  - family-names: Graff
    given-names: Mario
  - family-names: Téllez
    given-names: Eric S.
date-released: 2025-06-01
doi: 10.1234/gwti.code

GitHub Events

Total
  • Push event: 7
  • Create event: 2
Last Year
  • Push event: 7
  • Create event: 2