news-tracking 1.0.2.202209211638

Programs used to manage experiments with the document_tracking infrastructure.

Track news stories from news articles

This project is the front-end of the document_tracking package, which proposes algorithms to track news documents (long articles, telegraphic dispatches, etc.) into news stories (group of documents reporting similar events). It manipulates data following the document_tracking_resources format, and all your datasets should complain with it.

Installation

pip install news_tracking

Utilities

Once installed, you will be provided new commands which act as a front-end to the utilities your need.

• news_tracking_miranda: run the Miranda et al. algorithm (see the document_tracking package for more information). on you datasets. It needs to have a model, as the algorithm is supervised.
• news_tracking_miranda_training: train and export a model for the Miranda et al. algorithm.
• news_tracking_kmeans: run K-Means on a dataset, it can act as a baseline algorithm.
• news_tracking_evaluation: evaluate the clustering of algorithms using Standard and BCubed metrics.