Skip to main content

Sentence tagger for biomedical abstracts.

Project description

This module contains a fully-standalone implementation of the PIBOSO tagger that won the ALTA2012 Shared task [1]. The features and algorithms used are described in [2].


The tagger (including a pre-trained model) is packaged as a Python module and distributed via pypi. Installing it should be as simple as

pip install piboso


hydrat [3] - automatically installed by pip TreeTagger [4] - must be manually installed


The path to the folder in which treetagger is located must be specified in configuration file. When invoked, piboso_tag will attempt to locate a configuration file at ~/.pibosorc and ./.pibosorc. If neither exists, it will generate a blank configuration file at ./.pibosorc. The path to treetagger should be set in this configuration file.

An alternative location for reading the configuration file can be specified with the -c command-line option.

Using the tagger

The tagger can be invoked with the script piboso_tag, that is automatically installed when the package is installed with pip. The simplest invocation is

piboso_tag -o <OUTPUT_PATH> <FILE TO TAG> <FILE TO TAG> …

If no files are specified on the command line, piboso_tag will read STDIN and interpret each line as a path to a file to be tagged. More detailed information about invoking piboso_tag can be obtained by invoking

piboso_tag –help

Files are assumed to be sentence tokenized and presented in a sentence-per-line format. The output produced by piboso-tag is in a CSV format, for example:

subsample/1454068-1,background subsample/1454068-2,background subsample/1454068-3,outcome subsample/1454088-1,background subsample/1454088-2,background subsample/1454088-3,background subsample/1454088-4,background

The first item in each record is the path of the file and the sentence number separated by a dash. Sentences are enumerated from 1. The second item is the label assigned to the sentence.


Marco Lui <>

[1] [2] [3] [4]

Project details

Release history Release notifications | RSS feed

This version


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

piboso-1.tar.gz (24.3 MB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page