v0.8.0

What's Changed

Added

• Docker image snapshots are built and pushed on dockerhub at each commit
• new Dockerfile.local that does not clone from github
• End 2 end evaluation using MeasEVAL (#164)

Changed

• Updated to Grobid version 0.8.0
• Updated to Dropwizard version 4.x (from version 1.x)
• Updated training data, removed some leftover callout references that were partially removed
• Updated models and evaluations (available here)

Fixed

• Fixed and improved the word2number that now supports also fractions and other constructs #176, #110, #91
• Fixed the segmentation issue for the quantified object. Now the spurious characters from PDF documents are removed #158

Full Changelog: v0.7.3...v0.8.0

0.7.3

Note

This release has been re-built on the 26th of June 2023 because of some issues with the docker image and the versioning.

What's Changed

• Fix notation lexicon #97
• Fix list and labelled sequence extraction with DL BERT models #153
• Improve recognition of composed units using sentence segmentation #155 #87
• Added more units in the lexicon
• Added missing log when exception are raised
• Upgrade to grobid 0.7.3 and support to JDK > 11
• Updated Docker image to support JDK 17 and use the gradle distribution script instead of the JAR directly
• Introduced Kotlin for new development
• Transitioned from circleci to github actions

Full Changelog: 0.7.2...v0.7.3

0.7.2

What's Changed

• Update to Grobid 0.7.2
• Create holdout set by @lfoppiano in #145
• Add additional DL and transformers models by @lfoppiano in #146
• Fix value parser's incorrect recognition by @lfoppiano in #141

Full Changelog: 0.7.1...0.7.2

0.7.1

Added

• New BidLSTM_CRF models for quantities, values and units parsing #129
• Update to Grobid 0.7.1 #137

Changed

• Use the grobid sentence segmentation for the quantified object sentence splitting #138
• Update docker image to allow DL models #142

Fixed

• Fixes incorrect boxes colors #125
• Fixed lexicon #134

0.7.0

In this release:

• various normalisation improvements to cover products of multiple units #95
• update to grobid 0.7.0 and new configuration file via yaml #123
• added additional units definitions
• various bugfixes
• add docker image #126

After updating to this release, it's important to update the models by running ./gradlew copyModels

0.6.0

First official release of grobid-quantities