GROBID

[!TIP] Getting started here.

Summary

GROBID (or Grobid, but not GroBid nor GroBiD) means GeneRation Of BIbliographic Data.

GROBID is a machine learning library for extracting, parsing and re-structuring raw documents such as PDF into structured XML/TEI encoded documents with a particular focus on technical and scientific publications. First developments started in 2008 as a hobby, following a suggestion by Laurent Romary (Inria, France). In 2011, the tool has been made available in open source. Work on GROBID has been steady as a side project since the beginning and is expected to continue as such, facilitated in particular to the continuous support of Inria.

The following functionalities are available:

• Header extraction and parsing from article in PDF format. The extraction here covers the usual bibliographical information (e.g. title, abstract, authors, affiliations, keywords, etc.).
• References extraction and parsing from articles in PDF format, around .87 F1-score against on an independent PubMed Central set of 1943 PDF containing 90,125 references, and around .90 on a similar bioRxiv set of 2000 PDF (using the Deep Learning citation model). All the usual publication metadata are covered (including DOI, PMID, etc.).
• Citation contexts recognition and resolution of the full bibliographical references of the article. The accuracy of citation contexts resolution is between .76 and .91 F1-score depending on the evaluation collection (this corresponds to both the correct identification of the citation callout and its correct association with a full bibliographical reference).
• Full text extraction and structuring from PDF articles, including a model for the overall document segmentation and models for the structuring of the text body (paragraph, section titles, reference and footnote callouts, figures, tables, data availability statements, etc.).
• PDF coordinates for extracted information, allowing to create "augmented" interactive PDF based on bounding boxes of the identified structures.
• Parsing of references in isolation (above .90 F1-score at instance-level, .95 F1-score at field level, using the Deep Learning model).
• Parsing of names (e.g. person title, forenames, middle name, etc.), in particular author names in header, and author names in references (two distinct models).
• Parsing of affiliation and address blocks.
• Parsing of dates, ISO normalized day, month, year.
• Consolidation/resolution of the extracted bibliographical references using the biblio-glutton service or the CrossRef REST API. In both cases, DOI/PMID resolution performance is higher than 0.95 F1-score from PDF extraction.
• Extraction and parsing of patent and non-patent references in patent publications.
• Extraction of Funders and funding information with optional matching of extracted funders with the CrossRef Funder Registry.
• Identification of copyrights' owner and license associated to the document, e.g. publisher or authors copyrights, CC-BY/CC-BY-NC/etc. license.

In a complete PDF processing, GROBID manages 68 final labels used to build relatively fine-grained structures, from traditional publication metadata (title, author first/last/middle names, affiliation types, detailed address, journal, volume, issue, pages, DOI, PMID, etc.) to full text structures (section title, paragraph, reference markers, head/foot notes, figure captions, etc.).

GROBID includes a comprehensive web service API, Docker images, batch processing, a JAVA API, a generic training and evaluation framework (precision, recall, etc., n-fold cross-evaluation), systematic end-to-end benchmarking on thousand documents and the semi-automatic generation of training data.

GROBID can be considered as production ready. Deployments in production includes ResearchGate, Semantic Scholar, HAL Research Archive, scite.ai, Academia.edu, Internet Archive Scholar, INIST-CNRS, CERN (Invenio), and many more. The tool is designed for speed and high scalability in order to address the full scientific literature corpus.

Requirements

• OpenJDK 21 for building GROBID from source
• Linux (64 bits) or macOS (Intel and ARM) for native builds
• [Optional] Python 3.8+ with JEP for Deep Learning models
• [Optional] NVIDIA GPU with CUDA support for faster Deep Learning models

[!TIP] We bump to OpenJDK 21, however some dependencies may require an earlier version, so we might increase the runtime backward compatibility to JDK 17+ in the next release, > 0.8.2.

For detailed installation instructions, including JDK setup and platform-specific requirements, see the Installation documentation.

GROBID should run properly "out of the box" on Linux (64 bits) and macOS (Intel and ARM). We cannot ensure currently support for Windows as we did before (help welcome!).

GROBID uses Deep Learning models relying on the DeLFT library, a task-agnostic Deep Learning framework for sequence labelling and text classification, via JEP. GROBID can run Deep Learning architectures (RNN or transformers with or without layout feature channels) or with feature engineered CRF (default), or any mixtures of CRF and DL to balance scalability and accuracy. These models use joint text and visual/layout information provided by pdfalto.

Note that by default the Deep Learning models are not used, only CRF are selected in the default configuration to accommodate "out of the box" hardware. For improved accuracy, you need to select the Deep Learning models to be used in the GROBID configuration file, according to your need and hardware capacities (in particular GPU availability and runtime requirements). Some GROBID Deep Learning models perform significantly better than default CRF, in particular for bibliographical reference parsing, so it is recommended to consider selecting them to use this tool appropriately.

Demo

Demo server

For testing purposes, two public GROBID demo servers are available thanks to HuggingFace, hosted as spaces.

A GROBID demo server with a combination of Deep Learning models and CRF models is available at the following address: https://kermitt2-grobid.hf.space/ or at https://huggingface.co/spaces/kermitt2/grobid. This demo runs however on CPU only. If you have GPU for your own server deployment, it will be significantly faster.

A faster demo with CRF only is available at https://kermitt2-grobid-crf.hf.space/ or https://huggingface.co/spaces/kermitt2/grobid-crf. However, accuracy is lower.

The Web services are documented here.

Warning: Some quota and query limitation apply to the demo server! Please be courteous and do not overload the demo server. For any serious works, you will need to deploy and use your own Grobid server, see the GROBID and Docker containers documentation for doing that easily and activate some Deep Learning models.

Try in Play With Docker

<a href="https://labs.play-with-docker.com/?stack=https://raw.githubusercontent.com/kermitt2/grobid/master/compose.yml"> <img src="https://raw.githubusercontent.com/play-with-docker/stacks/master/assets/images/button.png" alt="Try in PWD"/> </a>

Wait for 30 seconds for Grobid container to be created before opening a browser tab on port 8080. This demo container runs only with CRF models. Note that there is an additional 60s needed when processing a PDF for the first time for the loading of the models on the "cold" container. Then this Grobid container is available just for you during 4 hours.

Clients

For facilitating the usage GROBID service at scale, we provide clients written in Python, Java, node.js using the web services for parallel batch processing:

• <a href="https://github.com/kermitt2/grobid-client-python" target="_blank">Python GROBID c