Inter-annotator agreement for Brat annotation projects
Inter-annotator agreement for Brat annotation projects. For a quick overview of the output generated by
bratiaa, have a look at the example files. So far only text-bound annotations are supported, all other annotation types are ignored.
Install the package via pip.
pip install bratiaa
bratiaa expects that each first-level subdirectory of the annotation project contains the files of one annotator. It will automatically determine the set of files annotated by each annotator (files with the same relative path starting from the different annotators' directories). Here is a simple example:
example-project/ ├── annotation.conf ├── annotator-1 │ ├── doc-1.ann │ ├── doc-1.txt │ ├── doc-3.ann │ ├── doc-3.txt │ └── second │ ├── doc-2.ann │ └── doc-2.txt └── annotator-2 ├── doc-3.ann ├── doc-3.txt ├── doc-4.ann ├── doc-4.txt └── second ├── doc-2.ann └── doc-2.txt
In this example, we have two agreement documents: 'second/doc-2.txt' and 'doc-3.txt'. The other two documents are only annotated by a single annotator.
If you have a different project setup, you need to provide your own
input_generator function, yielding document objects with paths to the plain text and all corresponding ANN files (cf.
You can either use
bratiaa as a Python library or as a command-line tool.
import bratiaa as biaa project = '/path/to/brat/project' # instance-level agreement f1_agreement = biaa.compute_f1_agreement(project) # print agreement report to stdout biaa.iaa_report(f1_agreement) # agreement per label label_mean, label_sd = f1_agreement.mean_sd_per_label() # agreement per document doc_mean, doc_sd = f1_agreement.mean_sd_per_document() # total agreement total_mean, total_sd = f1_agreement.mean_sd_total()
For the token-level evaluation, please use your own tokenization function. This function should yield (start, end) offset tuples for any given string like the example function below.
import re import bratiaa as biaa def token_func(text): token = re.compile('\w+|[^\w\s]+') for match in re.finditer(token, text): yield match.start(), match.end() # token-level agreement f1_agreement = biaa.compute_f1_agreement('/path/to/brat/project' , token_func=token_func)
# instance-level agreement and heatmap brat-iaa /path/to/brat/project --heatmap instance-heatmap.png > instance-agreement.md # token-level agreement (not recommended) brat-iaa /path/to/brat/project -t --heatmap token-heatmap.png > token-agreement.md
The token-based evaluation of the command-line interface uses the generic pattern
'\S+' to identify tokens (splitting on whitespace) and hence is not recommended. Please use the Python interface with a language- and task-specific tokenizer instead.
For the output formats generated by the above commands, have a look at the example files.
For each multiply annotated document, we compute the number of true positives (TP), false positives (FP) and false negatives (FN) for each 2-combination of annotators, where each annotator contributes one set of annotations, via basic (multi)set operations. These numbers can later be aggregated along two dimensions: documents and/or labels. Based on the aggregated numbers we compute
F1 = (2*TP) / (2*TP + FP + FN) for each annotator pair. From these pair-wise F-scores, mean and standard deviation are reported (see Hripcsak & Rothschild, 2005).
An annotation instance pertaining to a certain document consists of a label and one or more start-end offset tuples (multiple start-end tuples in the case of discontinuous annotations). Two instances are considered identical if label and offset tuples match. Identical instances from a single annotator (on the same document) are considered as accidental - only unqiue annotation instances are used for calculating agreement.
Each annotation instance is split up into its overlapping tokens, e.g. if our tokenizer splits on whitespace, "[ORG Human Rights Watch]" and "[ORG Human Rights Wat]ch" both become "[ORG Human] [ORG Rights] [ORG Watch]". These split annotations are then treated as instances in the way described above with the only exception that we are dealing with multisets instead of sets, allowing for multiple token-based annotations with the same label and offsets in the case of overlapping annotations of the same type. For example, in "[LOC University of [LOC Jena]]" we have two overlapping location annotations resulting in four token-based annotations of which two are identical ("[LOC Jena]").
Be aware that "[ORG Human] [ORG Rights Watch]" and "[ORG Human Rights] [ORG Watch]" both become "[ORG Human] [ORG Rights] [ORG Watch]", that is, boundary errors between adjacent annotations of the same type are ignored!
Hripcsak, G., & Rothschild, A. S. (2005). Agreement, the f-measure, and reliability in information retrieval. Journal of the American Medical Informatics Association, 12(3), 296-298.
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size bratiaa-0.1.3-py3-none-any.whl (46.8 kB)||File type Wheel||Python version py3||Upload date||Hashes View hashes|
|Filename, size bratiaa-0.1.3.tar.gz (44.4 kB)||File type Source||Python version None||Upload date||Hashes View hashes|