AnnoCTR.py

"""The Anno CTR Dataset"""


import datasets
import json
import requests

logger = datasets.logging.get_logger(__name__)

_DESCRIPTION = """\
AnnoCTR consists of 400 cyber threat reports that have been obtained from commercial CTI vendors. The reports describe threat-related information such as tactics, techniques, actors, tools, and targeted industries. The reports have been annotated by a domain expert with named entities, temporal expressions, and cybersecurity-specific concepts. The annotations include mentions of organizations, locations, industry sectors, time expressions, code snippets, hacker groups, malware, tools, tactics, and techniques.

The dataset is split into three parts: train, dev, and test, with 60%, 15%, and 25% of the documents, respectively. The train set is used for model training, the dev set is used for model selection, and the test set is used for evaluation.

For further information on the annotation scheme, please refer to our paper and the annotation guidelines for the general concepts and cybersecurity-specific concepts.
"""


# TODO: Add a link to an official homepage for the dataset here
_HOMEPAGE = "https://github.com/boschresearch/anno-ctr-lrec-coling-2024"

# TODO: Add the licence for the dataset here if you can find it
_LICENSE = "The AnnoCTR corpus located in the folder AnnoCTR is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License (CC-BY-SA 4.0)."

_CITATION = """\

Lukas Lange, Marc Müller, Ghazaleh Haratinezhad Torbati, Dragan Milchevski, Patrick Grau, Subhash Pujari, Annemarie Friedrich. AnnoCTR: A Dataset for Detecting and Linking Entities, Tactics, and Techniques in Cyber Threat Reports. LREC-COLING 2024.

"""


_URL = "https://raw.githubusercontent.com/boschresearch/anno-ctr-lrec-coling-2024/d510b6949e1938d47c93a43eedd562dc538439dc/AnnoCTR/ner_json/"

_TRAINING_FILE="train.json"
_TEST_FILE = "test.json"
_DEV_FILE = "dev.json"

class AnnoCTRConfig(datasets.BuilderConfig):
    """The AnnoCTR dataset configuration"""

    def __init__(self, **kwargs):
        """BuilderConfig for Anno CTR dataset.
        Args:
          **kwargs: keyword arguments forwarded to super.
        """
        super(AnnoCTRConfig, self).__init__(**kwargs)


class AnnoCTRDataset(datasets.GeneratorBasedBuilder):
    """The Open NER dataset Entities Dataset."""

    VERSION = datasets.Version("1.0.0")

    # You will be able to load one or the other configurations in the following list with
    # data = datasets.load_dataset('my_dataset', 'first_domain')
    # data = datasets.load_dataset('my_dataset', 'second_domain')
    BUILDER_CONFIGS = [
        AnnoCTRConfig(name="all_tags", version=VERSION, description="Include all tags"),
        AnnoCTRConfig(name="everything", version=VERSION, description="Everything"),
    ]

    DEFAULT_CONFIG_NAME = "all_tags"  # It's not mandatory to have a default configuration. Just use one if it make sense.


    def _get_tags(self,url,col="all_tags"):
      names = set()
      r = requests.get(url)
      for line in r.text.splitlines():
          data = json.loads(line)
          tags = data[col]

          names = names.union(set(tags))

      return list(names)

    def _info(self):

        if self.config.name == "all_tags":
            the_url = f"{_URL}{_TRAINING_FILE}"

            logger.info("Loading the %s" % the_url)
            all_tags = self._get_tags(the_url,col="all_tags")

            logger.info("Found %d tags" % len(all_tags))

            features=datasets.Features(
                {
                    "tokens": datasets.Sequence(datasets.Value("string")),
                    "text": datasets.Value("string"),
                    "all_tags": datasets.Sequence(datasets.features.ClassLabel(names=all_tags))
                }
            )

        else:
            features=datasets.Features(
                {
                    "tokens": datasets.Sequence(datasets.Value("string")),
                    "text": datasets.Value("string"),
                    "ne_tags": datasets.Sequence(datasets.Value("string")),
                    "nc_tags": datasets.Sequence(datasets.Value("string")),
                    "te_tags": datasets.Sequence(datasets.Value("string")),
                    "ce_tags": datasets.Sequence(datasets.Value("string")),
                    "ci_tags": datasets.Sequence(datasets.Value("string")),
                    "all_tags": datasets.Sequence(datasets.Value("string"))
                }
            )

        #logger.info("Total names = %d" % len(our_names))

        dinfo = datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            supervised_keys=None,
            homepage=_HOMEPAGE,
            citation=_CITATION,
        )

        logger.info(dinfo)
        
        return dinfo

    def _split_generators(self, dl_manager):
        """Returns SplitGenerators."""
        urls_to_download = {
            "train": f"{_URL}{_TRAINING_FILE}",
            "dev": f"{_URL}{_DEV_FILE}",
            "test": f"{_URL}{_TEST_FILE}",
        }
        downloaded_files = dl_manager.download_and_extract(urls_to_download)

        return [
            datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files["train"]}),
            datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["dev"]}),
            datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["test"]}),
        ]

    def _generate_examples(self, filepath):
        logger.info("⏳ Generating examples from = %s", filepath)

        with open(filepath, encoding="utf-8") as f:
          for key, row in enumerate(f):
              data = json.loads(row)

              if self.config.name == "all_tags":
                  # Yields examples as (key, example) tuples
                  yield key, {
                      "tokens": data["tokens"],
                      "text": data['text'],
                      "all_tags": data['all_tags']
                  }

              if self.config.name == "everything":
                  # Yields examples as (key, example) tuples
                  data_new = data.copy()
                  data_new.pop("id")

                  yield key, data_new