opusparcus.json

{
  "overview": {
    "where": {
      "has-leaderboard": "no",
      "leaderboard-url": "N/A",
      "leaderboard-description": "N/A",
      "paper-bibtext": "```\n@InProceedings{creutz:lrec2018,\n  title = {Open Subtitles Paraphrase Corpus for Six Languages},\n  author={Mathias Creutz},\n  booktitle={Proceedings of the 11th edition of the Language Resources and Evaluation Conference (LREC 2018)},\n  year={2018},\n  month = {May 7-12},\n  address = {Miyazaki, Japan},\n  editor = {Nicoletta Calzolari (Conference chair) and Khalid Choukri and Christopher Cieri and Thierry Declerck and Sara Goggi and Koiti Hasida and Hitoshi Isahara and Bente Maegaard and Joseph Mariani and H\u00e9l\u00e8ne Mazo and Asuncion Moreno and Jan Odijk and Stelios Piperidis and Takenobu Tokunaga},\n  publisher = {European Language Resources Association (ELRA)},\n  isbn = {979-10-95546-00-9},\n  language = {english},\n  url={http://www.lrec-conf.org/proceedings/lrec2018/pdf/131.pdf}\n```",
      "paper-url": "[LREC](http://www.lrec-conf.org/proceedings/lrec2018/pdf/131.pdf)",
      "data-url": "[Website](http://urn.fi/urn:nbn:fi:lb-2018021221)",
      "website": "[Website](http://urn.fi/urn:nbn:fi:lb-2018021221)",
      "contact-name": "Mathias Creutz",
      "contact-email": "firstname dot lastname at helsinki dot fi"
    },
    "languages": {
      "is-multilingual": "yes",
      "license": "cc-by-nc-4.0: Creative Commons Attribution Non Commercial 4.0 International",
      "task-other": "N/A",
      "language-names": [
        "German",
        "English",
        "Finnish",
        "French",
        "Russian",
        "Swedish"
      ],
      "language-speakers": "Opusparcus is a paraphrase corpus for six European languages: German, English, Finnish, French, Russian, and Swedish. The paraphrases consist of subtitles from movies and TV shows.\n\nThe data in Opusparcus has been extracted from [OpenSubtitles2016](http://opus.nlpl.eu/OpenSubtitles2016.php), which is in turn based on data from [OpenSubtitles](http://www.opensubtitles.org/).",
      "license-other": "N/A",
      "task": "Paraphrasing",
      "communicative": "Models can be trained, e.g., for paraphrase detection and generation, that is, determining whether two given sentences mean the same thing or generating new paraphrases for a given sentence. ",
      "intended-use": "Opusparcus is a sentential paraphrase corpus for multiple languages containing colloquial language."
    },
    "credit": {
      "gem-added-by": "Mathias Creutz (University of Helsinki)"
    },
    "structure": {
      "data-fields": "- `sent1`: a tokenized sentence\n- `sent2`: another tokenized sentence, which is potentially a paraphrase of `sent1`.\n- `annot_score`: a value between 1.0 and 4.0 indicating how good an example of paraphrases `sent1` and `sent2` are. (For the training sets, the value is 0.0, which indicates that no manual annotation has taken place.)\n- `lang`: language of this dataset\n- `gem_id`: unique identifier of this entry\n\nAll fields are strings except `annot_score`, which is a float.",
      "structure-description": "For each target language, the Opusparcus data have been partitioned into three types of data sets: training, validation and test sets. The training sets are large, consisting of millions of sentence pairs, and have been compiled automatically, with the help of probabilistic ranking functions. The development and test sets consist of sentence pairs that have been annotated manually; each set contains approximately 1000 sentence pairs that have been verified to be acceptable paraphrases by two independent annotators.\n\nWhen you download Opusparcus, you must always indicate the language you want to retrieve, for instance:\n\n```\ndata = load_dataset(\"GEM/opusparcus\", lang=\"de\")\n```\n\nThe above command will download the validation and test sets for German. If additionally, you want to retrieve training data, you need to specify the level of quality you desire, such as \"French, with 90% quality of the training data\":\n\n```\ndata = load_dataset(\"GEM/opusparcus\", lang=\"fr\", quality=90)\n```\n\nThe entries in the training sets have been ranked automatically by how likely they are paraphrases, best first, worst last. The quality parameter indicates the estimated proportion (in percent) of true\nparaphrases in the training set. Allowed quality values range between 60 and 100, in increments of 5 (60, 65, 70, ..., 100). A value of 60 means that 60% of the sentence pairs in the training set are estimated to be true paraphrases (and the remaining 40% are not). A higher value produces a smaller but cleaner set. The smaller sets are subsets of the larger sets, such that the `quality=95` set is a subset of `quality=90`, which is a subset of `quality=85`, and so on.\n\nThe default `quality` value, if omitted, is 100. This matches no training data at all, which can be convenient, if you are only interested in the validation and test sets, which are considerably\nsmaller, but manually annotated.\n\nNote that an alternative to typing the parameter values explicitly, you can use configuration names instead. The following commands are equivalent to the ones above:\n\n```\ndata = load_dataset(\"GEM/opusparcus\", \"de.100\")\ndata = load_dataset(\"GEM/opusparcus\", \"fr.90\")\n```",
      "structure-labels": "Annotators have used the following scores to label sentence pairs in the test and validation sets:\n\n4: Good example of paraphrases (Dark green button in the annotation tool): The two sentences can be used in the same situation and essentially \"mean the same thing\".\n\n3: Mostly good example of paraphrases (Light green button in the annotation tool): It is acceptable to think that the two sentences refer to the same thing, although one sentence might be more specific\nthan the other one, or there are differences in style, such as polite form versus familiar form.\n\n2: Mostly bad example of paraphrases (Yellow button in the annotation tool): There is some connection between the sentences that explains why they occur together, but one would not really consider them to mean the same thing.\n\n1: Bad example of paraphrases (Red button in the annotation tool): There is no obvious connection. The sentences mean different things.\n\nIf the two annotators fully agreed on the category, the value in the `annot_score` field is 4.0, 3.0, 2.0 or 1.0.  If the two annotators chose adjacent categories, the value in this field will be 3.5, 2.5 or\n1.5.  For instance, a value of 2.5 means that one annotator gave a score of 3 (\"mostly good\"), indicating a possible paraphrase pair, whereas the other annotator scored this as a 2 (\"mostly bad\"), that is, unlikely to be a paraphrase pair.  If the annotators disagreed by more than one category, the sentence pair was discarded and won't show up in the datasets.\n\nThe training sets were not annotated manually. This is indicated by\nthe value 0.0 in the `annot_score` field.\n\nFor an assessment of of inter-annotator agreement, see Aulamo et al. (2019). [Annotation of subtitle paraphrases using a new web tool.](http://ceur-ws.org/Vol-2364/3_paper.pdf) In *Proceedings of the\nDigital Humanities in the Nordic Countries 4th Conference*, Copenhagen, Denmark.",
      "structure-example": "```\n{'annot_score': 4.0, 'gem_id': 'gem-opusparcus-test-1587', 'lang': 'en', 'sent1': \"I haven 't been contacted by anybody .\", 'sent2': \"Nobody 's contacted me .\"}\n```",
      "structure-splits": "The data is split into training, validation and test sets. The validation and test sets come in two versions, the regular validation and test sets and the full sets, called validation.full and test.full. The full sets contain all sentence pairs successfully annotated by the annotators, including the sentence pairs that were rejected as paraphrases. The annotation scores of the full sets thus range between 1.0 and 4.0. The regular validation and test sets only contain sentence pairs that qualify as paraphrases, scored between 3.0 and 4.0 by the annotators.\n\nThe number of sentence pairs in the data splits are as follows for each of the languages. The range between the smallest (`quality=95`) and largest (`quality=60`) train configuration have been shown.\n\n|       | train         | valid | test | valid.full | test.full |\n| ----- | ------        | ----- | ---- | ---------- | --------- |\n| de    | 0.59M .. 13M  |  1013 | 1047 |       1582 |      1586 |\n| en    |  1.0M .. 35M  |  1015 |  982 |       1455 |      1445 |\n| fi    | 0.48M .. 8.9M |   963 |  958 |       1760 |      1749 |\n| fr    | 0.94M .. 22M  |   997 | 1007 |       1630 |      1674 |\n| ru    | 0.15M .. 15M  |  1020 | 1068 |       1854 |      1855 |\n| sv    | 0.24M .. 4.5M |   984 |  947 |       1887 |      1901 |\n\nAs a concrete example, loading the English data requesting 95% quality of the train split produces the following:\n\n```\n>>> data = load_dataset(\"GEM/opusparcus\", lang=\"en\", quality=95)\n\n>>> data\nDatasetDict({\n    test: Dataset({\n        features: ['lang', 'sent1', 'sent2', 'annot_score', 'gem_id'],\n        num_rows: 982\n    })\n    validation: Dataset({\n        features: ['lang', 'sent1', 'sent2', 'annot_score', 'gem_id'],\n        num_rows: 1015\n    })\n    test.full: Dataset({\n        features: ['lang', 'sent1', 'sent2', 'annot_score', 'gem_id'],\n        num_rows: 1445\n    })\n    validation.full: Dataset({\n        features: ['lang', 'sent1', 'sent2', 'annot_score', 'gem_id'],\n        num_rows: 1455\n    })\n    train: Dataset({\n        features: ['lang', 'sent1', 'sent2', 'annot_score', 'gem_id'],\n        num_rows: 1000000\n    })\n})\n\n>>> data[\"test\"][0]\n{'annot_score': 4.0, 'gem_id': 'gem-opusparcus-test-1587', 'lang': 'en', 'sent1': \"I haven 't been contacted by anybody .\", 'sent2': \"Nobody 's contacted me .\"}\n\n>>> data[\"validation\"][2]\n{'annot_score': 3.0, 'gem_id': 'gem-opusparcus-validation-1586', 'lang': 'en', 'sent1': 'No promises , okay ?', 'sent2': \"I 'm not promising anything .\"}\n\n>>> data[\"train\"][1000]\n{'annot_score': 0.0, 'gem_id': 'gem-opusparcus-train-12501001', 'lang': 'en', 'sent1': 'Am I beautiful ?', 'sent2': 'Am I pretty ?'}",
      "structure-splits-criteria": "The validation and test sets have been annotated manually, but the training sets have been produced using automatic scoring and come in different size configurations depending on the desired quality level. (See above descriptions and examples for more details.)\n\nPlease note that previous work suggests that a larger and noisier training set is better than a\nsmaller and clean set. See Sj\u00f6blom et al. (2018). [Paraphrase Detection on Noisy Subtitles in Six\nLanguages](http://noisy-text.github.io/2018/pdf/W-NUT20189.pdf). In *Proceedings of the 2018 EMNLP Workshop W-NUT: The 4th Workshop on Noisy User-generated Text*, and Vahtola et al. (2021). [Coping with Noisy Training Data Labels in Paraphrase Detection](https://aclanthology.org/2021.wnut-1.32/). In *Proceedings of the 7th Workshop on Noisy User-generated Text*.\n"
    },
    "what": {
      "dataset": "Opusparcus is a paraphrase corpus for six European languages: German, English, Finnish, French, Russian, and Swedish. The paraphrases consist of subtitles from movies and TV shows."
    }
  },
  "curation": {
    "original": {
      "rationale": "Opusparcus was created in order to produce a *sentential* paraphrase corpus for multiple languages containing *colloquial* language (as opposed to news or religious text, for instance).",
      "communicative": "Opusparcus provides labeled examples of pairs of sentences that have similar (or dissimilar) meanings.",
      "is-aggregated": "no",
      "aggregated-sources": "N/A"
    },
    "language": {
      "obtained": [
        "Crowdsourced"
      ],
      "found": [],
      "crowdsourced": [
        "Other crowdworker platform"
      ],
      "created": "N/A",
      "machine-generated": "N/A",
      "producers-description": "The data in Opusparcus has been extracted from [OpenSubtitles2016](http://opus.nlpl.eu/OpenSubtitles2016.php), which is in turn based on data from [OpenSubtitles.org](http://www.opensubtitles.org/).\n\nThe texts consists of subtitles that have been produced using crowdsourcing.",
      "topics": "The language is representative of movies and TV shows. Domains covered include comedy, drama, relationships, suspense, etc.",
      "validated": "validated by data curator",
      "pre-processed": "Sentence and word tokenization was performed.",
      "is-filtered": "algorithmically",
      "filtered-criteria": "The sentence pairs in the training sets were ordered automatically based on the estimated likelihood that the sentences were paraphrases, most likely paraphrases on the top, and least likely paraphrases on the bottom.\n\nThe validation and test sets were checked and annotated manually, but the sentence pairs selected for annotation had to be different enough in terms of minimum edit distance (Levenshtein distance). This ensured that annotators would not spend their time annotating pairs of more or less identical sentences."
    },
    "annotations": {
      "origin": "expert created",
      "rater-number": "11<n<50",
      "rater-qualifications": "Students and staff at the University of Helsinki (native or very proficient speakers of the target languages)",
      "rater-training-num": "0",
      "rater-test-num": "2",
      "rater-annotation-service-bool": "no",
      "rater-annotation-service": [],
      "values": "The development and test sets consist of sentence pairs that have been annotated manually; each set contains approximately 1000 sentence pairs that have been verified to be acceptable paraphrases by two independent annotators.\n\nThe `annot_score` field reflects the judgments made by the annotators. If the annnotators fully agreed on the category (4.0: dark green, 3.0: light green, 2.0: yellow, 1.0: red), the value of `annot_score` is 4.0, 3.0, 2.0 or 1.0.  If the annotators chose adjacent categories, the value in this field will be 3.5, 2.5 or 1.5.  For instance, a value of 2.5 means that one annotator gave a score of 3 (\"mostly good\"), indicating a possible paraphrase pair, whereas the other annotator scored this as a 2 (\"mostly bad\"), that is, unlikely to be a paraphrase pair.  If the annotators disagreed by more than one category, the sentence pair was discarded and won't show up in the datasets.\n\nAnnotators could also reject a sentence pair as being corrupted data.",
      "quality-control": "validated by another rater",
      "quality-control-details": "If the annotators disagreed by more than one category, the sentence pair was discarded and is not part of the final dataset."
    },
    "consent": {
      "has-consent": "no",
      "consent-policy": "N/A",
      "consent-other": "N/A"
    },
    "pii": {
      "has-pii": "yes/very likely",
      "no-pii-justification": "N/A",
      "is-pii-identified": "no identification",
      "pii-identified-method": "N/A",
      "is-pii-replaced": "N/A",
      "pii-replaced-method": "N/A"
    },
    "maintenance": {
      "has-maintenance": "no",
      "description": "N/A",
      "contact": "N/A",
      "contestation-mechanism": "N/A",
      "contestation-link": "N/A",
      "contestation-description": "N/A"
    }
  },
  "gem": {
    "rationale": {
      "contribution": "Opusparcus provides examples of sentences that mean the same thing or have very similar meaning. Sentences are available in six languages and the style is colloquial language.",
      "sole-task-dataset": "yes",
      "sole-language-task-dataset": "yes",
      "distinction-description": "There is another data set containing manually labeled Finnish paraphrases.",
      "model-ability": "Sentence meaning"
    },
    "curation": {
      "has-additional-curation": "yes",
      "modification-types": [
        "other"
      ],
      "modification-description": "Training sets have been prepared for each the \"quality levels\" 60% \u2013 95%.\n\nIn the original release, this task was left to the user of the data.",
      "has-additional-splits": "yes",
      "additional-splits-description": "There are two versions of the validations and test sets: the regular sets which only contain positive examples of paraphrases and the full sets containing all examples.",
      "additional-splits-capacicites": "In the original release, only the full validation and test sets were supplied. The \"regular sets\" have been added in order to make it easier to test on true parapahrases only. "
    },
    "starting": {
      "research-pointers": "Creutz (2018). [Open Subtitles Paraphrase Corpus for Six Languages](http://www.lrec-conf.org/proceedings/lrec2018/pdf/131.pdf), Proceedings of the 11th edition of the Language Resources and Evaluation Conference (LREC 2018).\n\nSj\u00f6blom et al. (2018). [Paraphrase Detection on Noisy Subtitles in Six Languages](http://noisy-text.github.io/2018/pdf/W-NUT20189.pdf). In Proceedings of the 2018 EMNLP Workshop W-NUT: The 4th Workshop on Noisy User-generated Text.\n\nAulamo et al. (2019). [Annotation of subtitle paraphrases using a new web tool.](http://ceur-ws.org/Vol-2364/3_paper.pdf) In Proceedings of the Digital Humanities in the Nordic Countries 4th Conference.\n\nSj\u00f6blom et al. (2020). [Paraphrase Generation and Evaluation on Colloquial-Style Sentences](https://aclanthology.org/2020.lrec-1.224/), Proceedings of the 12th Language Resources and Evaluation Conference (LREC). \n\nVahtola et al. (2021). [Coping with Noisy Training Data Labels in Paraphrase Detection](https://aclanthology.org/2021.wnut-1.32/). In Proceedings of the 7th Workshop on Noisy User-generated Text.\n"
    }
  },
  "results": {
    "results": {
      "model-abilities": "Sentence meaning\n\nIn a scenario of paraphrase detection, the model determines whether two given sentences carry approximately the same meaning.\n\nIn a scenario of paraphrase generation, the model generates a potential paraphrase of a given sentence.",
      "metrics": [
        "BLEU",
        "BERT-Score",
        "Other: Other Metrics"
      ],
      "other-metrics-definitions": "PINC",
      "original-evaluation": "The metrics mentioned above can be used to assess how well a generated paraphrase corresponds to a given reference sentence. The PINC score additionally assesses how different the surface forms are.",
      "has-previous-results": "yes",
      "current-evaluation": "See publications on using Opusparcus",
      "previous-results": "Sj\u00f6blom et al. (2020). [Paraphrase Generation and Evaluation on Colloquial-Style Sentences](https://aclanthology.org/2020.lrec-1.224/), Proceedings of the 12th Language Resources and Evaluation Conference (LREC)."
    }
  },
  "considerations": {
    "pii": {
      "risks-description": "The data only contains subtitles of publicly available movies and TV shows."
    },
    "licenses": {
      "dataset-restrictions": [
        "non-commercial use only"
      ],
      "dataset-restrictions-other": "N/A",
      "data-copyright": [
        "non-commercial use only"
      ],
      "data-copyright-other": "N/A"
    },
    "limitations": {
      "data-technical-limitations": "Some subtitles contain typos that are caused by inaccurate OCR.",
      "data-unsuited-applications": "The models might memorize individual subtitles of existing movies and TV shows, but there is no context across sentence boundaries in the data.",
      "data-discouraged-use": "A general issue with paraphrasing is that very small modifications in the surface form might produce valid paraphrases, which are however rather uninteresting. It is more valuable to produce paraphrases with clearly different surface realizations (e.g., measured using minimum edit distance)."
    }
  },
  "context": {
    "previous": {
      "is-deployed": "no",
      "described-risks": "N/A",
      "changes-from-observation": "N/A"
    },
    "underserved": {
      "helps-underserved": "no",
      "underserved-description": "N/A"
    },
    "biases": {
      "has-biases": "no",
      "bias-analyses": "N/A",
      "speaker-distibution": "What social bias there may be in the subtitles in this dataset has not been studied."
    }
  }
}