{ "overview": { "where": { "has-leaderboard": "no", "leaderboard-url": "N/A", "leaderboard-description": "N/A", "website": "[Website](https://turkunlp.org/paraphrase.html)", "data-url": "[Github](https://github.com/TurkuNLP/Turku-paraphrase-corpus)", "paper-url": "[ACL Anthology](https://aclanthology.org/2021.nodalida-main.29/)", "paper-bibtext": "```\n@inproceedings{kanerva-etal-2021-finnish,\n title = {Finnish Paraphrase Corpus},\n author = {Kanerva, Jenna and Ginter, Filip and Chang, Li-Hsin and Rastas, Iiro and Skantsi, Valtteri and Kilpel{\\\"a}inen, Jemina and Kupari, Hanna-Mari and Saarni, Jenna and Sev{\\'o}n, Maija and Tarkka, Otto},\n booktitle = {Proceedings of the 23rd Nordic Conference on Computational Linguistics (NoDaLiDa'21)},\n year = {2021},\n publisher = {Link{\\\"o}ping University Electronic Press, Sweden},\n url = {https://aclanthology.org/2021.nodalida-main.29},\n pages = {288--298}\n}\n```", "contact-name": "Jenna Kanerva, Filip Ginter", "contact-email": "jmnybl@utu.fi, figint@utu.fi" }, "languages": { "is-multilingual": "no", "license": "cc-by-sa-4.0: Creative Commons Attribution Share Alike 4.0 International", "task-other": "N/A", "language-names": [ "Finnish" ], "language-dialects": "written standard language, spoken language", "intended-use": "Paraphrase classification, paraphrase generation", "license-other": "N/A", "task": "Paraphrasing", "communicative": "The corpus provides naturally occurring Finnish paraphrases striving for low lexical overlap, thus supporting many different downstream applications requiring language understanding. " }, "credit": { "organization-type": [ "academic" ], "organization-names": "University of Turku", "creators": "Jenna Kanerva, Filip Ginter, Li-Hsin Chang, Iiro Rastas, Valtteri Skantsi, Jemina Kilpel\u00e4inen, Hanna-Mari Kupari, Aurora Piirto, Jenna Saarni, Maija Sev\u00f3n, Otto Tarkka (TurkuNLP / University of Turku)", "funding": "The Turku paraphrase corpus project was funded by the Academy of Finland, as well as the European Language Grid project through its open call for pilot projects. The European Language Grid project has received funding from the European Union\u2019s Horizon 2020 Research and Innovation programme under Grant Agreement no. 825627 (ELG).", "gem-added-by": "Jenna Kanerva, Filip Ginter (TurkuNLP / University of Turku)" }, "structure": { "data-fields": "The dataset consist of pairs of text passages, where a typical passage is about a sentence long, however, a passage may also be longer or shorter than a sentence. Thus, each example include two text passages (string), a manually annotated label to indicate the paraphrase type (string), and additional metadata.\n\nThe dataset include three different `modes`, plain, classification, and generation. The `plain` mode loads the original data without any additional preprocessing or transformations, while the `classification` mode directly builds the data in a form suitable for training a paraphrase classifier, where each example is doubled in the data with different directions (text1, text2, label) --> (text2, text1, label) taking care of the label flipping as well if needed (paraphrases with directionality flag < or >). In the `generation` mode, the examples are preprocessed to be directly suitable for paraphrase generation task. In here, paraphrases not suitable for generation are discarded (negative, and highly context-dependent paraphrases), and directional paraphrases are provided so that the generation goes from more detailed passage to the more general one in order to prevent model hallucination (i.e. model learning to introduce new information). The rest of the paraphrases are provided in both directions (text1, text2, label) --> (text2, text1, label).\n\n\nEach pair in `plain` and `classification` mode will include fields:\n\n`gem_id`: Identifier of the paraphrase pair (string)\n`goeswith`: Identifier of the document from which the paraphrase was extracted, can be `not available` in case the source of the paraphrase is not from document-structured data (string)\n`fold`: 0-99, data split into 100 parts respecting document boundaries, you can use this e.g. to implement crossvalidation safely as all paraphrases from one document are in one fold (int)\n`text1`: First paraphrase passage (string)\n`text2`: Second paraphrase passage (string)\n`label`: Manually annotated labels (string)\n`binary_label`: Label turned into binary with values `positive` (paraphrase) and `negative` (not-paraphrase) (string)\n`is_rewrite`: Indicator whether the example is human produced rewrite or naturally occurring paraphrase (bool)\n\nEach pair in `generation` mode will include the same fields expect `text1` and `text2` are renamed to `input` and `output` in order to indicate the generation direction. Thus the fields are:\n\n`gem_id`: Identifier of the paraphrase pair (string)\n`goeswith`: Identifier of the document from which the paraphrase was extracted, can be `not available` in case the source of the paraphrase is not from document-structured data (string)\n`fold`: 0-99, data split into 100 parts respecting document boundaries, you can use this e.g. to implement crossvalidation safely as all paraphrases from one document are in one fold (int)\n`input`: The input paraphrase passage for generation (string)\n`output`: The output paraphrase passage for generation (string)\n`label`: Manually annotated labels (string)\n`binary_label`: Label turned into binary with values `positive` (paraphrase) and `negative` (not-paraphrase) (string)\n`is_rewrite`: Indicator whether the example is human produced rewrite or naturally occurring paraphrase (bool)", "structure-example": "```\n{\n 'gem_id': 'gem-turku_paraphrase_corpus-train-15',\n 'goeswith': 'episode-02243',\n 'fold': 0,\n 'text1': 'Mit\u00e4 merkityst\u00e4 sill\u00e4 on?',\n 'text2': 'Mit\u00e4 v\u00e4li\u00e4 sill\u00e4 edes on?',\n 'label': '4',\n 'binary_label': 'positive',\n 'is_rewrite': False\n}\n```", "structure-splits": "The corpus include 3 splits: train, validation, and test.", "structure-splits-criteria": "The data is split randomly into the three section with a restriction of all paraphrases from the same document (movie, TV episode, news article, student translation, or exam question) being in the same section. All splits are manually annotated." }, "what": { "dataset": "This is a Finnish paraphrase corpus which consists of pairs of text passages, where a typical passage is about a sentence long. It can be used to either identify or generate paraphrases." } }, "curation": { "original": { "is-aggregated": "yes", "aggregated-sources": "movie and TV series subtitles (82%)\nnews articles (9%)\ndiscussion forum messages (8%)\nuniversity translation exercises (1%)\nuniversity course essays and exams (<1%)", "rationale": "The dataset is fully manually annotated. The dataset strives for interesting paraphrases with low lexical overlap, thus the annotation is two fold. First the paraphrases are manually extracted from two related documents, where the annotators are instructed to extract only interesting paraphrases. In the second phrase, all extracted paraphrases are manually labeled given the annotation scheme.\n\nThe annotation scheme is:\n4 : paraphrase in all reasonably possible contexts\n3 : paraphrase in the given document contexts, but not in general\n2 : related but not paraphrase\nDuring annotation also labels 1 (unrelated) and x (skip, e.g. wrong language) were used, however, the insignificant amount of examples annotated with these labels were discarded from the released corpus.\n\nThe following flags are annotated to label 4 paraphrases:\n< : txt1 is more general than txt2; txt2 is more specific than txt1 (directional paraphrase where txt2 can be replaced with txt1 in all contexts but not to the other direction)\n> : txt2 is more general than txt1; txt1 is more specific than txt2 (directional paraphrase where txt1 can be replaced with txt2 in all contexts but not to the other direction)\ni : minor traceable difference (differing in terms of grammatical number or case, 'this' vs 'that', etc.)\ns : style or strength difference (e.g. equivalent meaning, but one of the statements substantially more colloquial than the other)\n\nFor paraphrases where the annotated label was something else than label 4 without any flags, the annotators had an option to rewrite the text passages so that the rewritten paraphrase pair formed label 4 (universal) paraphrase. This was used for cases where simple edit would turn e.g. contextual or directional paraphrase into universal one. For the rewritten examples, both the original and the rewritten pairs are available with corresponding labels annotated.", "communicative": "Representing text passages with identical meaning but different surface realization." }, "language": { "found": [ "Multiple websites", "Offline media collection", "Other" ], "crowdsourced": [], "created": "N/A", "machine-generated": "N/A", "validated": "validated by data curator", "is-filtered": "not filtered", "filtered-criteria": "N/A", "obtained": [ "Found", "Other" ], "producers-description": "The movie and TV series subtitles are extracted from OPUS OpenSubtitles2018 collection, which is based on data from [OpenSubtitles](http://www.opensubtitles.org/).\nThe news articles are collected from two Finnish news sites, YLE and HS, during years 2017-2020.\nDiscussion forum messages are obtained from the Finnish Suomi24 discussion forum released for academic use (http://urn.fi/urn:nbn:fi:lb-2020021801).\nUniversity translation exercises, essays and exams are collected during the project.", "pre-processed": "N/A" }, "annotations": { "origin": "expert created", "rater-number": "2