Intel/WEC-Eng

1# WEC-Eng2A large-scale dataset for cross-document event coreference extracted from English Wikipedia. </br>3 4- **Repository (Code for generating WEC):** https://github.com/AlonEirew/extract-wec5- **Paper:** https://aclanthology.org/2021.naacl-main.198/6 7### Languages8 9English10 11## Load Dataset12You can read in WEC-Eng files as follows (using the **huggingface_hub** library):13 14```json15from huggingface_hub import hf_hub_url, cached_download16import json17REPO_ID = "datasets/Intel/WEC-Eng"18splits_files = ["Dev_Event_gold_mentions_validated.json",19                "Test_Event_gold_mentions_validated.json",20                "Train_Event_gold_mentions.json"]21wec_eng = list()22for split_file in splits_files:23    wec_eng.append(json.load(open(cached_download(24        hf_hub_url(REPO_ID, split_file)), "r")))25```26 27## Dataset Structure28 29### Data Splits30- **Final version of the English CD event coreference dataset**<br>31    - Train - Train_Event_gold_mentions.json 32    - Dev - Dev_Event_gold_mentions_validated.json33    - Test - Test_Event_gold_mentions_validated.json34 35|                             | Train   | Valid | Test |36| -----                       | ------ | ----- | ----  |37| Clusters                    | 7,042  |  233  | 322   |38| Event Mentions              | 40,529 |  1250 | 1,893 |39 40- **The non (within clusters) controlled version of the dataset (lexical diversity)**<br>41    - All (experimental) - All_Event_gold_mentions_unfiltered.json42 43### Data Instances44 45```json46{47        "coref_chain": 2293469,48        "coref_link": "Family Values Tour 1998",49        "doc_id": "House of Pain",50        "mention_context": [51            "From",52            "then",53            "on",54            ",",55            "the",56            "members",57            "continued",58            "their"59  ],60  "mention_head": "Tour",61  "mention_head_lemma": "Tour",62  "mention_head_pos": "PROPN",63  "mention_id": "108172",64  "mention_index": 1,65  "mention_ner": "UNK",66  "mention_type": 8,67  "predicted_coref_chain": null,68  "sent_id": 2,69  "tokens_number": [70    50,71    51,72    52,73    5374  ],75  "tokens_str": "Family Values Tour 1998",76  "topic_id": -177}78```79 80### Data Fields81 82|Field|Value Type|Value|83|---|:---:|---|84|coref_chain|Numeric|Coreference chain/cluster ID|85|coref_link|String|Coreference link wikipeida page/article title|86|doc_id|String|Mention page/article title|87|mention_context|List[String]|Tokenized mention paragraph (including mention)|88|mention_head|String|Mention span head token|89|mention_head_lemma|String|Mention span head token lemma|90|mention_head_pos|String|Mention span head token POS|91|mention_id|String|Mention id|92|mention_index|Numeric|Mention index in json file|93|mention_ner|String|Mention NER|94|tokens_number|List[Numeric]|Mentions tokens ids within the context|95|tokens_str|String|Mention span text|96|topic_id|Ignore|Ignore|97|mention_type|Ignore|Ignore|98|predicted_coref_chain|Ignore|Ignore|99|sent_id|Ignore|Ignore|100 101## Citation102```103@inproceedings{eirew-etal-2021-wec,104    title = "{WEC}: Deriving a Large-scale Cross-document Event Coreference dataset from {W}ikipedia",105    author = "Eirew, Alon  and106      Cattan, Arie  and107      Dagan, Ido",108    booktitle = "Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",109    month = jun,110    year = "2021",111    address = "Online",112    publisher = "Association for Computational Linguistics",113    url = "https://aclanthology.org/2021.naacl-main.198",114    doi = "10.18653/v1/2021.naacl-main.198",115    pages = "2498--2510",116    abstract = "Cross-document event coreference resolution is a foundational task for NLP applications involving multi-text processing. However, existing corpora for this task are scarce and relatively small, while annotating only modest-size clusters of documents belonging to the same topic. To complement these resources and enhance future research, we present Wikipedia Event Coreference (WEC), an efficient methodology for gathering a large-scale dataset for cross-document event coreference from Wikipedia, where coreference links are not restricted within predefined topics. We apply this methodology to the English Wikipedia and extract our large-scale WEC-Eng dataset. Notably, our dataset creation method is generic and can be applied with relatively little effort to other Wikipedia languages. To set baseline results, we develop an algorithm that adapts components of state-of-the-art models for within-document coreference resolution to the cross-document setting. Our model is suitably efficient and outperforms previously published state-of-the-art results for the task.",117}118```119 120 121## License122We provide the following data sets under a <a href="https://creativecommons.org/licenses/by-sa/3.0/deed.en_US">Creative Commons Attribution-ShareAlike 3.0 Unported License</a>. It is based on content extracted from Wikipedia that is licensed under the Creative Commons Attribution-ShareAlike 3.0 Unported License123 124## Contact125If you have any questions please create a Github issue at https://github.com/AlonEirew/extract-wec.