{"url":"/dataset/fmc-mwo2kg","name":"FMC-MWO2KG","full_name":"The MWO2KG Failure Mode Classification Dataset","description_markdown":"The Failure Mode Classification dataset released in the paper [\"MWO2KG and Echidna: Constructing and exploring knowledge graphs from maintenance data\"](https://journals.sagepub.com/doi/10.1177/1748006X221131128) by  Stewart et al. The goal is to label a given observation (made by a maintainer) with the corresponding Failure Mode Code.\r\n\r\nEach row contains an observation made by a maintainer, followed by a comma, followed by the Failure Mode, for example:\r\n\r\n    falure,Breakdown\r\n\r\nAs they are written in technical language, there are often spelling/grammatical/tokenisation errors made in the observations - these are typical of maintenance work orders.\r\n\r\nThe dataset comprises 502 (observation, label) pairs (for training), 62 pairs (for validation) and 62 pairs (for testing). The labels are taken from a set of 22 failure mode codes from ISO 14224. In order to pull a list of observations in which to label, we ran MWO2KG over the data once and exported a list of all entities labelled as ‘observation’ (such as ‘leaking’, ‘not working’) by the Named Entity Recognition model. We then removed all results that were incorrectly predicted as observations by the NER model and proceeded to label each observation with the most appropriate failure mode code using a text editor.\r\n\r\nThe source code of the above paper (which also includes this dataset) is located on [GitHub](https://github.com/nlp-tlp/mwo2kg-and-echidna).\r\n\r\nThe direct link to the data (`train.txt`, `dev.txt`, and `test.txt`) is available [here](https://github.com/nlp-tlp/mwo2kg-and-echidna/tree/main/mwo2kg/failure_mode_classification/input_data).","description_withheld":null,"homepage":"https://github.com/nlp-tlp/mwo2kg-and-echidna/tree/main/mwo2kg/failure_mode_classification/input_data","introduced_date":"2022-11-05","introduced_date_note":null,"introduced_by":null,"license":{"name":"Apache 2.0","url":"https://github.com/nlp-tlp/mwo2kg-and-echidna/blob/main/LICENSE"},"modalities":[{"name":"Texts","url":"/datasets/modality/texts"}],"tasks":[{"name":"Text Classification","url":"/task/text-classification","datasets_with_task":"/datasets/task/text-classification"}],"languages":[{"name":"English","url":"/datasets/language/english"}],"variants":["FMC-MWO2KG"],"data_loaders":[],"num_papers_in_archive":1,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28"},"benchmarks":[{"leaderboard":"/sota/text-classification-on-fmc-mwo2kg-1","task":"Text Classification","dataset_variant":"FMC-MWO2KG","rows":1,"metrics":["Macro F1","Micro F1"],"first_row_in_archive_order":{"model":"Flair","paper":"/paper/mwo2kg-and-echidna-constructing-and-exploring","metrics":{"Macro F1":"0.459","Micro F1":"0.597"},"code_links":[{"title":"nlp-tlp/mwo2kg-and-echidna","url":"https://github.com/nlp-tlp/mwo2kg-and-echidna"}]},"note":"rows are the archive's own order at snapshot; nothing here re-ranks them"}],"papers_with_a_benchmark_row":[{"paper":"/paper/mwo2kg-and-echidna-constructing-and-exploring","title":"MWO2KG and Echidna: Constructing and exploring knowledge graphs from maintenance data","date":"2022-11-05","rows_on_this_dataset":1,"code_links":1,"syntology":null}],"syntology_totals":{"read_at":"2026-09-25T09:33:49+00:00","papers_with_samples":0,"samples_harvested":0,"samples_ran":0,"samples_unverified":0,"pointer_only_for_licence":0,"papers_with_no_sample_that_ran":0,"note":"the per-paper counts above, summed; not a rate"},"papers_note":"The archive never published its papers-using-dataset list; these are papers with a leaderboard row on this dataset's benchmarks."}