For transparency, the host requests that participants post links to external data sources here as you use them, no later than the competition deadline specified. Once it has been posted, you do not need to post it again.
", "messages": [ { "id": 1272656, "postDate": "2021-04-13T16:57:15.360Z", "content": "For transparency, the host requests that participants post links to external data sources here as you use them, no later than the competition deadline specified. Once it has been posted, you do not need to post it again.
", "rawMarkdown": "For transparency, the host requests that participants post links to external data sources here as you use them, no later than the **competition** deadline specified. Once it has been posted, you do not need to post it again.", "votes": 25 }, { "id": 1276691, "postDate": "2021-04-17T21:05:53.427Z", "content": "There has been another post about it, but just to be sure: I use data from the RichContextCompetition.
", "rawMarkdown": "There has been another [post](https://www.kaggle.com/c/coleridgeinitiative-show-us-the-data/discussion/231374) about it, but just to be sure: I use data from the [RichContextCompetition](https://coleridgeinitiative.org/richcontext/richcontextcompetition/).", "votes": 9 }, { "id": 1272687, "postDate": "2021-04-13T17:25:29.800Z", "content": "@poedator Thanks for your question - I checked with @maggiemd and @philculliton about the rules - and they came up with this great response. Thanks so much!
", "rawMarkdown": "@poedator Thanks for your question - I checked with @maggiemd and @philculliton about the rules - and they came up with this great response. Thanks so much!", "votes": 3 }, { "id": 1295135, "postDate": "2021-05-06T08:21:36.760Z", "content": "I see that some participants are using external databases like this one https://www.kaggle.com/mlconsult/score-57ish-with-additional-govt-datasets
\nIs it allowed?
Hi @kulishyurii
\nFrom the code requirements page on the the competition overview:
\nThis is a Code Competition
\nSubmissions to this competition must be made through Notebooks. In order for the \"Submit\" button to be active after a commit, the following conditions must be met:
\nCPU Notebook <= 9 hours run-time\nGPU Notebook <= 9 hours run-time\nInternet access disabled\nFreely & publicly available external data is allowed, including pre-trained models\nSubmission file must be named submission.csv\nTo be eligible for prizes, one need to disclose the external data sources used before the deadline specified. For those who don't care about prizes, well, I don't think there's any way to force them to disclose it.
", "rawMarkdown": "To be eligible for prizes, one need to disclose the external data sources used before the deadline specified. For those who don't care about prizes, well, I don't think there's any way to force them to disclose it.", "votes": 1 }, { "id": 1299875, "postDate": "2021-05-10T05:32:35.843Z", "content": "Can someone please point me toward where or what is the deadline to disclose the external data sources?
\nI have seen several references to this deadline existing, but not to what the date of the deadline actually is.
", "rawMarkdown": "Can someone please point me toward where or what is the deadline to disclose the external data sources? \n\nI have seen several references to this deadline existing, but not to what the date of the deadline actually is." }, { "id": 1300758, "postDate": "2021-05-10T17:25:21.970Z", "content": "The goal is that the data sources are shared as you use them, but no later than the competition deadline.
", "rawMarkdown": "The goal is that the data sources are shared as you use them, but no later than the competition deadline. ", "votes": 2 }, { "id": 1340879, "postDate": "2021-06-08T10:08:32.677Z", "content": "Hi @kulishyurii. I'm from Russia. Would you like to join? I work in Research DS team in VTB`
", "rawMarkdown": "Hi @kulishyurii. I'm from Russia. Would you like to join? I work in Research DS team in VTB`" } ] }, { "id": 1273017, "postDate": "2021-04-14T02:39:53.630Z", "content": "I'm exploring the use of the Open Academic Graph and the KDD cup datasets to pretrain models and learn priors on distributions of citation patterns. These are open datasets available free of charge to the general public, is it rules-compliant to use these outside datasets?
", "rawMarkdown": "I'm exploring the use of the [Open Academic Graph](https://www.microsoft.com/en-us/research/project/open-academic-graph/) and the [KDD cup datasets](http://www.cs.cornell.edu/projects/kddcup/datasets.html) to pretrain models and learn priors on distributions of citation patterns. These are open datasets available free of charge to the general public, is it rules-compliant to use these outside datasets?", "votes": 2 }, { "id": 1347982, "postDate": "2021-06-13T16:05:02.650Z", "content": "https://github.com/awesomedata/awesome-public-datasets
\nhttps://www.kaggle.com/datasets (https://www.kaggle.com/stefanoleone992/complete-collection-of-kaggle-datasets)
\nhttps://www.kaggle.com/mlconsult/100000-govt-datasets-api-json-to-df
wikipedia (almost any googlable proper nouns are contained here)
\nhttps://en.wikipedia.org/wiki
almost all high public LB notebooks are using
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
\nbert MLM model
\nhttps://www.kaggle.com/tungmphung/coleridge-mlm-model
We are using SciBERT, which is trained on a scientific paper set. Paper: https://arxiv.org/abs/1903.10676 Github: https://github.com/allenai/scibert
", "rawMarkdown": "We are using SciBERT, which is trained on a scientific paper set. Paper: https://arxiv.org/abs/1903.10676 Github: https://github.com/allenai/scibert", "votes": 1 }, { "id": 1300604, "postDate": "2021-05-10T15:26:15.933Z", "content": "I referred to this page. https://www.data.gov/
", "rawMarkdown": "I referred to this page. https://www.data.gov/", "votes": 1 }, { "id": 1283809, "postDate": "2021-04-25T09:49:48.477Z", "content": "When using an external pretrained model (not developed ourselves) do we need to reference the published model or the datasets that have been used for this model?
", "rawMarkdown": "When using an external pretrained model (not developed ourselves) do we need to reference the published model or the datasets that have been used for this model? ", "votes": 1, "replies": [ { "id": 1283904, "postDate": "2021-04-25T11:43:52.607Z", "content": "@jasperdemeijer Just link to the pretrained model (participants can lookup other details).
", "rawMarkdown": "@jasperdemeijer Just link to the pretrained model (participants can lookup other details).", "votes": 1 } ] }, { "id": 1309626, "postDate": "2021-05-16T06:49:20.480Z", "content": "[https://healthdata.gov/browse?q=&sortBy=relevance]
\n(url)
What is the specified deadline?
", "rawMarkdown": "What is the specified deadline?", "votes": 2, "replies": [ { "id": 1300757, "postDate": "2021-05-10T17:24:34.457Z", "content": "Thanks for the question, I updated my original post. The goal is that people share the data reference as they use it but no later than the competition deadline.
", "rawMarkdown": "Thanks for the question, I updated my original post. The goal is that people share the data reference as they use it but no later than the competition deadline. ", "votes": 2 }, { "id": 1300781, "postDate": "2021-05-10T17:51:36.990Z", "content": "Thanks. This is different (better) than previous competitions where one had to disclose at least one week before competition end.
", "rawMarkdown": "Thanks. This is different (better) than previous competitions where one had to disclose at least one week before competition end." } ] }, { "id": 1293289, "postDate": "2021-05-04T18:08:32.320Z", "content": "what if one wants to build a dataset from scratch specifically for this project? i guess it should be allowed, but i am not sure about what would need to be disclosed. thanks, sorin.
", "rawMarkdown": "what if one wants to build a dataset from scratch specifically for this project? i guess it should be allowed, but i am not sure about what would need to be disclosed. thanks, sorin.", "votes": 2 }, { "id": 1273928, "postDate": "2021-04-14T19:15:12.903Z", "content": "I consider using articles texts available on CHORUS site
", "rawMarkdown": "I consider using articles texts available on [CHORUS site](http://search.chorusaccess.org/) ", "votes": 2 }, { "id": 1283994, "postDate": "2021-04-25T13:21:37.877Z", "content": "I am reading this ebook:
\nhttps://study.sagepub.com/richcontext/student-resources/e-book-files/epub-files
Woow , amazing
", "rawMarkdown": "Woow , amazing", "votes": -2 }, { "id": 1361979, "postDate": "2021-06-23T06:34:50.170Z", "content": "Hi @maggiemd most top-ranking teams yet to post links to their external data sources here. Please clarify as to what is your / competition hosts policy this being the case.
", "rawMarkdown": "Hi @maggiemd most top-ranking teams yet to post links to their external data sources here. Please clarify as to what is your / competition hosts policy this being the case.", "replies": [ { "id": 1364384, "postDate": "2021-06-24T23:14:54.943Z", "content": "If data was already posted then there is no requirement to re-post data that has already been identified.
", "rawMarkdown": "If data was already posted then there is no requirement to re-post data that has already been identified. " } ] }, { "id": 1361578, "postDate": "2021-06-22T23:31:10.057Z", "content": "Datasets:
\nhttps://www.kaggle.com/khubchandani/data-set-800-2
\nModel pre-trained on Kaggle platform:
\nhttps://www.kaggle.com/khubchandani/bert-mlm-v6
I used public datasets:
\nhttps://www.kaggle.com/chienhsianghung/coleridge-additional-gov-datasets-22000popular
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
I used datasets from https://www.nsf.gov/statistics/surveys.cfm and https://huggingface.co/models.
", "rawMarkdown": "I used datasets from https://www.nsf.gov/statistics/surveys.cfm and https://huggingface.co/models." }, { "id": 1361289, "postDate": "2021-06-22T17:37:19.063Z", "content": "datasets from https://www.icpsr.umich.edu/web/pages/
", "rawMarkdown": "datasets from https://www.icpsr.umich.edu/web/pages/ " }, { "id": 1361116, "postDate": "2021-06-22T15:03:25.883Z", "content": "https://www.kaggle.com/mlconsult/100000-govt-datasets-api-json-to-df
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
Dataset from RichContext Competition
\nhttps://coleridgeinitiative.org/richcontext/richcontextcompetition/
\nhttps://www.kaggle.com/ks2019/rich-context-coleridge
I used a manually-made blacklist below.
\n'USGS','GWAS','ECLS','aDAS','NCDC','NDBC','UDS','GTD','ISC','DGP','EDC','FDA','TSE','DEA','CDA','IDB','NGDC','JODC','EDM','FADN','LRD','DBDM','DMC','WSC','national longitudinal survey','education longitudinal survey','census bureau','data appendix','data file user','supplementary data','data supplement','major field of study'
https://www.kaggle.com/oviazlo/labels-from-covid-dataset-v1
\nhttps://www.kaggle.com/allen-institute-for-ai/CORD-19-research-challenge
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
https://www.data.gov/ and https://www.kaggle.com/mlconsult/bigger-govt-dataset-list
", "rawMarkdown": "https://www.data.gov/ and https://www.kaggle.com/mlconsult/bigger-govt-dataset-list" }, { "id": 1360642, "postDate": "2021-06-22T08:45:51.583Z", "content": "https://www.healthypeople.gov/
", "rawMarkdown": "https://www.healthypeople.gov/\n" }, { "id": 1360096, "postDate": "2021-06-21T19:46:15.720Z", "content": "https://www.kaggle.com/robrichter/usefull-data-ver3
", "rawMarkdown": "https://www.kaggle.com/robrichter/usefull-data-ver3" }, { "id": 1360093, "postDate": "2021-06-21T19:44:20.050Z", "content": "", "rawMarkdown": "https://paperswithcode.com/" }, { "id": 1359439, "postDate": "2021-06-21T09:40:08.630Z", "content": "We're using Flair (https://github.com/flairNLP/flair) with the pretrained embeddings mentioned in the documentation.
", "rawMarkdown": "We're using Flair (https://github.com/flairNLP/flair) with the pretrained embeddings mentioned in the documentation. " }, { "id": 1359246, "postDate": "2021-06-21T07:12:20.323Z", "content": "Public Data on Kaggle
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
I added rule-base screening to this data.
", "rawMarkdown": "Public Data on Kaggle\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list\n\nI added rule-base screening to this data." }, { "id": 1358753, "postDate": "2021-06-20T17:44:32.620Z", "content": "I used pre-trained model. I used model which i have trained offline, uploaded and used in Kaggle notebook for this competition.
", "rawMarkdown": "I used pre-trained model. I used model which i have trained offline, uploaded and used in Kaggle notebook for this competition. " }, { "id": 1358244, "postDate": "2021-06-20T09:46:20.260Z", "content": "", "rawMarkdown": "https://www.data.gov/" }, { "id": 1357653, "postDate": "2021-06-19T20:59:48.810Z", "content": "the external data I used is from https://catalog.data.gov/dataset
", "rawMarkdown": "the external data I used is from https://catalog.data.gov/dataset" }, { "id": 1355268, "postDate": "2021-06-18T07:41:51.080Z", "content": "Our team is using data from this paper: https://arxiv.org/pdf/2101.10273.pdf (TDMSci: A Specialized Corpus for Scientific Literature Entity Tagging of Tasks Datasets and Metrics by Hou et al)
", "rawMarkdown": "Our team is using data from this paper: https://arxiv.org/pdf/2101.10273.pdf (TDMSci: A Specialized Corpus for Scientific Literature Entity Tagging of Tasks Datasets and Metrics by Hou et al)" }, { "id": 1354720, "postDate": "2021-06-17T19:36:54.643Z", "content": "External data from:
\nhttps://coleridgeinitiative.org/richcontext/richcontextcompetition/
\nhttps://github.com/Coleridge-Initiative/RCDatasets/blob/master/datasets.json
Pretrained models from:
\nhttps://huggingface.co/models
\nhttps://github.com/allenai/scibert
NER Fine Tuning framework from:
\nhttps://github.com/ebanalyse/NERDA
https://data.cdc.gov/
\nhttps://www.ncdc.noaa.gov/
I just added a some extra datasets from https://figshare.com/articles/dataset/Scientific_Data_recommended_repositories_June_2015/1434640/14 to Ken Miller's commendable efforts and compiled them (rather lazily, sorry 😅) in one place here
\nPS: did not help my public score, but maybe it found a few extra hits in the private LB data! 😛
", "rawMarkdown": "I just added a some extra datasets from https://figshare.com/articles/dataset/Scientific_Data_recommended_repositories_June_2015/1434640/14 to Ken Miller's [commendable efforts ](https://www.kaggle.com/mlconsult/bigger-govt-dataset-list) and compiled them (rather lazily, sorry 😅) in one place [here](https://www.kaggle.com/slashie/coleridgeextradsets) \n\nPS: did not help my public score, but maybe it found a few extra hits in the private LB data! 😛" }, { "id": 1349557, "postDate": "2021-06-14T22:30:05.437Z", "content": "google's big bird tf code, plus the pre trained model stored in the storage bucket they provide on https://github.com/google-research/bigbird
4.6.1 version of hugginface transformers (not yet installed in Kaggle kernels)
rich context datasets from this GitHub repository https://github.com/Coleridge-Initiative/RCDatasets
custom labelling of the training set with datasets existing in .json files, but not mentioned in train.csv
https://huggingface.co/models
\nhttps://spacy.io/models
\nhttps://paperswithcode.com/datasets
https://www.kaggle.com/yk1598/479k-english-words
", "rawMarkdown": "https://www.kaggle.com/yk1598/479k-english-words" }, { "id": 1346606, "postDate": "2021-06-12T14:05:26.850Z", "content": "Pre-trained models from https://huggingface.co/models
", "rawMarkdown": "Pre-trained models from https://huggingface.co/models" }, { "id": 1346467, "postDate": "2021-06-12T12:32:26.427Z", "content": "spacy 3.0.6 from https://www.kaggle.com/dathudeptrai/spacy-dependencies and scispacy model from https://allenai.github.io/scispacy/ and the dataset from https://github.com/Coleridge-Initiative/RCDatasets, https://coleridgeinitiative.org/richcontext/richcontextcompetition/, manually whitelist/blacklist labels from training data (pseudo labels).
", "rawMarkdown": "spacy 3.0.6 from https://www.kaggle.com/dathudeptrai/spacy-dependencies and scispacy model from https://allenai.github.io/scispacy/ and the dataset from https://github.com/Coleridge-Initiative/RCDatasets, https://coleridgeinitiative.org/richcontext/richcontextcompetition/, manually whitelist/blacklist labels from training data (pseudo labels)." }, { "id": 1345209, "postDate": "2021-06-11T11:58:56.890Z", "content": "\ncountry names in text:
\nhttps://gist.github.com/kalinchernev/486393efcca01623b18d
https://nlp.johnsnowlabs.com/models
", "rawMarkdown": "https://nlp.johnsnowlabs.com/models" }, { "id": 1344585, "postDate": "2021-06-11T03:05:24.640Z", "content": "", "rawMarkdown": "https://huggingface.co/models\n\n" }, { "id": 1340669, "postDate": "2021-06-08T07:18:57.613Z", "content": "Pre-trained models from https://huggingface.co/models
", "rawMarkdown": "Pre-trained models from https://huggingface.co/models" }, { "id": 1340433, "postDate": "2021-06-07T23:24:50.587Z", "content": "We are going to use AbbreviationDetector from https://github.com/allenai/scispacy
", "rawMarkdown": "We are going to use AbbreviationDetector from [https://github.com/allenai/scispacy](url)" }, { "id": 1340062, "postDate": "2021-06-07T16:32:01.630Z", "content": "BioBERT is also worth considering. Github:https://github.com/dmis-lab/biobert
", "rawMarkdown": "BioBERT is also worth considering. Github:https://github.com/dmis-lab/biobert" }, { "id": 1312725, "postDate": "2021-05-18T07:57:09.400Z", "content": "Great topic
", "rawMarkdown": "Great topic" }, { "id": 1312384, "postDate": "2021-05-18T02:54:32.457Z", "content": "Hi @maggiemd, just to be clear, the declaration deadline is before the competition ends, not a week before, is that correct?
", "rawMarkdown": "Hi @maggiemd, just to be clear, the declaration deadline is before the competition ends, not a week before, is that correct?", "replies": [ { "id": 1340334, "postDate": "2021-06-07T19:22:17.310Z", "content": "The goal is that you share the data information \"as you use them\" and not wait until the competition deadline.
", "rawMarkdown": "The goal is that you share the data information \"as you use them\" and not wait until the competition deadline. ", "votes": 1 } ] }, { "id": 1388736, "postDate": "2021-07-15T07:23:51.803Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1361553, "postDate": "2021-06-22T23:05:01.540Z", "rawMarkdown": "", "isDeleted": true }, { "id": 1355830, "postDate": "2021-06-18T14:59:38.213Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1276691, "author_name": "Hannes Öhler", "author_url": "", "post_date": "2021-04-17T21:05:53.427000", "content": "There has been another post about it, but just to be sure: I use data from the RichContextCompetition.
", "votes": 9, "replies": [] }, { "id": 1272687, "author_name": "Julia Lane", "author_url": "", "post_date": "2021-04-13T17:25:29.800000", "content": "@poedator Thanks for your question - I checked with @maggiemd and @philculliton about the rules - and they came up with this great response. Thanks so much!
", "votes": 3, "replies": [] }, { "id": 1295135, "author_name": "KulishYurii", "author_url": "", "post_date": "2021-05-06T08:21:36.760000", "content": "I see that some participants are using external databases like this one https://www.kaggle.com/mlconsult/score-57ish-with-additional-govt-datasets
\nIs it allowed?
Hi @kulishyurii
\nFrom the code requirements page on the the competition overview:
\nThis is a Code Competition
\nSubmissions to this competition must be made through Notebooks. In order for the \"Submit\" button to be active after a commit, the following conditions must be met:
\nCPU Notebook <= 9 hours run-time\nGPU Notebook <= 9 hours run-time\nInternet access disabled\nFreely & publicly available external data is allowed, including pre-trained models\nSubmission file must be named submission.csv\nTo be eligible for prizes, one need to disclose the external data sources used before the deadline specified. For those who don't care about prizes, well, I don't think there's any way to force them to disclose it.
", "votes": 1, "replies": [] }, { "id": 1299875, "author_name": "mic mylin", "author_url": "", "post_date": "2021-05-10T05:32:35.843000", "content": "Can someone please point me toward where or what is the deadline to disclose the external data sources?
\nI have seen several references to this deadline existing, but not to what the date of the deadline actually is.
", "votes": 0, "replies": [] }, { "id": 1300758, "author_name": "Maggie", "author_url": "", "post_date": "2021-05-10T17:25:21.970000", "content": "The goal is that the data sources are shared as you use them, but no later than the competition deadline.
", "votes": 2, "replies": [] }, { "id": 1340879, "author_name": "VDA", "author_url": "", "post_date": "2021-06-08T10:08:32.677000", "content": "Hi @kulishyurii. I'm from Russia. Would you like to join? I work in Research DS team in VTB`
", "votes": 0, "replies": [] } ] }, { "id": 1273017, "author_name": "sfsf", "author_url": "", "post_date": "2021-04-14T02:39:53.630000", "content": "I'm exploring the use of the Open Academic Graph and the KDD cup datasets to pretrain models and learn priors on distributions of citation patterns. These are open datasets available free of charge to the general public, is it rules-compliant to use these outside datasets?
", "votes": 2, "replies": [] }, { "id": 1347982, "author_name": "Khoi Nguyen", "author_url": "", "post_date": "2021-06-13T16:05:02.650000", "content": "https://github.com/awesomedata/awesome-public-datasets
\nhttps://www.kaggle.com/datasets (https://www.kaggle.com/stefanoleone992/complete-collection-of-kaggle-datasets)
\nhttps://www.kaggle.com/mlconsult/100000-govt-datasets-api-json-to-df
wikipedia (almost any googlable proper nouns are contained here)
\nhttps://en.wikipedia.org/wiki
almost all high public LB notebooks are using
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
\nbert MLM model
\nhttps://www.kaggle.com/tungmphung/coleridge-mlm-model
We are using SciBERT, which is trained on a scientific paper set. Paper: https://arxiv.org/abs/1903.10676 Github: https://github.com/allenai/scibert
", "votes": 1, "replies": [] }, { "id": 1300604, "author_name": "ochaaaaaaan", "author_url": "", "post_date": "2021-05-10T15:26:15.933000", "content": "I referred to this page. https://www.data.gov/
", "votes": 1, "replies": [] }, { "id": 1283809, "author_name": "Jasper de Meijer", "author_url": "", "post_date": "2021-04-25T09:49:48.477000", "content": "When using an external pretrained model (not developed ourselves) do we need to reference the published model or the datasets that have been used for this model?
", "votes": 1, "replies": [ { "id": 1283904, "author_name": "Sirish Somanchi", "author_url": "", "post_date": "2021-04-25T11:43:52.607000", "content": "@jasperdemeijer Just link to the pretrained model (participants can lookup other details).
", "votes": 1, "replies": [] } ] }, { "id": 1309626, "author_name": "GUNER", "author_url": "", "post_date": "2021-05-16T06:49:20.480000", "content": "[https://healthdata.gov/browse?q=&sortBy=relevance]
\n(url)
What is the specified deadline?
", "votes": 2, "replies": [ { "id": 1300757, "author_name": "Maggie", "author_url": "", "post_date": "2021-05-10T17:24:34.457000", "content": "Thanks for the question, I updated my original post. The goal is that people share the data reference as they use it but no later than the competition deadline.
", "votes": 2, "replies": [] }, { "id": 1300781, "author_name": "CPMP", "author_url": "", "post_date": "2021-05-10T17:51:36.990000", "content": "Thanks. This is different (better) than previous competitions where one had to disclose at least one week before competition end.
", "votes": 0, "replies": [] } ] }, { "id": 1293289, "author_name": "flokilum", "author_url": "", "post_date": "2021-05-04T18:08:32.320000", "content": "what if one wants to build a dataset from scratch specifically for this project? i guess it should be allowed, but i am not sure about what would need to be disclosed. thanks, sorin.
", "votes": 2, "replies": [] }, { "id": 1273928, "author_name": "Poe Dator", "author_url": "", "post_date": "2021-04-14T19:15:12.903000", "content": "I consider using articles texts available on CHORUS site
", "votes": 2, "replies": [] }, { "id": 1283994, "author_name": "林湧森 (Dyson Lin)", "author_url": "", "post_date": "2021-04-25T13:21:37.877000", "content": "I am reading this ebook:
\nhttps://study.sagepub.com/richcontext/student-resources/e-book-files/epub-files
Woow , amazing
", "votes": -2, "replies": [] }, { "id": 1361979, "author_name": "Khubchandani", "author_url": "", "post_date": "2021-06-23T06:34:50.170000", "content": "Hi @maggiemd most top-ranking teams yet to post links to their external data sources here. Please clarify as to what is your / competition hosts policy this being the case.
", "votes": 0, "replies": [ { "id": 1364384, "author_name": "Maggie", "author_url": "", "post_date": "2021-06-24T23:14:54.943000", "content": "If data was already posted then there is no requirement to re-post data that has already been identified.
", "votes": 0, "replies": [] } ] }, { "id": 1361578, "author_name": "Khubchandani", "author_url": "", "post_date": "2021-06-22T23:31:10.057000", "content": "Datasets:
\nhttps://www.kaggle.com/khubchandani/data-set-800-2
\nModel pre-trained on Kaggle platform:
\nhttps://www.kaggle.com/khubchandani/bert-mlm-v6
I used public datasets:
\nhttps://www.kaggle.com/chienhsianghung/coleridge-additional-gov-datasets-22000popular
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
I used datasets from https://www.nsf.gov/statistics/surveys.cfm and https://huggingface.co/models.
", "votes": 0, "replies": [] }, { "id": 1361289, "author_name": "zagros ", "author_url": "", "post_date": "2021-06-22T17:37:19.063000", "content": "datasets from https://www.icpsr.umich.edu/web/pages/
", "votes": 0, "replies": [] }, { "id": 1361116, "author_name": "sneezy_giraffe", "author_url": "", "post_date": "2021-06-22T15:03:25.883000", "content": "https://www.kaggle.com/mlconsult/100000-govt-datasets-api-json-to-df
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
Dataset from RichContext Competition
\nhttps://coleridgeinitiative.org/richcontext/richcontextcompetition/
\nhttps://www.kaggle.com/ks2019/rich-context-coleridge
I used a manually-made blacklist below.
\n'USGS','GWAS','ECLS','aDAS','NCDC','NDBC','UDS','GTD','ISC','DGP','EDC','FDA','TSE','DEA','CDA','IDB','NGDC','JODC','EDM','FADN','LRD','DBDM','DMC','WSC','national longitudinal survey','education longitudinal survey','census bureau','data appendix','data file user','supplementary data','data supplement','major field of study'
https://www.kaggle.com/oviazlo/labels-from-covid-dataset-v1
\nhttps://www.kaggle.com/allen-institute-for-ai/CORD-19-research-challenge
\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list
https://www.data.gov/ and https://www.kaggle.com/mlconsult/bigger-govt-dataset-list
", "votes": 0, "replies": [] }, { "id": 1360642, "author_name": "Peter KK", "author_url": "", "post_date": "2021-06-22T08:45:51.583000", "content": "https://www.healthypeople.gov/
", "votes": 0, "replies": [] }, { "id": 1360096, "author_name": "Robert", "author_url": "", "post_date": "2021-06-21T19:46:15.720000", "content": "https://www.kaggle.com/robrichter/usefull-data-ver3
", "votes": 0, "replies": [] }, { "id": 1360093, "author_name": "ymgw55", "author_url": "", "post_date": "2021-06-21T19:44:20.050000", "content": "", "votes": 0, "replies": [] }, { "id": 1359439, "author_name": "Lea Demelius", "author_url": "", "post_date": "2021-06-21T09:40:08.630000", "content": "We're using Flair (https://github.com/flairNLP/flair) with the pretrained embeddings mentioned in the documentation.
", "votes": 0, "replies": [] }, { "id": 1359246, "author_name": "", "author_url": "", "post_date": "2021-06-21T07:12:20.323000", "content": "", "votes": 0, "replies": [] }, { "id": 1358753, "author_name": "", "author_url": "", "post_date": "2021-06-20T17:44:32.620000", "content": "", "votes": 0, "replies": [] }, { "id": 1358244, "author_name": "", "author_url": "", "post_date": "2021-06-20T09:46:20.260000", "content": "", "votes": 0, "replies": [] }, { "id": 1357653, "author_name": "", "author_url": "", "post_date": "2021-06-19T20:59:48.810000", "content": "", "votes": 0, "replies": [] }, { "id": 1355268, "author_name": "", "author_url": "", "post_date": "2021-06-18T07:41:51.080000", "content": "", "votes": 0, "replies": [] }, { "id": 1354720, "author_name": "", "author_url": "", "post_date": "2021-06-17T19:36:54.643000", "content": "", "votes": 0, "replies": [] }, { "id": 1353307, "author_name": "", "author_url": "", "post_date": "2021-06-17T03:36:50.150000", "content": "", "votes": 0, "replies": [] }, { "id": 1351576, "author_name": "", "author_url": "", "post_date": "2021-06-16T12:17:34.817000", "content": "", "votes": 0, "replies": [] }, { "id": 1349557, "author_name": "", "author_url": "", "post_date": "2021-06-14T22:30:05.437000", "content": "", "votes": 0, "replies": [] }, { "id": 1348273, "author_name": "", "author_url": "", "post_date": "2021-06-13T22:32:26.893000", "content": "", "votes": 0, "replies": [] }, { "id": 1347755, "author_name": "", "author_url": "", "post_date": "2021-06-13T13:25:46.477000", "content": "", "votes": 0, "replies": [] }, { "id": 1346708, "author_name": "", "author_url": "", "post_date": "2021-06-12T14:57:29.037000", "content": "", "votes": 0, "replies": [] }, { "id": 1346606, "author_name": "", "author_url": "", "post_date": "2021-06-12T14:05:26.850000", "content": "", "votes": 0, "replies": [] }, { "id": 1346467, "author_name": "", "author_url": "", "post_date": "2021-06-12T12:32:26.427000", "content": "", "votes": 0, "replies": [] }, { "id": 1345209, "author_name": "", "author_url": "", "post_date": "2021-06-11T11:58:56.890000", "content": "", "votes": 0, "replies": [] }, { "id": 1344865, "author_name": "", "author_url": "", "post_date": "2021-06-11T07:18:28.877000", "content": "", "votes": 0, "replies": [] }, { "id": 1344585, "author_name": "", "author_url": "", "post_date": "2021-06-11T03:05:24.640000", "content": "", "votes": 0, "replies": [] }, { "id": 1340669, "author_name": "", "author_url": "", "post_date": "2021-06-08T07:18:57.613000", "content": "", "votes": 0, "replies": [] }, { "id": 1340433, "author_name": "", "author_url": "", "post_date": "2021-06-07T23:24:50.587000", "content": "", "votes": 0, "replies": [] }, { "id": 1340062, "author_name": "", "author_url": "", "post_date": "2021-06-07T16:32:01.630000", "content": "", "votes": 0, "replies": [] }, { "id": 1323073, "author_name": "", "author_url": "", "post_date": "2021-05-26T00:28:25.770000", "content": "", "votes": 0, "replies": [] }, { "id": 1312725, "author_name": "", "author_url": "", "post_date": "2021-05-18T07:57:09.400000", "content": "", "votes": 0, "replies": [] }, { "id": 1312384, "author_name": "", "author_url": "", "post_date": "2021-05-18T02:54:32.457000", "content": "", "votes": 0, "replies": [ { "id": 1340334, "author_name": "", "author_url": "", "post_date": "2021-06-07T19:22:17.310000", "content": "", "votes": 1, "replies": [] } ] }, { "id": 1388736, "author_name": "", "author_url": "", "post_date": "2021-07-15T07:23:51.803000", "content": "", "votes": 0, "replies": [] }, { "id": 1361553, "author_name": "", "author_url": "", "post_date": "2021-06-22T23:05:01.540000", "content": "", "votes": 0, "replies": [] }, { "id": 1355830, "author_name": "", "author_url": "", "post_date": "2021-06-18T14:59:38.213000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1272656": "For transparency, the host requests that participants post links to external data sources here as you use them, no later than the **competition** deadline specified. Once it has been posted, you do not need to post it again.", "1276691": "There has been another [post](https://www.kaggle.com/c/coleridgeinitiative-show-us-the-data/discussion/231374) about it, but just to be sure: I use data from the [RichContextCompetition](https://coleridgeinitiative.org/richcontext/richcontextcompetition/).", "1272687": "@poedator Thanks for your question - I checked with @maggiemd and @philculliton about the rules - and they came up with this great response. Thanks so much!", "1295135": "I see that some participants are using external databases like this one https://www.kaggle.com/mlconsult/score-57ish-with-additional-govt-datasets\nIs it allowed?", "1273017": "I'm exploring the use of the [Open Academic Graph](https://www.microsoft.com/en-us/research/project/open-academic-graph/) and the [KDD cup datasets](http://www.cs.cornell.edu/projects/kddcup/datasets.html) to pretrain models and learn priors on distributions of citation patterns. These are open datasets available free of charge to the general public, is it rules-compliant to use these outside datasets?", "1347982": "https://github.com/awesomedata/awesome-public-datasets\nhttps://www.kaggle.com/datasets (https://www.kaggle.com/stefanoleone992/complete-collection-of-kaggle-datasets)\nhttps://www.kaggle.com/mlconsult/100000-govt-datasets-api-json-to-df", "1327270": "wikipedia (almost any googlable proper nouns are contained here)\nhttps://en.wikipedia.org/wiki", "1311420": "We are using SciBERT, which is trained on a scientific paper set. Paper: https://arxiv.org/abs/1903.10676 Github: https://github.com/allenai/scibert", "1300604": "I referred to this page. https://www.data.gov/", "1283809": "When using an external pretrained model (not developed ourselves) do we need to reference the published model or the datasets that have been used for this model? ", "1309626": "[https://healthdata.gov/browse?q=&sortBy=relevance]\n(url)", "1300174": "What is the specified deadline?", "1293289": "what if one wants to build a dataset from scratch specifically for this project? i guess it should be allowed, but i am not sure about what would need to be disclosed. thanks, sorin.", "1273928": "I consider using articles texts available on [CHORUS site](http://search.chorusaccess.org/) ", "1283994": "I am reading this ebook:\nhttps://study.sagepub.com/richcontext/student-resources/e-book-files/epub-files", "1285253": "Woow , amazing", "1361979": "Hi @maggiemd most top-ranking teams yet to post links to their external data sources here. Please clarify as to what is your / competition hosts policy this being the case.", "1361578": "Datasets:\nhttps://www.kaggle.com/khubchandani/data-set-800-2\nModel pre-trained on Kaggle platform:\nhttps://www.kaggle.com/khubchandani/bert-mlm-v6", "1361563": "I used public datasets:\nhttps://www.kaggle.com/chienhsianghung/coleridge-additional-gov-datasets-22000popular\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list ", "1361545": "I used datasets from https://www.nsf.gov/statistics/surveys.cfm and https://huggingface.co/models.", "1361289": "datasets from https://www.icpsr.umich.edu/web/pages/ ", "1361116": "https://www.kaggle.com/mlconsult/100000-govt-datasets-api-json-to-df\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list", "1361065": "Dataset from RichContext Competition\nhttps://coleridgeinitiative.org/richcontext/richcontextcompetition/\nhttps://www.kaggle.com/ks2019/rich-context-coleridge", "1360985": "I used a manually-made blacklist below.\n'USGS','GWAS','ECLS','aDAS','NCDC','NDBC','UDS','GTD','ISC','DGP','EDC','FDA','TSE','DEA','CDA','IDB','NGDC','JODC','EDM','FADN','LRD','DBDM','DMC','WSC','national longitudinal survey','education longitudinal survey','census bureau','data appendix','data file user','supplementary data','data supplement','major field of study'", "1360960": "https://www.kaggle.com/oviazlo/labels-from-covid-dataset-v1\nhttps://www.kaggle.com/allen-institute-for-ai/CORD-19-research-challenge\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list", "1360887": "https://www.data.gov/ and https://www.kaggle.com/mlconsult/bigger-govt-dataset-list", "1360642": "https://www.healthypeople.gov/\n", "1360096": "https://www.kaggle.com/robrichter/usefull-data-ver3", "1360093": "https://paperswithcode.com/", "1359439": "We're using Flair (https://github.com/flairNLP/flair) with the pretrained embeddings mentioned in the documentation. ", "1359246": "Public Data on Kaggle\nhttps://www.kaggle.com/mlconsult/bigger-govt-dataset-list\n\nI added rule-base screening to this data.", "1358753": "I used pre-trained model. I used model which i have trained offline, uploaded and used in Kaggle notebook for this competition. ", "1358244": "https://www.data.gov/", "1357653": "the external data I used is from https://catalog.data.gov/dataset", "1355268": "Our team is using data from this paper: https://arxiv.org/pdf/2101.10273.pdf (TDMSci: A Specialized Corpus for Scientific Literature Entity Tagging of Tasks Datasets and Metrics by Hou et al)", "1354720": "External data from:\nhttps://coleridgeinitiative.org/richcontext/richcontextcompetition/\nhttps://github.com/Coleridge-Initiative/RCDatasets/blob/master/datasets.json\n\nPretrained models from:\nhttps://huggingface.co/models\nhttps://github.com/allenai/scibert\n\nNER Fine Tuning framework from:\nhttps://github.com/ebanalyse/NERDA", "1353307": "https://data.cdc.gov/ \nhttps://www.ncdc.noaa.gov/ \n", "1351576": "I just added a some extra datasets from https://figshare.com/articles/dataset/Scientific_Data_recommended_repositories_June_2015/1434640/14 to Ken Miller's [commendable efforts ](https://www.kaggle.com/mlconsult/bigger-govt-dataset-list) and compiled them (rather lazily, sorry 😅) in one place [here](https://www.kaggle.com/slashie/coleridgeextradsets) \n\nPS: did not help my public score, but maybe it found a few extra hits in the private LB data! 😛", "1349557": "1. google's big bird tf code, plus the pre trained model stored in the storage bucket they provide on https://github.com/google-research/bigbird \n\n2. 4.6.1 version of hugginface transformers (not yet installed in Kaggle kernels)\n\n3. rich context datasets from this GitHub repository https://github.com/Coleridge-Initiative/RCDatasets\n\n4. custom labelling of the training set with datasets existing in .json files, but not mentioned in train.csv", "1348273": "https://huggingface.co/models\nhttps://spacy.io/models\nhttps://paperswithcode.com/datasets ", "1347755": "https://huggingface.co/models", "1346708": "https://www.kaggle.com/yk1598/479k-english-words", "1346606": "Pre-trained models from https://huggingface.co/models", "1346467": "spacy 3.0.6 from https://www.kaggle.com/dathudeptrai/spacy-dependencies and scispacy model from https://allenai.github.io/scispacy/ and the dataset from https://github.com/Coleridge-Initiative/RCDatasets, https://coleridgeinitiative.org/richcontext/richcontextcompetition/, manually whitelist/blacklist labels from training data (pseudo labels).", "1345209": "[https://huggingface.co/models](url)\n\ncountry names in text:\n[https://gist.github.com/kalinchernev/486393efcca01623b18d](url)", "1344865": "https://nlp.johnsnowlabs.com/models", "1344585": "https://huggingface.co/models\n\n", "1340669": "Pre-trained models from https://huggingface.co/models", "1340433": "We are going to use AbbreviationDetector from [https://github.com/allenai/scispacy](url)", "1340062": "1. Corpus available under nltk\n2. https://www.usa.gov/\n3. https://unstats.un.org/home/\n4. https://allenai.github.io/scispacy/\n", "1323073": "BioBERT is also worth considering. Github:https://github.com/dmis-lab/biobert", "1312725": "Great topic", "1312384": "Hi @maggiemd, just to be clear, the declaration deadline is before the competition ends, not a week before, is that correct?", "1388736": "", "1361553": "", "1355830": "" } }