There is a newer version of this record available.

Dataset Open Access

MESINESP2 Corpora: Annotated data for medical semantic indexing in Spanish

Gasco, Luis; Krallinger, Martin

Citation Style Language JSON Export

  "publisher": "Zenodo", 
  "DOI": "10.5281/zenodo.4612275", 
  "language": "spa", 
  "title": "MESINESP2 Corpora: Annotated data for medical semantic indexing in Spanish", 
  "issued": {
    "date-parts": [
  "abstract": "<p>Annotated corpora for MESINESP2 shared-task (Spanish BioASQ track, see <a href=\"\"></a>). BioASQ 2021 will be held at CLEF 2021 (scheduled in Bucharest, Romania in September)&nbsp;<a href=\"\">;</a></p>\n\n<p><strong>Introduction:</strong><br>\nThese corpora contain the data for each of the sub-tracks of MESINESP2 shared-task:</p>\n\n<ul>\n\t<li><strong>Track 1- Medical indexing</strong>: &nbsp;\n\n\t<ul>\n\t\t<li><em><strong>Training set: </strong></em>It contains all spanish records from LILACS and IBECS databases at the Virtual Health Library (VHL) with non-empty abstract written in Spanish.&nbsp;We have filtered out empty abstracts and non-Spanish abstracts.&nbsp;&nbsp;We have built the training dataset with the data crawled on 01/29/2021. This means that the data is a snapshot of that moment and that may change over time since LILACS and IBECS usually add or modify indexes after the first inclusion in the database.&nbsp;We distribute two different datasets:\n\n\t\t<ul>\n\t\t\t<li><strong>Articles training set:&nbsp;</strong>This corpus contains the set of 237574 Spanish scientific papers in VHL that have at least one DeCS code assigned to them.</li>\n\t\t\t<li><strong>Full training set</strong>: This corpus contains the whole set of 249474 Spanish documents from VHL that have at leas one DeCS code assigned to them.</li>\n\t\t</ul>\n\t\t</li>\n\t\t<li><strong>Development set:&nbsp;</strong>We provide a development set manually indexed by expert annotators. This dataset includes 1065 articles annotated with DeCS by three expert indexers in this controlled vocabulary. The articles were initially indexed by 7 annotators, after analyzing the Inter-Annotator Agreement among their annotations we decided to select the 3 best ones, considering their annotations the valid ones to build the test set. From those 1065 records:\n\t\t<ul>\n\t\t\t<li>213 articles were annotated by more than one annotator. We have selected de union between annotations.</li>\n\t\t\t<li>852 articles were annotated by only one of the three selected annotators with better performance.</li>\n\t\t</ul>\n\t\t</li>\n\t\t<li><strong>Test set:&nbsp;</strong>To be published&nbsp;</li>\n\t</ul>\n\t</li>\n\t<li><strong>Track 2- Clinical trials</strong>: &nbsp;<br>\n\t<ul>\n\t\t<li><strong>Training set:&nbsp;</strong>The training dataset contains records from&nbsp;<a href=\"\">Registro Espa&ntilde;ol de Estudios Cl&iacute;nicos (REEC)</a>. REEC doesn&#39;t&nbsp;provide documents with the structure title/abstract needed in BioASQ, for that reason we have built artificial abstracts based on the content available in the data crawled using the REEC&nbsp;<a href=\"\">API</a>.&nbsp;Clinical trials are not indexed with DeCS terminology, we have used as training data a set of 3592 clinical trials that were automatically annotated in the first edition of MESINESP and that were published as a&nbsp;<a href=\"\">Silver Standard outcome</a>. Because the performance of the models used by the participants was variable, we have only selected predictions from runs with a MiF higher than 0.30, which corresponds with the submission of the best three teams. We have selected the union of all codes assigned by those team.</li>\n\t\t<li><strong>Development set: </strong>We provide a development set manually indexed by expert annotators. This dataset includes 147 clinical trials annotated with DeCS by seven expert indexers in this controlled vocabulary.</li>\n\t</ul>\n\t</li>\n\t<li><strong>Track 3- Patents:&nbsp;</strong>To be published</li>\n</ul>\n\n<p><strong>Files structure:</strong></p>\n\n<p><strong></strong> contains the corpora generated for the shared task. Content:</p>\n\n<ul>\n\t<li>Subtrack1:\n\t<ul>\n\t\t<li>Train\n\t\t<ul>\n\t\t\t<li>training_set_track1_all.json: Full training set for sub-track 1.</li>\n\t\t\t<li>training_set_track1_only_articles.json:&nbsp;Articles training set for sub-track 1.</li>\n\t\t</ul>\n\t\t</li>\n\t\t<li>Test\n\t\t<ul>\n\t\t\t<li>development_set_subtrack1.json: Manually annotated&nbsp;development set for sub-track 1.</li>\n\t\t</ul>\n\t\t</li>\n\t</ul>\n\t</li>\n\t<li>Subtrack2:\n\t<ul>\n\t\t<li>Train\n\t\t<ul>\n\t\t\t<li>training_set_subtrack2.json: Training set for sub-track 2.</li>\n\t\t</ul>\n\t\t</li>\n\t\t<li>Test\n\t\t<ul>\n\t\t\t<li>development_set_subtrack2.json:&nbsp;Manually annotated&nbsp;development set for sub-track 2.</li>\n\t\t</ul>\n\t\t</li>\n\t</ul>\n\t</li>\n\t<li>Subtrack3: This folder is empty. Data for sub-track&nbsp;3 will be published soon.</li>\n</ul>\n\n<p>&nbsp;</p>\n\n<p><strong>DeCS2020.tsv</strong> contains a DeCS table with the following structure:</p>\n\n<ul>\n\t<li>DeCS code</li>\n\t<li>Preferred descriptor (the preferred label in the <code>Latin Spanish Decs&nbsp;</code>2020 set)</li>\n\t<li>List of synonyms (the descriptors and synonyms from&nbsp;<code>Latin Spanish DeCS 2020, separate by pipes)</code></li>\n</ul>\n\n<p>&nbsp;</p>\n\n<p><strong>DeCS2020.obo&nbsp;</strong>contains the *.obo file with the hierarchical relationships between DeCS descriptors.</p>\n\n<p>&nbsp;</p>\n\n<p>&nbsp;</p>\n\n<p>For further information, please visit&nbsp;<a href=\"\"></a>&nbsp;or email us at</p>", 
  "author": [
      "family": "Gasco, Luis"
      "family": "Krallinger, Martin"
  "note": "Funded by the Plan de Impulso de las Tecnolog\u00edas del Lenguaje (Plan TL).", 
  "version": "1.0.0", 
  "type": "dataset", 
  "id": "4612275"
All versions This version
Views 2,166131
Downloads 48135
Data volume 37.4 GB4.1 GB
Unique views 1,615108
Unique downloads 18715


Cite as