LASIGE researchers André Neves, André Lamúrias and Francisco Couto participated, as a team, in the first edition of the BioASQ Task on Medical Semantic Indexing in Spanish (BioASQ MESINESP).
BioASQ organizes international challenges on biomedical semantic indexing and question answering (QA). The challenges include tasks relevant to hierarchical text classification, machine learning, information retrieval, QA from texts and structured data, multi-document summarization and many other areas. The MESINESP task goal was to automatically index biomedical Spanish articles with structured medical vocabulary (DeCS terms).
The full results for that task are available here. The team submitted four system variants, namely LasigeBioTM TXMC F1, LasigeBioTM TXMC P and X-BERT BioASQ, and X-BERT BioASQ F1. The latter achieved the second best value for the main evaluation measure (label-based micro F-measure), and, in general, their systems were placed on top positions in other evaluation metrics.
André Lamúrias and André Neves are currently developing work on the FCT financed project DeST, lead by Francisco Couto.
Updated on 06/07/2020 to correct the info of the results of the challenge, after the MESINESP organization communicated that it had committed errors in the evaluation process. The LASIGE team was originally considered the winner team.