| | Data source | Literature | Year | Size | Label number | Labeling method | NER | Method |
| | Supervised learning | | EMR | Aramaki et al. [10] | 2010 | 3012 notes | A, O (2) | H | CRF | Pattern-based | | Sohn et al. [11] | 2011 | 237 notes | A, O (2) | H | cTAKES | Pattern-based, DT C4.5 | | Henriksson et al. [26] | 2015 | 400 notes | A, I, O (3) | H | CRF | Word embedding, RF | | Casillas et al. [12] | 2016 | n/a | A, O (2) | H | FreeLing-Med | Pattern-based, SVM, RF | | Literature | Peng et al. [16] | 2016 | 18,410 abstracts | A, O (2) | H, DS | Dictionary, tmChem, DNorm | Feature-based, SVM | | Social media | Segura-Bedmar et al. [33] | 2015 | 84,000 messages | A, I (2) | DS | GATE | Shallow linguistic kernel, distant supervision | | Nikfarjam et al. [17] | 2015 | 8800 blog sentences, 3200 tweets | A, I, O (3) | H | CRF | Word embedding, CRF | | Jenhani et al. [18] | 2016 | 80,000 tweets | A, O (2) | R, ODIN | Dictionary, Stanford CoreNLP | Rule-base, feature-based, DT, SVM, LR, NB | | Liu et al. [34] | 2016 | 1800 blog sentences, 500 tweets | A, O (2) | H | MetaMap | Feature-based, tree kernel-based, ensemble method |
| | Semisupervised learning | | EMR | Taewijit and Theeramunkong [13] | 2016 | 1.5 M sentences | A, I (2) | DS | MetaMap | Distant supervision, OpenIE [35], pattern-based | | Literature | Kang et al. [36] | 2014 | 1644 abstracts | A, O (2) | H | Peregrine | Hierarchical graph-based, shortest path | | Social media | Liu and Chen [37] | 2015 | 400 sentences | A, I, O (3) | H | MetaMap | Dependency tree, TSVM [38] |
| | Unsupervised learning | | EMR | Wang et al. [39] | 2009 | 25,074 notes | None | None | MedLEE | Co-occurrence | | Literature | Xu and Wang [14] | 2014 | 119 M sentences | None | None | Parse tree | Pattern-based, ranking | | Social media | Feldman et al. [15] | 2015 | 0.1~1 M messages | None | None | Dictionary, pattern | HPSG-based parser, postprocessing of relation merging |
|
|
