Table 1.
Extensive successful previous work on negation detection in clinical text.
Table 2.
Characteristics of four corpora with negation annotations.
Table 3.
In the MiPACQ and SHARP corpora, the named entities (NEs) are annotated with different semantic groups which occur with different frequencies (left columns).
Figure 1.
The SHARPn Polarity Module is an Attribute Discovery algorithm. Training and evaluations use gold standard NEs (skip NER).
Figure 2.
Significance bands of model performance for each test corpus.
These are labeled with successive letters from right to left in Table 4.
Table 4.
Performance (F1 score) in practical negation detection situations.
Figure 3.
Learning curve for i2b2 training data on various corpora.
For each proportion of the i2b2 corpus (x axis), the reported F-score (y axis) is an average of 5 randomly sampled runs.
Table 5.
Average F-score with and without frustratingly easy domain adaptation (FEDA).
Figure 4.
The effect of named entity length (in number of words) on performance for each of 6 training configurations.
SHARP, MiPACQ, and i2b2 test sets are used for evaluation.
Figure 5.
The effect of named entity semantic group on the F-score of 6 models.
SHARP, MiPACQ, and i2b2 test sets are used for evaluation.
Table 6.
Top negation context features in a multi-corpus model, by chi-square value; and feature rank in domain-specific models.