.evaluate

class tieval.evaluate.Span(start: int, end: int)

Bases: object

An annotation span build to check if two entities overlap.

.metrics module

tieval.evaluate.metrics.temporal_awareness(prediction: Set[TLink], annotation: Set[TLink]) → float

Compute the temporal awareness of a system.

Temporal awareness is a f1 measure that takes into account the temporal closure of a system. For more information refer to the original paper.

Parameters:

• prediction (Set[TLink]) – The TLinks predicted by the system.

• annotation (Set[TLink]) – The reference TLinks.

tieval.evaluate.metrics.temporal_precision(prediction: Set[TLink], annotation: Set[TLink]) → Tuple[int, int]

Compute precision by taking into account the temporal closure of the annotations.

Parameters:

• prediction (Set[TLink]) – The TLinks predicted by the system.

• annotation (Set[TLink]) – The reference TLinks.

tieval.evaluate.metrics.temporal_recall(prediction: Set[TLink], annotation: Set[TLink]) → Tuple[int, int]

Compute recall by taking into account the temporal closure of the predictions.

Parameters:

• prediction (Set[TLink]) – The TLinks predicted by the system.

• annotation (Set[TLink]) – The reference TLinks.