Information extraction¶
Information Extraction (IE) tasks employ a wide array of data structures. We support the following, all of which can be
imported from metametric.structures.ie
:
Mention
: A span of text, defined by its starting and ending indices (inclusive or exclusive) in a passage of text.Relation
: A typed binary relation between twoMention
s.RelationSet
: A collection ofRelation
s.Trigger
: A typed, event-denotingMention
.Argument
: A typed, entity-denotingMention
that satisfies a certainrole
in an event.Entity
: An entity, represented as a collection ofMention
s that refer to it.EntitySet
: A collection ofEntity
s.Membership
: A membership relation between a particularMention
and theEntity
it refers to.Event
: A complete event, represented by a particularTrigger
together with all of itsArgument
s.EventSet
: A collection ofEvent
s.
One can define a wide array of metrics based on these data structures.
Coreference Resolution¶
We support three of the most widely used metrics for coreference resolution, including
\(\text{MUC}\) (muc
; paper),
\(B^3\) [b_cubed_precision
, b_cubed_recall
; paper)
and \(\text{CEAF}_{\phi_4}\) (ceaf_phi4
; paper), as well as a metric
suite (coref_suite
) that includes all of these, plus the commonly reported average of all three. These metrics can be
imported from metametric.metrics.coref
.