Skip to content

Mappings Comparison

Roque López edited this page Apr 11, 2017 · 34 revisions

Context

KNEWS extracts its frame types starting from the output of Boxer and UKB (or alternatively Babelfy). WordNet synsets from UKB is aligned to the event predicate symbol from Boxer, then the WordNet synsets is mapped to a FrameNet frame using a pre-existent mapping [file](https://github.com/valeriobasile/learningbyreading/blob/master/resources/framenet-wordnet-map.txt). However, this mapping is still small. Recently, STLab developed better mappings within Framester and made them [available](https://github.com/framester/Framester/tree/master/fn-babelnet-mappings).

## Objectives In this experiment we evaluated the t was performed some experiments in order to evaluate the calculate some statistics such as number of frame instances, frame types, role types and element types

Results

Knews Mappings

Item All values Unique values
Instances 114517 114517
Types 97154 2274
Elements 154422 3830
Roles 154422 222

Framester Mappings

Item All values Unique values
Instances 114212 114212
Types 95347 2013
Elements 154095 3824
Roles 154095 222

Resources

Item WordNet BabelNet
Frames 881 870
Synsets 9080 8169

Clone this wiki locally