Semantic features or concepts play an important role in understanding the meaning of a word in a given context. The features depend upon the dataset that provides context and has a sparse representation of the word as a concept. These semantic features can be either handpicked or generated by analyzing its co-occurrence with the other words. In this paper, we predict the semantic features of a concept(noun) type. We use the features of the McRae dataset for a given concept and create a mapping of those features to the embeddings obtained from the BERT model for that specific concept. Our goal is to decode the meaning of the BERT embeddings using the feature vectors of the McRae dataset.
-
Notifications
You must be signed in to change notification settings - Fork 0
Extract semantic features to understand the meaning of words using a 2-layer NN with BERT embeddings and McRae features.
License
div5yesh/semantic-feature-extraction
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
About
Extract semantic features to understand the meaning of words using a 2-layer NN with BERT embeddings and McRae features.
Topics
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published