The goal of this project is to implement a bi-directional LSTM functional neural network that can classify named entities. the dataset has been extracted from GMB corpus and it is structured in a way that makes it easier to train a model for named entity recognition or part of speech tagging.
Included entities:
- geo = Geographical Entity
- org = Organization
- per = Person
- gpe = Geopolitical Entity
- tim = Time indicator
- art = Artifact
- eve = Event
- nat = Natural Phenomenon
the dataset is available at kaggle
the used network is a simple Bi-Directional LSTM build with tensorflow keras functional API with the below architecture.
The notebook is divided into sections:
- Data Exploration
- Data visualization
- Modeling