Sequence Labelling is a task of Natural Language Processing (NLP). Its main objective is to tag a sequence of tokens contained in a sentence.
On the other hand, Named Entity Recognition (NER) is a subtask of Sequence Labelling and its principal goal is to identify and classify named entity mentions in unstructured text into pre-defined categories such as person names, locations, time expressions, organizations, etc.
In this repository, different solutions are compared to solve a NER problem using a spanish database.
This is a task for the NLP course, CC6205 of the University of Chile. Here, you can find a baseline for the task, which is a basic solution created by the assistant professor Pablo Badilla.
Data: CoNLL 2002 Spanish.
Model / Metric |
Macro AVG | ||
---|---|---|---|
F1 | Precision | Recall | |
BILMST (hidden dim = 512, layers = 3, dropout = 0.2) | 0.6664 +- 0.026 | 0.745 +- 0.0115 | 0.615 +- 0.035 |