=========
NoReCfine
=========

For this targeted sentiment analysis, we used a training dataset in Norwegian with corresponding word embeddings.

We will be working with the recently released norecfine_,
a dataset for finegrained sentiment analysis in Norwegian.
The texts in the dataset have been annotated with respect to polar expressions, targets and holders of opinion but
we will here be focusing on identification of targets and their polarity only.
The underlying texts are taken from a corpus of professionally authored reviews
from multiple news-sources and across a wide variety of domains, including
literature, games, music, products, movies and more.


.. _norecfine: https://www.researchgate.net/publication/337671672_A_Fine-grained_Sentiment_Dataset_for_Norwegian


====================
NLPL Word Embeddings
====================

the word embeddings used are taken from the nlpl_ datasets,
using the *Norwegian-Bokmaal CoNLL17* corpus, with a vocabulary size of 1,182,371.

.. _nlpl: http://vectors.nlpl.eu/repository/#