TSV-formatted embeddings released by Mark Granroth-Wilding
==========================================================

These embeddings accompany the paper:
    Unsupervised Learning of Cross-Lingual Symbol Embeddings Without Parallel Data
    Mark Granroth-Wilding and Hannu Toivonen (2019).
    In proceedings of the Society for Computation in Linguistics (SCiL)

They are trained using the Xsym method described in the paper.

All embeddings for language pairs referred to in the paper, as well
as code for data preparation and model trainined are released at:
    https://mark.granroth-wilding.co.uk/papers/unsup_symbol/

Each set of embeddings is made available in two different formats,
both standard formats for storing vector embeddings. These embeddings
are stored in the TSV (tab-separated value) format used, for example,
by https://projector.tensorflow.org/. They are accompanied by metadata
identifying the language and character for each embedding. The TSV
file, together with the metadata, can be loaded into
https://projector.tensorflow.org/ to explore the vector space in a
3D projection.

See the website for more details about formats.

Characters are represented as their unicode character, with a prefix
denoting its language, e.g. fi:a is the name for the Finnish letter
a. Note that in the embeddings for the Finnish-Estonian language pair,
fi:a and et:a are considered distinct characters and have separate
embeddings. See the paper for more details.
