In this folder you will find the code and data from the following publication: 

Goran Glavaš and Ivan Vulić. Discriminating between Lexico-Semantic Relations with the Specialization Tensor Model. 2018. 
In Proceedings of the Conference of the North American Chapter of the Association of Computational Linguistics: Human Language Technologies (NAACL-HLT). 
New Orleans, Lousiana, USA.

The LAST version of the code can be found in the following repository: 
https://github.com/codogogo/stm


In the "code" folder you'll find two main scripts: 

1.  *stm-train.py* trains an instance of the STM model using provided data;
2.  *stm-predict.py* predicts labels for pairs of words, using a model previously trained with stm-train.py. 

You can see the descriptions of mandatory and optional arguments by running the srcipts with option *-h*

In the "data" folder you'll find the datasets used to evaluate the STM model in the paper, namely: 

1. The WN-LS dataset (in English, Spanish, German and Croatian variants)
2. CogaLex-V dataset (used also in previous work)

For questions regarding the code and/or data, contact:
goran@informatik.uni-mannheim.de 