Explorando métodos non-supervisados para calcular a similitude semántica textual

Pablo Gamallo, Martin Pereira-Fariña

Abstract: This paper presents some unsupervised methods for detecting semantic textual similarity, which are based on distributional models and dependency parsing. The systems are evaluated using the dataset realased by the ASSIN Shared Task co-located with PROPOR 2016. The more basic methods offer better behavior than the more complex ones, which include syntactic-semantic information in sentence analysis. Finally, the use of distributional models built automatically from corpora provides results comparable to strategies that use external lexical resources built manually.