OIEC-PT

The OIEC-PT (Open Information Extraction Corpus-Portuguese) is the corpus resulting from the manual and theory-driven annotation of sentences from the Portuguese section of the Parallel Universal Dependencies (PUD). It is composed of two complementary subsets: OIEC-PT-SILVER and OIEC-PT-GOLD, which reflect different stages of maturity in the annotation guidelines and annotator consensus.

OIEC-PT-SILVER comprises 200 sentences annotated during the first two stages of the process and was used to calibrate and refine the structural and semantic guidelines.

OIEC-PT-GOLD, in turn, contains 100 sentences annotated in the final stage, after the consolidation of the annotation rules and the resulting agreement among specialists. Due to its higher syntactic-semantic consistency and factual reliability, it serves as the gold standard for the evaluation of Open IE systems in Portuguese.

Together, the two subsets constitute the full OIEC-PT corpus, totaling 300 sentences and 473 validated extractions. This resource combines textual diversity with theoretical rigor, enabling both the supervised training of models and the standardized evaluation of Open Information Extraction methods in Portuguese.