Anotação de Entidades Mencionadas na área do Gaming

This paper aims to analyse the effects of including gaming entities in performance NER system, for English language and a machine translation industrial context customer support content. To identify classify (by Named Entity Recognition (NER) model), three new categories were created added already used annotation typology: GAME NAME, FEATURE CURRENCY. A set reference annotations (gold standard) was also developed, allowing not only training system but evaluation its accuracy more objective way, namely by counting number that identifies categorises correctly. In scope this work, 6618 sentences from 7 clients manually annotated, constituting gold standard which then train evaluate system. The experiments assess whether existing improved when trained with specifically domain if it could handle typology identifying categorizing them results both auspicious positive, demonstrating relevance greater investment domain-specific entity recognition, service text processing.