The introduction of criteria for assessing an aligned parallel Persian-English corpus at the sentence level
Masoumeh Mashayekhi, Morteza Analoui, Behrouz Minaei Bidgoli · International Journal of Engineering and Technology · 2019
Bilingual corpora are a collection of writings that serve as an example of the relationship between two languages for linguistic and translational applications.Examining the effectiveness of the corpora is one of the essential requirements for working with them.Therefore, the validity of the work based on the corpora is to check their quality.Scientists have identified four main attributes for corpses.These four features are representativeness, limited size, machine-readable shape, standard reference.To evaluate an entity, we need to evaluate these four properties.The limited size and intelligibility of a machine in electronic compartments are certain because they are otherwise unusable.Representativeness means to put a sample set of language variations for the language in question in the corpus.In fact, the corpus has a linguistic diversity.To evaluate this property, we examine the complexity and diversity of the figure and compute the degree of compliance with Ziff's law.For the standardization of each pair, we combine several of the following characteristics: alignment, translation, command, punctuation, separation, characterization.Finally, a fuzzy system uses the final evaluation of these criteria and uses a fuzzy rule base and fuzzy inputs of the introduced evaluators to obtain a fuzzy result for the quality of the entity.