Home

The main goal of the project is to develop a prototype system for a hybrid machine translator between Spanish and the native languages in Perú, as there are more than 40 original languages spoken in the country. Like the hybrid approach suggests, it requires the automatization of the linguistic knowledge, and also the computational mining of resources and annotated corpus for Machine Learning tasks. In this sense, a multidisciplinary effort is being made by linguists and computer scientists, building computational linguistic resources from scratch for a specific language as a case study: Shipibo-konibo. The partial results has been promising and they are presented in the section of publications. As the project is progressing, we are aiming to apply novel methods of representation learning for improving specific results in scarse-resource language scenarios.