Amazighe Named Entity Recognition (NER) system is a potentially vital pretreatment for so many Natural Language preprocessing (NLP) applications, and more specifically: Machine Translation (MT). In this paper, we present an Amazighe named entity recognition system based on a synthetic study of the language-specificity and the amazighe named entities. The article focuses on the methodological choices to resolve ambiguities of this language, by exploiting existing ressources in a related languages.