Named Entity Recognition for an Indonesian Based Language Tweet using Multinomial Naive Bayes Classifier

Abstract

In Natural Languange Processing (NLP), Named Entity Recognition (NER) is a sub discussion that is widely used for research. the main task of Named Entity Recognition (NER) is to help identify and detect the entity names from a word in a sentence. The data sources we use are a real time Indonesian language tweets that often occur, which the number of letter each tweet is limited to 280 characters. The words contained in that Indonesian language tweets can refer to the name of the entity, location, or organization, so to determine the name of that entity, it must be considered first by looking at the word patterns around it. In Indonesia, an average tweet posted from an account at least is 1-3 tweets per day which contain a formal and non-formal contents that made this a difficult challenge to provide the right entity naming. In this research, we are naming the entities from the Indonesian language tweets by using the Multinomial Naive Bayes Classifier algorithm. The system uses precision, recall, and f-measure as evaluation metrics. Naming this entity is able to classify with a value of f-1 reaching 80%.