Exploratory Analysis of Rinconada Bikol Language-Nabua Text Corpus

Any text corpus is an essential aspect of a language in which study is carried out or continues since the corpus of that language can be used to perform functional language study. Hence, this study focused on the development of the Rinconada Bikol Language-Nabua (RBL-Nabua) text corpus. RBL-Nabua texts were collected in social media, learning modules and other sources that will serve as the dataset. Corpus Linguistics tools and techniques were used to analyze and evaluate the developed RBL-Nabua corpus, such as tokenization, term frequency, parts-of-speech tagging, collocation, and visualization. This study managed to collect five thousand seven-hundred twenty (8754) RBL-Nabua words, and manually POS and Polarity tagged. A bag-of-words sentiment analyzer was also implemented by training five hundred RBL—Nabua sentences labeled with positive, negative, and neutral using different machine learning algorithms. Experimental result shows that it obtained an F1-Score of 57.83%, 52.08%, 39.98%, and 36.71% for Logistic Regression, Support Vector Machine, Decision Tree Classifier, and Gaussian Naive Bayes, respectively. This study could help develop natural language processing applications such as sentiment analysis, text classification, morphological analysis, language translation, and others in future studies. This study will also help in the preservation of the RBL-Nabua language as well as its affiliated culture.