BALINESE AUTOMATIC TEXT SUMMARIZATION USING GENETIC ALGORITHM

A summary contains the important idea of a text. However, summarizing a text requires one to read its entire content. In this study, text summarization is done automatically by applying a genetic algorithm to optimize the weight of five sentence features. The features include positive and negative keywords, the similarity between sentences and titles, the similarity between sentences, and cosine similarity. The collection of documents in this study are Balinese text stories. The summarization technique used is the extraction technique which eliminates unnecessary sentences, without changing the structure of the original sentence. The score of a sentence is generated by multiplying the feature value of each sentence by the weight of the feature. Summarization of the text is done by sorting the sentences based on the score. At the training stage, the best weight combination is chosen based on the average fitness value. Evaluation of the proposed method is carried out using 50 test data in the form of Balinese text stories. From the test results, it can be concluded that the fitness value of the feature weights is affected by the crossover and mutation rate of the genetic algorithm. Furthermore, accuracy is also influenced by the compression parameters used.