Distributed sentiment analysis of an agglutinative language via Spark by applying machine learning methods

摘要

Currently there are a large amount of studies conducted on the field of sentiment analysis worldwide. Due to continuous drastic increase of amount of data, all natural language processing systems require special approach oriented to it. Implementing and evaluating such mechanism was a strong incentive to conduct this research. In this paper, sentiment analysis was implemented on Kazakh language via Spark on the basis of data set taken from Kazakh books. The data was initially unbalanced. There was achieved training F1 measure over 90 %.