A Case Study of Sentiment Orientation Identification for Polish Texts

Abstract

In order to make rational decisions and react quickly to changes in the business environment, organizations, especially those operating in the e-business setting, need to constantly monitor numerous information sources on the Internet, e.g., electronic media or opinions published by users at various portals. Majority of the opinions and therefore data concerning enterprises is stored in various forms in broadly understood social media. The available resources are presented in a variety forms, both highly structured and free-style form. In order to identify the emotional attitude of the published texts, there is a well recognised need to automate the overall process and to employ sentiment analysis techniques. In this paper, we show how the already available resources for the Polish language, such as parsers, semantic networks, thesauri, general and domain specific corpora can be further extended and used as a cornerstone for more advanced applications. A case study of Sentiment Orientation Identification system for Polish texts is presented together with the obtained results from the conducted experiments.