Heading
Building a positive image of an organization represents the central activity of public relations that is often achieved through media relations. Today, the evaluation and measurement of the effects of the media represent important issues in the fi eld of public relations (PR), while a positive or a negative image can be studied as a result of the PR activities. Since the Internet represents an increasingly important means of communication, the fundamental objective of this study was to identify the online image of the passenger transport organizations (Croatia Airlines, Croatian Railways and the Zagreb Electric Tram) on the internet portals through the analysis of usergenerated content. This paper studies the method of computer sentiment analysis of user comments in two news portals - 24sata.hr and Index.hr. In parallel with the human sentiment analysis, the paper examines the accuracy of sentiment analysis of Croatian language, testing the sentiment of the users of Croatian news portals (more precisely, users’ comments on media reports about the organizations in the domain of passenger transport - Croatia Airlines, Zagreb Electric Tram and Croatian Railways). Based on the sentiment analysis and word frequencies in the users’ comments, a computer lexicon of sentiment words for the domain of transportation was generated. Also, this paper identifi es the key challenges of the sentiment analysis for the Croatian language, but also for other morphologically rich Slavic languages, based on the comparison, correlation analysis and measurement of the relative error of computer sentiment analyzes in relation to the human sentiment analysis. The results indicate the potential use of sentiment analysis to identify the positive, negative and neutral emotions towards organizations dealing with the passenger transport. The paper concludes that the users of Croatian news portal have the predominantly negative sentiment towards the passenger transport organizations, i.e. they have a negative online image.
