Classification and Clustering of Emotive Microblogs in Albanian: Two User-Oriented Tasks

摘要: Human emotion analysis has continuously stimulated studies in different disciplines and it is spurring interest among the computer scientists too. Particularly, growing popularity of Micro-blogging platforms, generated large amounts data, which turn represent an attractive source to study social media users, especially user-generated content analysis, such as opinion mining sentiment analysis. In this paper, we propose analyze micro-blogging order characterize users individually when writing posts with emotional content. The two-fold considers at granularity levels, one refers textual units allows us capture state expressed by user, other collections summarize lexicon used user. particular, first case, focus on a sentence-based detection problem, aimed classifying into set pre-defined categories. second performed through keyword extraction approach, finding representative generic word sets form prototypes unit clusters. Extensive experiments conducted under perspectives, yet always centered around reveal interesting findings terms classification accuracy, clustering incoherence versus classifications perspectives valuable efforts user profiling.