Paper abstract:

Sentiment - analysis is a form of natural language processing that gives us information about the sentiment (emotion) of the text - is it positive (towards something or someone), negative or neutral;

All the examples presented in the Presentation are from the political corpus of the Georgian language.

The political discourse can be defined as political communication used by politicians in their public speeches, planned (inaugural texts), or unplanned speeches (television programs, participation in discussions, parliamentary speeches).

Sentiment analysis research has mainly three interrelated and at the same time distinct levels (Bing Liu, 2016). these are:

o Document level;

o Sentence level;

o Aspect level;

At the document level, the task is to determine whether the overall sentiment expresses negative or positive sentiment (Bing Liu, 2016).

Sentence level is the determination of what each sentence expresses - negative, positive, or neutral.

When defining sentiment at the aspect level, we consider several aspects: whom we value (it can be a specific person and most of the time, it is a politician), what we value (mostly, it is a concrete action of a specific politician), how we value it (negatively, positively or neutrally) (Bing Liu, 2016). In this Presentation, based on examples, we will show why the level of the aspect of sentiment research has the greatest advantage.

In this Presentation, we will talk about the role of adjectives in determining sentiment - usually, a semantically neutral noun can be accompanied by an adjective with a positive or negative sentiment word, and accordingly, the sentiment of the opinion will change - it will become positive or negative.

Usually, an adjective with positive or negative content can be attached to a semantically neutral noun and, accordingly, the sentiment of the opinion will change - it will become positive or negative; However, we may have a completely different picture if the noun also has its own sentiment. Based on the combinations and analyzing the contexts, it was found that the most important component in the process of sentiment determination is not individual words, but phrases, and in many cases, not noun phrases (NP), but verb phrases (VP) play an important role in sentiment determination;

We highlight the rules that we have identified directly as a result of observing the empirical material.