Sentiment analysis, Using computer programs and word dictionaries to automatically measure emotions in text, Tokenization, Splitting text into individual words (tokens), Stop words, Very common words removed before analysis (e.g., "la", "de", "the"), Unit of analysis, The thing that gets a score: a word, a tweet, a document, Normalizing, Dividing counts by text length so long and short texts can be compared, Full vs. pruned data, All tweets vs. only tweets containing at least one dictionary word, False positive, A word counted as an emotion it doesn't express in that text, Validity, The degree to which a research tool (like a translated sentiment dictionary) accurately measures the specific concepts or emotions it claims to measure, Domain problem, The challenge where a sentiment dictionary misinterprets text because words take on different meanings depending on the specific context or subject area

DEFINITIONS

от
Печат
За вграждане

Табло

Визуален стил

Опции

Шаблон за превключване

Възстановяване на авто-записаната: ?