A lemma is the canonical dictionary form representing related inflected words. Adjectives commonly use masculine singular forms, countable nouns singular forms, and verbs infinitives. section 5.6.5
Dictionary-based sentiment search works best when reviews and lexicons use the same representation. If a lexicon contains lemmas but preprocessing keeps surface word forms, sentiment detection can miss matches and show low sensitivity. section 5.6.5
Matching a Portuguese sentiment lexicon
Choose a lemma-based resource such as SentiLex-lem-PT02.txt when appropriate, then use spaCy’s Portuguese model to normalize each review, extract alphabetic lemmas, remove stop words, and compare the resulting list with the lexicon. Inspect the output and validate sentiment results against sample reviews, because language morphology and resource differences can affect performance. section 5.6.5 section 5.6.3
pip install spacy python -m spacy download pt_core_news_sm
Install spaCy and download the Portuguese language model required for the lemmatization workflow. section 5.6.5
