# Sentiment analysis, experimenting with lexicons
###### tags: `3. Text Experiments`

I ran 9 custom lexicons across the context matches and full articles.

Lexicons:
- AFINN standard lexicon, values -5 to +5
- AFINN extreme values. Only include words that have values 3-5.
- AFINN categorical. Simply -1 and +1 for pos and neg values
- sentsum_PT_score - hand-annotated ADJectives, Peeter (POS+, NEG-, REAL-)
- sentsum_KV_score - hand-annotated ADJectives, Kristiina (POS+, NEG-, REAL-)
- sentsum_PT_real - hand-annotated ADJectives, Peeter (only REAL-)
- sentsum_KV_real - hand-annotated ADJectives, Kristiina (only REAL-)
- sentsum_LK_score - hand-annotated NOUNS, Laur (POS+, NEG-)
- sentsum_combined_PT_KV, hand-annotated ADJecitves, Kristiina & Peeter combined (POS+, NEG-, REAL-)

Kristiina & Peeter are working on expanding the custom lexicons via wordnet too.