From this section, you can download in a csv file the lexicon used to calculate the different indexes.

The file has 9 columns:

• keyword: the word or group of words (ranging from 1 to 10 words). Lemmanization and the Porter stemming algorithm were applied.
• ngram: the number of ngram in the group of word.
• total_class: the number of time the word or group of words was classified in either the Monetary Policy (MP) or the Economic Outlook (EC) category.
• mp_acco: the probability of belonging to the MP category with a dovish (accomodative) inclination.
• mp_neut: the probability of belonging to the MP category with a neutral inclination.
• mp_rest: the probability of belonging to the MP category with a hawkish (restrictive) inclination.
• ec_nega: the probability of belonging to the EC category with a negative inclination.
• ec_neut: the probability of belonging to the EC category with a neutral inclination.
• ec_posi: the probability of belonging to the EC category with a positive inclination.

Lastest Version :
21/05/2021 : ECB press conferences until Dec 2020 (included): Download here