Skip to content

Latest commit

 

History

History
55 lines (51 loc) · 3.27 KB

File metadata and controls

55 lines (51 loc) · 3.27 KB

Casts Features

Contains cast data since November, 1st 2024, enriched with categories, topics, embeddings, ML features and language.

Dune link

Updated every 2 hours, lagging 2 to 3 hours behind real time data.

Column Type Description
day DATE YYYY-MM-DD
timestamp TIMESTAMP Cast timestamp as recorded by the Farcaster hub.
hash STRING Cast hash.
fid INTEGER Author fid.
user_name STRING Author username at time of cast.
num_follower INTEGER Author's number of followers at time of cast.
num_follower_bin INTEGER num_follower bin from 0 to 9, based on the same hour distribution.
num_following INTEGER Author's number of followed accounts at time of cast.
num_following_bin INTEGER num_following bin from 0 to 9, based on the same hour distribution.
num_casts_same_hour INTEGER Number of times the user casted in the same hour.
text STRING Cast text.
text_len INTEGER Cast text length.
text_len_bin INTEGER text_len bin from 0 to 9, based on the same hour distribution.
embeds JSON Array of embeddings as recorded by the Farcaster hub.
mentions JSON Array of mentions. As recorded by the Farcaster hub.
mentions_pos JSON Array of mentions positions. As recorded by the Farcaster hub.
embeds_num INTEGER Number of embeddings.
mentions_num INTEGER Number of mentions.
parent_fid INTEGER Parent cast fid when it's a reply.
parent_hash STRING Parent cast hash when it's a reply.
parent_url STRING Parent url when it's casted in a channel.
category INTEGER Category classification into one of 11 categories.
category_label STRING Category label, one of c_arts, c_business, c_crypto, c_culture, c_misc, c_money, c_nature, c_politics, c_sports, c_tech_science.
topic INTEGER Topic classification into one of 60 topics.
topic_label STRING Topic label.
q_clear INTEGER Is it clear? (from 0 to 100)
q_audience INTEGER Does it have a clear target audience? (from 0 to 100)
q_info INTEGER Is it informative? (from 0 to 100)
q_easy INTEGER Is it easy to understand for the general layman public? (from 0 to 100)
q_verifiable INTEGER Is it verifiable online? (from 0 to 100)
q_personal INTEGER Is it personal? (from 0 to 100)
q_funny INTEGER Is it funny? (from 0 to 100)
q_meme_ref INTEGER Is it a reference to a well known meme? (from 0 to 100)
q_emo_res INTEGER Does it elicit an emotional response? (from 0 to 100)
q_happiness INTEGER Does it convey happiness? (from 0 to 100)
q_curiosity INTEGER Does it trigger curiosity? (from 0 to 100)
q_aggressivity INTEGER Is it aggressive? (from 0 to 100)
q_surprise INTEGER Does it have an element of surprise? (from 0 to 100)
q_interesting_ask INTEGER Does it ask an interesting question? (from 0 to 100)
q_call_action INTEGER Does it come with an explicit call to action? (from 0 to 100)
c_xxx INTEGER c_columns indicate how likely the cast is to be in that category. (from 0 to 100)
t_xxx INTEGER t_columns indicate how likely the cast is to be in that topic. (from 0 to 100)
dim_1 to dim_32 FLOAT Reduced embedding dimensions. These were calculated such that, when combined with the q_, c_ and t_ features, will conserve 50% variance from the original 512 dimensions embedding.
predict_like INTEGER Output of the Likemeter model (0 to 100).
language STRING Autodetected language from the text.