Use different algorithm for trending tags (#7697)

This commit is contained in:
Eugen Rochko 2018-06-01 19:19:44 +02:00 committed by GitHub
parent 8d6e4e0485
commit 6e6c0e9613
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -2,17 +2,16 @@
class TrendingTags
KEY = 'trending_tags'
HALF_LIFE = 1.day.to_i
MAX_ITEMS = 500
EXPIRE_HISTORY_AFTER = 7.days.seconds
THRESHOLD = 5
class << self
def record_use!(tag, account, at_time = Time.now.utc)
return if disallowed_hashtags.include?(tag.name) || account.silenced?
return if disallowed_hashtags.include?(tag.name) || account.silenced? || account.bot?
increment_vote!(tag.id, at_time)
increment_historical_use!(tag.id, at_time)
increment_unique_use!(tag.id, account.id, at_time)
increment_vote!(tag.id, at_time)
end
def get(limit)
@ -24,8 +23,16 @@ class TrendingTags
private
def increment_vote!(tag_id, at_time)
redis.zincrby(KEY, (2**((at_time.to_i - epoch) / HALF_LIFE)).to_f, tag_id.to_s)
redis.zremrangebyrank(KEY, 0, -MAX_ITEMS) if rand < (2.to_f / MAX_ITEMS)
expected = redis.pfcount("activity:tags:#{tag_id}:#{(at_time - 1.day).beginning_of_day.to_i}:accounts").to_f
expected = 1.0 if expected.zero?
observed = redis.pfcount("activity:tags:#{tag_id}:#{at_time.beginning_of_day.to_i}:accounts").to_f
if expected > observed || observed < THRESHOLD
redis.zrem(KEY, tag_id.to_s)
else
score = ((observed - expected)**2) / expected
redis.zadd(KEY, score, tag_id.to_s)
end
end
def increment_historical_use!(tag_id, at_time)
@ -40,12 +47,6 @@ class TrendingTags
redis.expire(key, EXPIRE_HISTORY_AFTER)
end
# The epoch needs to be 2.5 years in the future if the half-life is one day
# While dynamic, it will always be the same within one year
def epoch
@epoch ||= Date.new(Date.current.year + 2.5, 10, 1).to_datetime.to_i
end
def disallowed_hashtags
return @disallowed_hashtags if defined?(@disallowed_hashtags)