我们第一次尝试的是排行榜,但它似乎更看重纯粹的数量(想想“修好了吗?修好了吗?修好了吗?你好?”)而不是质量。好吧,是的,“质量”很难量化,尽管可以进行一些明显的调整。
我们开始通过以下方式识别真正有帮助的用户:
寻找另一位用户是第一个回答的帖子
-- 目标:查找第一个非 OP 回复是非 SonarSourcer 的主题
WITH
SonarSourcers AS (
SELECT u.id AS user_id
FROM groups g
INNER JOIN group_users gu ON g.id=gu.group_id
INNER JOIN users u ON u.id = gu.user_id
WHERE g.name='sonarsourcers'
),
Tagged_topics AS (
SELECT tt.topic_id
FROM topic_tags tt
JOIN tags t on t.id=tt.tag_id
WHERE name = 'me-too'
),
-- 查找由普通用户创建的“常规”主题
topic_user AS (
SELECT id as topic_id, user_id, created_at
FROM topics
LEFT JOIN SonarSourcers ss USING(user_id)
LEFT JOIN tagged_topics tt on topics.id = tt.topic_id
WHERE ss.user_id IS NULL -- 排除 SonarSourcers 发起的主题
AND tt.topic_id IS NULL -- 排除标记为 me-too 的主题
AND user_id > 0 -- 排除 DiscoBot 的“欢迎”PM/教程主题
AND visible = TRUE
AND archived = FALSE
AND archetype='regular'
AND created_at::DATE > '2023-07-01'
),
-- 查找用户主题的第一个非 OP 回复
min_response AS (
SELECT p.topic_id, tu.created_at, MIN(post_number) as post_number
FROM posts p
JOIN topic_user tu USING(topic_id)
WHERE p.post_type = 1
AND p.user_id != tu.user_id
AND p.post_number > 1
AND p.hidden = false
AND p.deleted_at IS NULL
GROUP BY topic_id, tu.created_at
)
SELECT p.topic_id, p.user_id, mr.created_at::DATE as thread_date
FROM posts p
JOIN min_response mr ON p.topic_id = mr.topic_id AND p.post_number=mr.post_number
LEFT JOIN SonarSourcers ss ON p.user_id=ss.user_id
LEFT JOIN user_badges ub on p.id = ub.post_id and ub.badge_id=110
WHERE ss.user_id IS NULL -- 排除 SonarSourcer 是第一个回复的主题
AND ub.user_id IS NULL -- 排除已授予徽章的主题
ORDER BY mr.created_at DESC
一旦我们确定了这些帖子,我们就会评估用户的回复,然后要么授予回复用户“用户互助”徽章,要么在帖子中添加一个(非员工不可见)的“me-too”标签。 (顺便说一句,这个用户回复报告还有助于快速发现我们云服务中的事件。
)
从那里开始,就可以进一步奖励有帮助的用户,并识别他们以进行进一步培养。