今天刚面的，题目不是notification, video call, violating content，是个Live story stream 的题目。\n\n-baidu 1point3acres\n首先，一点点小建议，大家面试的时候不要看到不是常见题目就慌，LZ 就是太慌，frame 说的太乱，该展示出来的实力，没展示出来。然后面试官的口音还蛮重的，说话又快，要自己把握好节奏。. ----\n\n下面是题目 sql 很简单\n\nTable name: info_stream_views\npost_id (BIGINT) – ID of info stream post\n\nviewer_id (BIGINT)-- ID of person who has seen the post\nRelationship (STRING) – relationship of the viewer and post author Relationship can be: {Friend, Followee, Unconnected}. 1point 3acres\nduration(DOUBLE)-- time spent in seconds of viewer on the post\nds(STRING)-- date of the view\n\n\nWhat’s the number of posts with more than 60s views from unconnected audiences in the past 7 days?Post_reactions: ..\npost_id (INTEGER) -- ID of post\nviewer_id (INTEGER) -- ID of the user who viewed the content\nPost_action (STRING) – {like, comment, reshare}. ----\nds (STRING) – date of the actions\n\n\nWhat’s the average number of reactions for friends and unconnected posts in last 7 days?\n\n\nProduct\n\n1. The social media platform is a “social” app, we want to know if friends' content being shown to users are more “social”, or drive more social engagement than unconnected contents from strangers/creators, how would you validate that hypothesis using the data above?\n2. If we are introducing unconnected content to Info stream for the first time (before Info stream is only friends content), how would we know that this is successful? Let’s focus on metrics and experimentation.