I'm working on articles suggestions. There are lots of them. Think about YouTube videos suggestions. To avoid suggesting article once again I'd like to remember which were seen already by particular user. I have lot's of users and I'd like to avoid infinitely growing database of history as well.
Every article has MongoDB ObjectId. I use Redis and Go language.
I think that Bloom Filter could solve this problem, because false positives are ok in this case. I'd like to avoid false negatives, but it's not 100% obligatory.
I don't know any sensible alternative to Bloom Filter in this area. Should I?
I can't find any implementation of Bloom Filter for Redis in Go. Can somebody advice me is it the best choice, how can I write my own or is there any existing implementation?
For anyone stumbling across this post, if you're using Redis in this situation, then you probably want to use its HyperLogLog feature; Antirez wrote about it at length here. No need to figure out how to make a bloom filter in Redis, it has something just as good built in (and it can also tell you approximately how many items have been inserted in it).
As @jumand mentioned, Redis has a module named RedisBloom which provides a Bloom Filter, Cuckoo Filter, Count-Min Sketch and TopK services.