4 ms·
It's many pronged approach. User timelines are loaded for "reputable" users, (see blog for computing reputation) and search input is used with primative bayesia
by timhastings 16y ago
It's many pronged approach. User timelines are loaded for "reputable" users, (see blog for computing reputation) and search input is used with primative bayesian filtering.
In my experience, I have found Tweet source (web, Tweetdeck, API etc.) to be one of the most effective quality/authenticity indicators.
- iampims 16y agoUsing the tweet source is such a smart idea! So far I rely on username filtering / patterns. Toughest is to get rid of job postings ! Good luck with Tagwalk ! You've got a new user ;)
- timhastings 16y agoNice one, thanks, and good luck with Firetweets! There is tons of job tweets, but most have "job" in the username and/or app source. Also any app source that contains "bot" or "feed" is most likely safe to ignore.