Normally a Twitter bot is used to reply to relevant tweets that mention an account or business, or to direct messages, or to retweet tweets that are relevant to an account. Replying to random tweets is probably not a good idea an may result in your account being suspended by Twitter. Refer to Twitter's automation policies,
http://support.twitter.com/articles/76915-automation-rules-and-best-practices
Looking at your bot it is learning from the tweets that it is following (you can see these in the bot's "Training & Chat Logs" page by browsing "responses"). A Twitter bot will only responds to mentions, or tweets from a tweet search that match its "Reply Keywords/Hashtags", your reply keywords are empty, so it will not reply to tweets that do not mention it.
If you add reply keywords it will respond to searched tweets with those keywords, only if it has a trained response that is a "good" match. Here a "good" match normally means a 90% match. You can change that match % required in the bot's "Learning & Settings" using the "Discussion Match %", a lower value will have more matches, more replies.
For learning the default learning rate is 50%, since a 90% match is required, it would need to learn the same response multiple times to have a successful match. You can increase the learning rate from "Learning & Settings" using the "Learning Rate %", i.e. setting it to 100% would learn any tweet instantly. Also responses that use keywords will be more likely to match a tweet.
|