You are viewing a single comment's thread from:
RE: Today's Truffle Picks: Quality Steemit Posts that deserve more Rewards and Attention! (11.07.2018)
Thanks for support and showcasing my work. Have a wonderful truffle day!
Thanks for support and showcasing my work. Have a wonderful truffle day!
You're welcome, and thank you, too. @trufflepig seems to be a big fan of yours :-D. How often did he select you already?
It used to be almost weekly and then it started showing up a lot more. This week alone this is my third time as it seems to be showing up every other day now. I’m wondering if things need to be adjusted slightly or if there was a recent change that has caused this in the code?
I’m not sure if my own interaction with it in regards to leaving a comment and an upvote has any influence with it either. I noticed today it did not post a comment. I did notice the upvote so I thought I would come and check things out.
I really wish I had the time to give back to the other posts it is finding. Sadly with all the other things, I’m involved with. I’m just lacking the time and the voting power to do much these days.
It is quite an interesting project, to say the least. I have enjoyed it showing up.
As far as totals I have no idea. Quite a lot of times. It really seems to love the length, amount of photos/screenshots, and my formatting. Not sure if it also uses comment engagement as well. What ever I'm doing it really loves it. I don't mind tossing it a small truffle when it shows up either.
No, I guess you just write good content that is valuable and should get paid more :-). But what does change every week is the training data. Old data is discareded and the Machine Learning model is retrained every Sunday on newer posts.
Well, no, he cannot be bribed with comments or upvotes, the algorithm doesn't take interactions with the bot into account. However, there was admittedly one single incident where @trufflepig made a biased evaluation :-D.
Yes, sometimes his comments don't make it through and are not included in the Blockchain, I should probably retry than just skip and continue upon an error.
Thanks :-)
Basically, if there are many posts in the training set that also show similar features to yours and that received high payouts, the model also predicts a high payout for your posts.
Interesting, I’m wondering what effects community votes are having on its predictions than. Even more so if this project will turn into a self-fulfilling prophecy at some point. It seems to understand in my case the chance of a community upvote for my gaming posts. My garden posts on the other hand well those are dead on arrival.
While I’m not quite sure if this curation post would qualify. You might want to check out on discord if you have not already Curation Collective. https://discord.gg/B8JFmJ4
The channel Curation-compilations might be interest to you. While they are not large upvotes that could help moves posts closer to predicted values. This is of course if you are even looking to alter this experiment in that way.