The Easiest Way to Discover Posts on Steemit that You Like? Me! (01.04.2018)

Hound Dog at Your Service

Finding good content on Steemit among thousands of posts published each day can be exhausting. But don't you worry, I can help! My name is @hounddog and I am your personalized post recommender. I hunt and track down posts that you like based on your previous upvotes and interests. I read through every publication and, by using Natural Language Processing, can support you to discover content that is right for you! It's very easy and the only thing I want in return is a vote from you. Just upvote this post and comment to receive recommendations. The details are explained below.

How It Works

1. Upvote this post, giving a 100% upvote
   • optionally follow me
   • and resteem this post
2. Reply to this post
   • with just: @hounddog to get personalized recommendations based on your last 3 upvotes
   • or with: @hounddog https://steemit.com/steemit/@myfavoriteauthor/my-favorite-post to find similar posts to the one in your comment
   • or with: @hounddog search: Some keywords or text like I love Jimi Hendrix or the first Matrix movie to search for the text and keywords
3. Wait for about a minute and refresh this webpage
4. I will recommend new, recent posts to you
   • 3 recommendations for your vote
   • 5 if you additionally follow me
   • or even 10 if you upvoted, followed, and resteemed this post

Please, give only 100% upvotes, anything less won't get you very far!

You can get multiple recommendations if you upvote my recommendations to you and reply to this post with @hounddog another time. Make sure to vote on other posts in between, otherwise you may get the same recommendations again!

IMPORTANT: Recommendations are given until 06.04.2018 (UTC). Find a newer post of mine after that date for new recommendations.

Example

Here is an example using the trending post Pretty girls and fast cars in Melbourne's Formula 1 by @hitwill. Let's say you upvoted my post, followed me, and replied with:

@hounddog https://steemit.com/steemit/@hitwill/pretty-girls-and-fast-cars-in-melbourne-s-formula-1

I would recommend the following posts:

1: Pumping Poo Around Melbourne Warning This Post is about Shit! by @scooter77 (24% match)
2: Beautiful Buildings of Melbourne Part One The U.S.S. Minnowprize! by @sammosk (19% match)
3: Melbourne Formula 1 Race Results by @jeff001 (68% match)
4: A Wedding at the Melbourne Zoo by @scooter77 (19% match)
5: Formula 1 Cars 2018 Halo by @jeff001 (21% match)

FAQs

I upvoted your post, but I did not receive any recommendation?!

Did you give a 100% upvote? If not, only 100% upvotes count. If yes, wait a tiny bit longer, maybe I am currently being in maintenance mode. Finally, leave a comment to let my developer know that something is not right and need to be fixed. Thank You!

Can I use your service if I sold my vote to services such as @smartsteem?

Well, you can give it a try. But my suggestions are based on the upvotes of your account. If your account voted for crap, you will get crap in return. Your vote on Steemit counts, so think twice about whether you sell it to a bot! Alternatively, you can add an example post (like @hounddog https://steemit.com/steemit/@myfavoriteauthor/my-favorite-post) and I will search for other posts similar to your example.

Can I get some more recommendations?

Yes, just upvote my first recommendations to you and reply to this post with @hounddog again.

Can I send you SBD to get even more recommendations?

Well, no! Sure send me some SBD if you have no use for them, but do not expect anything in return. You pay for this service with your vote only. This is recommendation socialism, baby! If you can give a lot, you have to give a lot, if you cannot, you only have to give a little!

How do you give recommendations?

In a nutshell, I read all posts of the last two weeks and, first, filter all short ones, non English ones, and publications drowning in spelling mistakes. Secondly, I extract the essence out of each contribution. I do this by embedding the posts into a vector space using a term frequency inverse document frequency encoding. Next, I look up the last 3 posts you voted for with at least 25% voting weight (or, alternatively, just read the example post you provided) and project these into the space, too. Finally, I search the space for the articles closest to your choice by maximizing the cosine similarity between your votes (or the example post you gave) and the last two weeks' publications.

Are you related to @trufflepig?

Yes, I am designed and coded by his developer @smcaterpillar, too. And yes, I even do use parts of @trufflepig's codebase. However, he tries to satisfy the taste of the masses whereas I try to cater to your needs only.

Best with barks,

HoundDog