Using sentiment analysis to analyze how positive or negative your steemit account is
I created a new tool that allows you to see how positive or negative the language that any steemit account is.
It's on the same page as the word cloud. Just type in an account name and then click "Generate".
Here's mine
If you have any ideas for future features or improvements, please comment and let me know!
I use this package for the actual sentiment analysis: https://www.npmjs.com/package/sentiment
You can find my repo here: https://github.com/evanmarshall/steemit-cloud/
Cool tool! I'll publish a post about my word cloud and sentiment tomorrow with a link to your tool.
If you feel the calling, I'd like to have a tool that provides me with 100 recent posts that I'd like to read.
These posts need to be found based on my account activity such as upvotes, follows, commenting, resteem, etc.
If you want some ideas about the pseudocode, I am willing to exchange ideas, if you are willing to implement.
If you can implement that, you'll fill a huge gap in Steemit and make a big change in Steemit.com, because most users I know complain about the difficulty of relevant content.
As far as I know, there isn't a tool that provides that.
I think that's a great idea and another steemian is also thinking about it. Check out the conversation here: https://steemit.com/technology/@lextenebris/programming-beginning-the-search-for-discovery
Essentially, I think there are many different recommendation algorithms and potential UIs. How would you like to use a tool like that? Would it be a separate webpage that would just show 100 recommendations or would it be a bot that tagged you when it thinks you'll like something?
Thank you for the link. I’ll check it out.
Do you mean theoretical algorithms or are there already some that are implemented on Steemit data?
If there already some implementations, I’d appreciate some links, because I can’t seem to find them.
I prefer a page like your word cloud. I enter my username on the top of the page and I receive 100 posts, broken down into 10 pages.
I don’t mind waiting for a few seconds until it loads. Neither do I mind a few ads on that page.
I wouldn’t like that idea. It’s not privacy friendly. Suppose that it tags me with a completely irrelevant, irritating post.
Moreover, it might be perceived as me spamming posts, if I understood the idea correctly.
By the way, I don’t think I get what you mean with this sentence. Do you mean commenting with my username?
If so, how can I find the posts and comments that mention my username? I can’t find a method to do that except the google solution and it takes a while until those pages are indexed.
I don't know any algorithms already implemented using steem data. I do know that recommendation algorithms are plentiful and there are many implemented examples. In other words, repurposing existing code should be able to do the job.
I like the separate page too but I'm not sure how making users go to a separate site is friction for usage.
The alternative of a bot commenting with your username and a specific tag is you can use something like https://www.asksteem.com/search?q=%22%40egmracer01%22 with a specific tag and I don't have to reinvent the search interface.
Believe me, I'd go to any website to find content relevant to my account. Trending and hot tabs don't work for me anyway. But of course, that's only me. I can't speak for others.