Prepare for action - help out categorizing the contents on hive!

in HiveDevs13 days ago

To train a neural network with the contents of hive, we need to prepare data by tagging as many posts as we can. You will probably all agree that the user provided tags aren't a good basis to work with, so those have to be added manually.

Screenshot_2022-08-04-14-02-54-04_a1b1bbe5f63d5b96c1a0f87c197ebfae.jpg

I'm in the last steps of finishing up a platform where anyone with a hive account can participate to get this done.
There's more administrational work to be done though, and we need volunteers. First off, the categories have to be defined. It's useful to start with the most Facebook used, but it should be adapted to our needs, and probably evolve constantly. And when it's running, it would also be good to have a few people checking the submissions once in a while.

If you're interested in helping to get this done, join the DeepHive Discord and introduce yourself!

Another way to currently support the project is a vote for my witness, which will not only help to keep my api node running, but also allow me to spend some of the funds on creating the next generation of hive social media.

Sort:  

This is the reason, I vote for your witness.

It sounds very interesting, at least to me ( neural network :p), and I would love to join. Can you please list all the beneficial things for the community emerging from this project? Thank you!

!LUV
!HBIT

I can't speak for the whole community, certainly not the current type that's focused on earnings.
We will provide a platform for mainstream users who are looking for content to be entertained with, the monetary effects will not be highlighted. If you use in in that way, you will have to search less for content you care about, and the suggestions will hopefully raise retention and engagement.

Thank you for the fast explanation. I didn't ask if it will be monetized, we already have so many things on Hive to do that, just what it will be able to do. I noticed when I search something on almost all our frontends ( Ecency is a bit faster/better) the results are a bit scarce, so I think this is going to be a great thing.
I used to update steemit Wikipedia in the ancient past, it was about all things steem, so I am familiar with cataloging. Anyways, I don't know how it is going to be done, I guess you will do that part and give us some sort of a tool, but I am looking forward to it and I would like to be a part of it.

That's the only way that can work. Vote should be like a donation. Nothing more.

Tags on hives are trash and the content is very disconnected. Recommended content doesn't work.

All that is necessary for a working user experience ( and working add revenue).

I remember all the people " wow hive is so innovative because of the rewarding system, it can replace add revenue".

back to the roots :D

With the difference, it could work on web3 in a less or even not destructive way.

IMO to train the AI, it would be easier to scrape a lot of content that is already categorized and then look at what ai is doing on hive after.

This is the way to go! To build real organic user base the monetary aspect might need to be “turned off” as most people are just here for the benefits.
If more people are here for the content primarily, then there’s a much better room for growth as most users start getting tired when their content starts having less votes.

so if the AI is not the one categorizing the posts/content
what will it do?

how will it work better than the tags we already got?

thank you :)

A neural network needs to be trained with curated datasets. You feed a lot of those into it, and the more it gets the better it can categorize future content by itself.
This will allow users to define a fixed set of interests to start with, but it's only one of the angles we're taking. For example, users of the dedicated platform will be able to opt in to tracking their behavior which lets us refine the results.

okay - so the categorizing help is only needed for the beginning?
to help the AI start?

thanks :)

Yes. It might need to be refined later, but most of the work is getting things started.

Will a use case for this be a better search function for Hive (he says with hope, lol)?
!BEER

It's not the main focus right now, but something that could later be added for sure

I've been wondering for a long time if I would see something like this on Hive, we urgently need to categorize the content that is published here.

count me in :)

Don't tell this @urun, he'll like it too much.

I'm in! I'm in the discord and reacted to your message there. Looking forward to helping out.

Hello @pharesim.
I would like to help. I just don't quite understand what is requiered and what are the benefits for HIVE users.

I'd be interested in making sure the alt health ones are covered, also gardening/homesteading. Is that what you mean?

There ya go. Finally some ai hive hashtag stuff

There is a simple way to add tags to old posts - just add a level 1 comment with tags or links to objects.

For example, by adding #neuralnetwork tag in this comment, you will link the original post to that tag, and the post will appear in the news feed for that tag.
In addition to tags, you can also use objects (terms, companies, products, places, pages, lists, etc.). For example, you can create an object on Hive for a particular place and then link to it from your posts and comments. All these posts will appear on the page of that object. Here is a tutorial about objects on Hive.

By linking tags or objects from posts, users accumulate expertise, and you can check it in user profiles. Similarly, you can click any object or tag and see the users with the greatest expertise in them.

image.png

Maybe this protocol can be useful for the needs of your project as well.