[Journey] 1 million UVs/month in 12 months using AI generated content. Let's do it!

[ANewBeginning]

this I don't understand. You could easily train a model which predicts with 99% if a site is an autoblog/spam site.

Well, google works in mysterious ways

 

[Alexion]

The simple answer is they don't bother. It's not that important to them as you might think. While it is a big deal for us, this venture is nothing for them.
Name one autoblog that's been around for more than 1 year. Chances are you will hardly find any.

They do let you have your fun. It's up to you to make some money out of it or not. Sooner or later, they will indeed catch up. That's not there the money is for them.

It makes more sense to let people create thousands of thousands of auto blogs, so they can get an idea of what's going it, instead of just killing them on the spot.

 

[ANewBeginning]

Name one autoblog that's been around for more than 1 year.

something like walmart, home depot, amazon. Yup, they are BIG autoblogs running for MORE than 1 year. Those are the big ones (missing a ton of big ones as well), but there are THOUSANDS and thousands of autoblogs out there running for more than a year, and even if they don't last a year, guess what! You just build tons more of them!

The beauty of spam is that the key is in VOLUME. Volume is everything. Spam in volume

 

[Fardin2012]

pegasus takes 10 to 15 minutes for 1000 words article.

Can you please give me SSH code for how to install pegasus? I want to test it on dedicated server.

 

[Sartre]

Can you please give me SSH code for how to install pegasus? I want to test it on dedicated server.

https://huggingface.co/tuner007/pegasus_paraphrase
works fine with Python 3.9. Let me know what problems you encounter.

 

[azooz3001]

if you are interested to try this one in arabic website, im here to help in testing

 

[AlexTheGeek]

Hi there!
I'm doing much of the same as you and I've come pretty far. My question is: can you go a little bit more into how you weight the content of the different articles? How do you choose what to include and not?

 

[Sartre]

Hi there!
I'm doing much of the same as you and I've come pretty far. My question is: can you go a little bit more into how you weight the content of the different articles? How do you choose what to include and not?

I have several methods, obviously, I don't want to reveal everything, because some of them have very limited scope and are easily replicable.

As I said before, for example, I will make a list article, gather 10 items I want to cover, find reviews for each, and find 1 paragraph for each that contains as many specific words as possible. I'm using Levenshtein distance for comparing different text with each other now.

Can you show a sample generated article of yours?

 

[AlexTheGeek]

I have several methods, obviously, I don't want to reveal everything, because some of them have very limited scope and are easily replicable.

As I said before, for example, I will make a list article, gather 10 items I want to cover, find reviews for each, and find 1 paragraph for each that contains as many specific words as possible. I'm using Levenshtein distance for comparing different text with each other now.

Can you show a sample generated article of yours?

Ok, guess I have to find a solution myself then hehe
As I said, I'm still working on putting together the parts to create the full article. The rewrite is easy, the structure is the hard part.

I will also implement a function that adds in pictures at certain points by using the keyword that appears the most in a paragraph I think, but might still use manual editing.

 

[Sartre]

The rewrite is easy

using Pegasus, T5, BERT or something else?

The pics are a great idea as you can see in my earlier posts I'm doing that too!

 

[AlexTheGeek]

using Pegasus, T5, BERT or something else?

The pics are a great idea as you can see in my earlier posts I'm doing that too!

Pegasus, but might try GPT-3 one time after tuning it. We'll see. This would save me 3/4 of the time I have to use on my weekly articles!

 

[blinks]

Very good thread. I'm also working on a project like this. using the AI method, paraphrasing.

Publish 5 - 10 articles per day. I built about 5 blogs, next month, index drop. Are you using Cloudflare?

 

[Sartre]

Very good thread. I'm also working on a project like this. using the AI method, paraphrasing.

Publish 5 - 10 articles per day. I built about 5 blogs, next month, index drop. Are you using Cloudflare?

yeah I'm using cloudflare. What do you mean by index drop?

 

[blinks]

yeah I'm using cloudflare. What do you mean by index drop?

I mean, in the first month, the index was stable and reached the hundreds. In the second and third months, the index dropped to less than 10. I checked with the method site:domainname.com

Do you do backlink optimization?

 

[PrincZ]

damn, following closely. really cool to know all this paraphase.

 

[Sartre]

I mean, in the first month, the index was stable and reached the hundreds. In the second and third months, the index dropped to less than 10. I checked with the method site:domainname.com

Do you do backlink optimization?

I don't build backlinks I don't believe they are necessary to make bank anymore. I think time/money is better spent making more content.

It sounds to me like your content isn't good enough if it was dropped.

 

[DanDD]

I don't build backlinks I don't believe they are necessary to make bank anymore. I think time/money is better spent making more content.

It sounds to me like your content isn't good enough if it was dropped.

What ad platforms are you using?

 

[hehe2444]

@brrupsz do you also have internal links on each of your articles?

 

[Niffo]

Do u have meta description for the article, If so how do you create it?

 

[blinks]

I don't build backlinks I don't believe they are necessary to make bank anymore. I think time/money is better spent making more content.

It sounds to me like your content isn't good enough if it was dropped.

On average, how many words are in an article you write, 500 or 1000? What libraries do you use in the "semantically relevant keywords" method?