Original Articles Automatic Creation... Possible?

[VladAros]

Hi all,
I have developed a "small" script to generate original content automatically based on keywords and known articles. The script scrambles more than the sequence of sentences: it scrambles a set of words and combines them with related keywords and key sentences mined from the Internet.
The grammar check and meaning relation is almost absent, that's the next (and much more difficult) step.
Therefore, most the content as no meaning for a human but may have for a machine.
What's your opinion? Will work with Google and allow the increment of PR?

Here's a small sample based on "women shoes" and 2 ezine articles:

"Make their in even popular become footwear which they are should extent addresses considerably. Into an uncomfortable pair of local are certainly problem woman. Problem provide as shoes you can case with that many can be worn inches designed to support occasion and comfortable. Soles when shoes. Woman's hit almost however. Means with just for at least. Shoes tackle shoes now without needing also an excellent major everything which for so long and protect were in the find so attractive the one style died out comfortable remain. Which first conditions 1970s and stunning sandals a stunning so why or more curve. Platforms for a night strength and that sole so that traction you non-slip traction to their village shoe village after designer dress materials smallest fashion."

If you want, post a small paragraph (6-7 sentences) and your keywords and I will generate a new text based on that.

 

[Michael]

Google can detect reading level and grammar, so that paragraph will have about as much SEO benefit as it being "spaghetti women's shoes domaining footwear turkey sandwich" because G would slap both for being spam even though they are "original". If you could crack this nut you would be rich, but I doubt it can be done without a team of MIT graduates

 

[VladAros]

Thanks for your opinion. And, I completely agree with you.

Google analyses adjective, verb, noun, etc... patterns but I doubt they correlate meaning. That would take to much time and power to do it when indexing content. And if they had the technology they would have perfect translations, what is far from the reality. But, I'm sure they are researching on that.

To map the text to acceptable grammatical patterns is not impossible. What for now it almost impossible is to correlate meaning.

This script was the result of 1 day work (my own), and BTW, I have a team of "MIT level" post-graduates. Maybe I will give them this to improvement as an hobby.

 

[Jolin]

I don't know how is that going to work. For instance, the site will definitely have a high bounce rate which may trigger something to google. Apart from that, users might even report your site since it practically means nothing. Furthermore, you won't get some of the natural direct traffic or even automatic bookmarking submission from the visitors.
If it's short term, then at least it may contribute something (which I highly doubt). But in terms of long-term, I would definitely take a pass on this.

 

[barefoottech]

Ah! Ah! So you are the person that is creating all those crappy articles.

 

[VladAros]

Not really, I never published anything like that. Except here!
For me this is just an hobby in a major natural language project.

DJuqa, you look like a "professional" minisite builder, do you think your "excellent" articles make a difference for google? There are hundreds of mini-articles exactly like yours with the same keywords and key-sentences, small changes don't matter for a content indexer. BTW, I know some persons that worked with Google in natural text and speech precessing, so I have trusty sources.

In conclusion, crappy is just a relative point of view.

 

[n00b1]

I read the first sentence of the 'new article' and had to stop. I could write better when I was 6.

 

[VladAros]

That was the point of this thread. How close do you think Google goes in terms of text analysis. What's natural to a human, even to a child, is very difficult to a machine to do in real or pseudo-real time.

 

[enlytend]

Search engines are only one part of the earnings equation.

If your content is so bad that you scare away visitors, your site won't convert. If your site earns nothing, what does it matter if you're #1 or #101?

Not to mention if you succeed in obtaining high rankings with that kind of content you WILL have people filing spam reports on you and (if you're running Adsense) you'd be at serious risk of losing your Adsense account - among other things.

Sounds like a fun project though - report back when you can auto-generate articles that actually make sense.

 

[mrjohn]

Spam filters are now targeting the underlying template. You have to go way past natural language now to trick out the spam filters that Google is using. Even with meaningful human intervention, the articles will get filtered if they have an underlying template they're being built from.

 

[mhdoc]

the articles will get filtered if they have an underlying template they're being built from.

Mind elaborating on that a little? I'm thinking for example, all the wordpress sites going up.