Skip to content

Bug report

Hello Sven,

Before starting, a big THANK YOU for this powerful tool that is GSA CG, I only recently acquired it but I am already very enthusiastic about the uses I will be able to make of it!
And above all, CONGRATULATION for the exemplary follow-up that the tool has been undergoing since its launch, as I understand that updates are regularly made to push the tool even further (+ the fact that you are particularly attentive to users' requests).

As I got to grips with GSA CG and the tests I was able to perform to better understand the software, I encountered several problems, and was unable to produce truly optimal results. I have several ideas to propose but first of all, here are some bugs (or misuses ?) I encountered during my sessions :

1) Foreign language content in the text : I generate texts in French, but my test theme being "marketing" many technical words are in English. So, I guess the tool has scraped foreign language contents. Unfortunately these elements were not filtered in the final rendering, and I end up with paragraphs entirely in English.
Is it possible to filter these unwanted contents ? Or, at least, offer the possibility to automatically filter these content once the generation is complete, even if it means losing a bit of time in the process ?
Here is a resource that might be useful, after some research on the subject (even if I don't know in which language you coded the software...): https://github.com/landrok/language-detector


2) When I ask the tool to generate exactly 5 anchors in my texts, I end up with less than 5 anchors in the content (between 1 and 5 anchors to be more precise), even though I didn't write "1-5" in the concerned field.



3) While I specify that I don't want H1 in my texts (see screenshot), I often end up with H1 at the beginning of my articles.
Moreover, these H1s are followed by H3 (which is not logical from a structural point of view).



4) I don't know if it's a bug or if it's something intentional, but empty paragraphs appear in my generation.



5) Although "Add first image as Wordpress Post Thumbnail" is checked, "wp-post-image" is not attributed to the first image in the article, but to a random one.



6) The alt tags of the images are almost all the same, instead of digging into the keywords and making the alt tags as different as possible from each other.

7) I probably misunderstood the function of the "Title with keywords" option, but when I enter additional keywords in the appropriate field, these keywords never appear in my titles.




Thank you very much for the time you spent reading this report, I wish you good luck for the rest!

On my side I will try to write down my ideas to flesh out the software :)

Take care !


Comments

  • SvenSven www.GSA-Online.de
    1. Language detection is happening by stop words and a lot other data. I would need the project backup with the wrongly build articles to debug it and maybe improve things. Using that language-detector is probably working better but I think the problem is something else here.

    2. It tried to place as many links as you specify, sometimes your other settings simply would not allow it or the text would be full of links in one place which looks unnatural. However, again a project backup is needed to give details.

    3. I see the problem and fixed it for next update

    4. Hard to tell what this could be but usually some "html variation" inserting empty paragraphs. If you have that turned off, please once again, send me the backup of the project.

    5. fixed in next update. Previously it inserted the first image at a random place and gave it that attribute, that wasn't always the first image displayed.

    6. alt attributes are defined in the image-listing (title) and are the default ones from scraping

    7. See that as a filter. It will accept found titles only if they match one of the inputs. Else it will try to generate other titles (titlegen.dat in your lanuage required).
    Thanked by 1Minidou
  • Thank you Sven,
    I PMed you the backup :)
  • SvenSven www.GSA-Online.de
    1. The issue should be fixed in next update. The article extraction based on it's html structure was improved.
    4. Fixed it in next update. It was a <p> or <div> inside another <p> tag which is handled as a closing <p> and an empty <p></p> when rendering.

    Thanked by 1Minidou
Sign In or Register to comment.