RE: Introducing Hive Report Card: A Tool to Analyse Content Complexity, User History, and More!
You are viewing a single comment's thread:
There was a design oversight in the tool. It was taking into account any markdown, URLs, or links to images. I've just fixed it, and tested it on your account:
As you use a lot of images on your post (18.22 on average!) this increased your writing complexity unintentionally due to my design oversight. It should be more accurate now that I have addressed this issue.
0
0
0.000
Aha, that makes a lot of sense. I do use a lot of photos. Looks much better now for sure.
Thanks for the adjustment.
I like this tool, and will use it occasionally to track my posts and writing.
I am the look out for other bugs and issues, so let me know if others come up in your use of the tool!
Thank you for your interest in it, it is really appreciated! I am sure this will be useful to everyone on Hive!
Ok will do.
If I check my last post my word count is 1324 with 29 photos in the post itself.
Looking in the tool it says in the post history (newest to oldest) section 1 to 3 (2 cross posts) of that post the word count is only 42 (x2) with 0 photos and at 3. Only 239 words with 65 images, it looks a bit off?
You are welcome, happy to help and try new tools, to report back.
Good catch. Cross Posts are technically posts on the chain. I will see what I can do to ensure that these are filtered out. I don't cross post much (if at all) - so I did not see this come up when I was testing my own data. This will make the tool even more accurate :)
Working on a solution now, which will likely be "If duplicate titles, de-deuplicate by retaining the earliest (oldest) post, with the highest word count... (Probably using the perm link)
I'll use your account to test my fixes further :D
Great, let me know any updates. :)
Would be interesting to see the changes.
Implemented a fix to filter out cross posts, using regex (character manipulation!)
If the permlink has hive-[xxxxxx] where the space between the square brackets are 6 digits, at the end of it; exclude the cross posts. Looks like it worked in my local testing, but please verify and let me know if it is doing anything strange. The change is live now :)
I just checked it out, it got a bit slower to fetch all.
It looks totally different now, especially in the readability scores. (Ps. english is my third language of learning but shouldn’t be taken into account)
How come there is none professional included now? By removing just the cross posts there should be some still left, right?
I don’t know how many posts I have written to be honest or how many cross posts, if the cross posts are filtered out: It dropped half in the tool.
Further in the post history it still says 239 words for the last post, but the post itself is 1324 words.
You work fast ;)
I’m back to packing some boxes now… we are moving next week.
Will reply later. Have a good evening/ night. It is morning here.
LLMs help a lot with code, once you know what the problem is. Will investigate the word count issue further.
I saw you solved the word count problem. Great job.
It still has issues, I've been working on them all day :P I will publish a post with some of the more major improvements later as an addendum to this one, because there are far too many people to reply to individually with updates to each issue :P
I noticed. The image count seems to much for each post.
I do like that you added replies and engagement. That is very useful.
Haha I can imagine. Lots of people helping too to detect bugs.
Cool looking forward to that post.
Enjoy your evening.
I just rolled out more changes, made a new post about it. I documented the image count issue, if you can give me an example of your recent post that has the mismatch on my new post I will investigate that some time next week.
My next week is very busy, though - so improvements won't come nearly as quick!
I will have a look around kater today or tomorrow how the image count is and let you know.
So busy… so I understand fully.
Have a wonderful morning.
Made improvements to the word count logic. More regex!
It doesn't match exactly with peakd, but it way more accurate than what it was before.