#data categorization

[matalab]

Let Artificial Intelligence do the tedious job, while you are drinking your coffee and thinking about your business strategy! https://matasoft.hr/QTrendControl/index.php/un-perplexed-spready/ai-driven-spreadsheet-processing-services

https://matasoft.hr/QTrendControl/index.php/un-perplexed-spready/ai-driven-spreadsheet-processing-services

#DataStrategy #AIDataSolutions #Analytics #PracticalAI #AISuperiority #DataProcessingPro

#BusinessIntelligence #DataSolutions #Automation #Speed #Efficiency #AISolutions #AIrevolution #SpreadsheetInnovation #AIProductivity #DataManagement #DataTransformation #AISolutions #BigData #AIProductivity #DataAnnotation #AIDriven #DataMagic #TaskAutomation #AIformulas #Productivity #DataCategorization #AIprecision #DataManagement #DataCategorization #DataClassification #DataLabeling #DataExtraction  #AIinsights #SpreadsheetMagic #UnPerplexedSpready #AITech #AI #DataAnalysis #DataTransformation #AIFuture #DataProcessing #AISolutions #DataEfficiency #Spreadsheets #AItransformation #SpreadsheetSolutions #Automation #ErrorReduction #AccuracyMatters #AIDataHelp #FastSolutions #Productivity #SpreadsheetHelp #NLP #LLM #DataEntry #ArtificialIntelligence #Spreadsheets #MachineLearning #DataAnnotation #DataPrep #BigData #DataProcessing #Automation #Matasoft #Accuracy #MachineLearning #Ecommerce #HealthTech #FinTech #Marketing #Software #Innovation #Matasoft #SpreadsheetAutomation #AIpoweredTools #EfficiencyWithAI #ExcelAutomation #GoogleSheetsIntegration

[sillylotrpolls]

About Reblog Graphs

Have you ever clicked on the "reblog graph" button of a post? I think they're one of the... well, maybe not greatest features on tumblr ever (polls are probably better), but they're still pretty darn neat.

I want to show some cool patterns I noticed on some recent posts of mine, but first I'll explain how reblog graphs work so you can more easily follow along.

This is pretty long with a bunch of pictures, so click the cut to read more.

How reblog graphs work

If you've never done so before, I invite you to click the notes button on this or any other post and then the icon with four circles. You will then see a bunch of dots connected by lines.

For example, if you click the graph for the "blorbo in Elvish" post, you get something like this:

Now, let's zoom in a bit. You can do this by using the mouse wheel and clicking and dragging around the graph until it's showing what you want. (I don't know how it works on mobile, but presumably it's similar to using Google maps?)

This next screenshot is the bit in the lower right of the graph shown above. However, the graph may not always display in the same way because reblog graphs are re-generated each time you click the "reblog graph" button.

Here you can see that I'm viewing the root post, which is the original post made by me. It's indicated by a circle with a dot inside. You can also see that six people reblogged that post. Each reblog shows as a dot with a line connecting it to the post it was reblogged from.

Now here's a cool thing about reblog graphs: they're interactive! You can click on any dot and see the post it represents and the reblog chain that led to it.

For example, clicking this dot that has several lines emanating out from it shows that it is "2 reblogs deep" and was posted by @cycas.

Got it? Close enough? Cool, now let me show you some neat things I noticed. :D

The Swedish Chef poll and very popular bloggers

My polls tend to average between 500 and 2000 votes, depending on subject matter. The Swedish Chef poll, however, took off and eventually garnered over 22,000 votes. How did that happen? A very popular blog reblogged it about five days in.

Initially, the graph looked like this. (This is the first 200 reblogs.) There's nothing unusual here. You can see that the root post had several reblogs, and that there's another cluster developing around a post by @zagreus. There are also several reblog chains where just one person reblogs someone else's reblog. Some of these chains peter out, while others find their way onto the dash of more popular bloggers, creating clusters.

A quick note about "popularity"

Yes, yes, it's all about "popular." However, it's not just about having a bunch of followers. What's more important is that the "popular" person reblogging your post has followers who are specifically interested in your post.

For instance, if I, @sillylotrpolls, make a poll about LazyTown, it's probably not going to get very many reblogs because my followers aren't here for that. However, if @silly-lazytown-polls reblogs the poll, that reblog might then get quite a lot of reblogs itself. It's not that silly-lazytown-polls has more followers than sillylotrpolls, it's that it has more followers specifically interested in LazyTown content. Make sense?

Back to the Swedish Chef poll

The poll eventually got over 5,000 reblogs. Since you can only add 200 reblogs to the graph at a time, you can roughly see how a post spread over time.

With 600 reblogs loaded, a new cluster bursts onto the scene. This is @bunjywunjy, who reblogged the post from @beecreeper who reblogged it from @soggypotatoes who reblogged the original.

Bunjywunjy didn't add any tags or comments, so I didn't even notice at first because it didn't show in my activity feed. However, I did notice a sudden uptick in notes on the post, which caused me to investigate. It had been five days since I posted the poll, and usually polls that are going to take off do so sooner than that.

By continuing to click the "load more reblogs" button I can see how the post further spread, especially from bunjywunjy's post.

When the post reached @beggars-opera (whose icon I am somewhat proud to announce I identified on sight), they added a screenshot of @stylishanachronism 's tags which said:

# all of these are incorrect it's the 'meat's back on the menu boys!' scene

This would become the dominant version of the post as it further spread. Interestingly, this was the only reblog of stylishanachronism's reblog. Literally thousands of people loved their tags and agreed with them, but they quite plausibly never saw it unless they specifically went looking.

By 3,200 reblogs, you can see even bigger clusters developing. @thebibliosphere shows up 10 reblogs deep, and leads to yet another cluster via @teaboot (12 reblogs deep).

Eventually, with all 5,371 reblogs loaded, the reblog graph looks like this:

Like I said: neat. :D

Cool, but if you've seen one, you've seen them all - right?

So what prompted this (extremely long) post was actually the reblog graph for my poll on inspirational LotR quotes.

Here's the reblog graph with 200 reblogs loaded:

And here's the graph with all 1,890 reblogs loaded:

It's just one big cluster around the root post. I've never seen that before!! Almost everyone reblogging this post saw it either because they follow this blog, saw it in the #lotr tags, or because their non-influencer friend reblogged it. (Or maybe it was in some kind of algorithm/the explore feed, but I have nearly zero experience with those.)

And this wasn't just a small post. This poll got over 15,000 votes and more than 4,000 notes. That puts it in the top 10 polls for this blog.

What does it mean? I have no idea. I would really like to know! But really, I got nothing. If you have a theory for why this particular poll should result in a reblog graph like this, I would very much like to hear it.

Orphan clusters

To round things off, I'd like to show another interesting facet of reblog graphs: orphan clusters.

This blog's current undisputed poll champion is the fmk wheel poll. That's not really a surprise, as it combined sex with a fun game where you just had to tell everyone what you got, which meant either a reply or a reblog. So it spread pretty far.

However, if you look at the graph, there's something odd going on.

This is with just 200 reblogs loaded:

Notice how some of the dots don't connect to the root post? That's because somewhere along the chain, a reblog was deleted.

This cluster in the bottom left got pretty big! This screenshot is at 800 reblogs loaded. The missing link is from a blog called @gendertaliban that doesn't exist anymore, as near as I can tell. That makes it impossible to trace the full path of any of these reblogs.

In conclusion

This concludes today's deep dive into a tumblr feature you probably never paid any attention to. Admittedly, there's not a huge use for it outside of determining which of your mutuals is an "influencer," and they get quite difficult to navigate after loading about 1000 reblogs, but I hope you enjoyed staring at dots and lines with me. :)

[catherinadodson1976blog]

[artemistorm]

I looked at the information that youtube had on me and was using for ad tracking (I have adblocker so...) and youtube thinks I work in real estate for a 'big company' 😂. Like what. Where did they get that? I'm unemployed. I've never worked real estate and I don't want to have anything to do with it. Youtube also thinks I'm female, not a parent, am upper middle class (apparently they judge you on what class they think you are??) and some categories had "not enough information" to make a judgement.

[froggywentaprincin]

At this point I’m invested in these statistics

If I did all the (teen) characters with polls about what elemental powers they would have, would you vote on them?

Would you be annoyed? I won’t spam them, maybe just 4 a day until they’re all done.

[un-pearable]

this is yet another random academia nitpick i won’t let go of for weeks but someone in the comments of the victor ninov broccumentary claimed that science is a “incredible field that can’t be simplified to storytelling like this” because “it goes against human nature to do” and writing good sci comm goes against scientific integrity.

and the whole statement is incredibly stupid but in true anthropologist fashion i must say what on earth about science goes against human nature. the desire to test a phenomenon and revise your theories based on the results is literally one of the benchmarks for early modern humans. this is the behavior in corvid’s (tool use) that people loose their shit over. communicating findings to others?? collaborative work to reach an arbitrary goal that won’t necessarily have a direct benefit on your immediate life???? that’s how we domesticated staple crops dumbass. it’s science all the way down

[venacoeurva]

If Tumblr DOES end up scraping our data for training it's irritating but at this point I'm just so tired of this shit I almost can't bring myself to care, really, which isn't helpful but what can I fucking do. I rely on social media to survive, and we keep having to move or the risk of having to move. Other sites will do it, other sites did it, and even if Tumblr decides not to sell data to midjourney and such or this is some advanced hearsay, scrapers would access our blogs anyway as they do any other site, so it feels kind of just... moot regarding training until actual regulation and legal precedents are set in place?

That being said, I wonder if it will fall through if it hasn't already with the implosion of Matt being himself and people lawyering up against him/Tumblr, and the scraping plans and data selling could very likely be a data privacy violation in at least some places

[angstics]

sorry im still on this but i wanna say it. i really like most of gerard’s makeup from the olden days but i dont see it as subversive to my ideas of gender norms. it was common for men in rock n roll to wear eye makeup. people may see that as subversive in dominant culture, and it def got some people mad, but ultimately doesnt mean much to me

[dollsahoy]

the bulk of the unattributed generative images I see (the kind that have notes full of people posting "it's AI") are room interiors

Granted, I have not followed decorating trends, but I somehow doubt that interior decorators are losing money to people who are prompting for images of rooms instead of going and hiring someone--or spending money themselves--to decorate one side of a single room for a single photo

and I feel like this is more a case of people just not thinking about photography and interior decoration as art in the same way they think of drawing and painting as art

so they are quick to post "it's AI" not so much in defense of photographers and interior decorators as they are because they don't like how generative tools are trained and used, while--in some cases, but of course not all--not even thinking about photographers and interior decorators as artists they see harmed by generative images

(and it is in extremely poor form for people to post generative images without saying that's what they are, but I suspect some of them are copying and reposting things that they don't realize are generative, either, because they were already unlabeled. it's extremely poor form to post ANY unattributed images, of any kind, and I am personally trying to be better about that, even when the images are official product photos) (and NO the generative images are NOT themselves unlabeled art reposts, because that's not how they work)

I dunno. The bulk of my "art" is creating three dimensional things and photographing them, and sometimes the results are more "obviously artistic" than others, so I may be taking this a bit more personally

[palin-tropos]

earlier tonight I had a little rant at @excalibutt about how disco elysium, if it’s a world made of information and not matter, that is running out of memory space, hence the pale… that explains the reincarnation concept of mazovian yaoi lmao. the world is trying to create new material, and stave off the end, but it’s saving space by reusing/recycling some old information, so you get repeating patterns

[if-loki-was-a-fox]

Me when I like something: "I wanna make diagrams and charts of this"

[unnonexistence]

ok we are doing Pacific Rim Data Visualization live on my blog tonight because i can't help myself

I'm going to put this post in the main pacrim tag in case anyone is interested, but i don't want to annoy anyone, so everything else is just going to be in my #unscientific aside tag. okay? okay.

basically i am making graphs of stuff hermann would have been making graphs of (& predictive models about).

so i'm thinking one nice way to show the increasing frequency of kaiju attacks over the years would be a bar graph along these lines:

2013-2025 is a pretty good time span in terms of number of bars, so it'll look nice, and it'll clearly show the increase over time. i might leave 2025 out because the movie takes place in january 2025 so it'll look funny to leave it in (only 1 attack pre-movie, i think).

now this would be a very broad-strokes graph. this is the kind of graph that just kind of says "hey there have been more and more giant monsters emerging from the ocean lately. that's probably bad". i want to do something with hermann's predictive models, & this isn't going to work for that.

i could do a more specific one, e.g. by month, but even for 2024 when there are a stupid number of attacks in the timeline, that's not ideal. some months are only going to have one or two & it'll look funny. also, his model gets down to "number of minutes between attacks", which is way too specific to use a bar graph.

for the predictive model, when i think about "increasing frequency of kaiju attacks" & hermann's speech about it*, what I picture is a line something like this:

ignore the scale, i just put whatever. the point is Number Go Up.

the problem though is that means the y axis needs to be number of kaiju attacks, and the x axis needs to be time. which like... idk how to make that work? this is probably just because idk what i'm doing, but like... the attacks are happening at distinct points in time. how are we defining "number of attacks"? it would have to be number of attacks over some timespan, right? i guess it could be Total Number Ever but that seems weird.

this maybe should be a dot plot of some sort, but the thing with the axes is throwing me off. I'm going to work on some of the actual data stuff and report back, but at some point i probably need to do some research about the types of graph i can make & what they're used for.

if anyone has ideas or otherwise wants to talk about this PLEASE let me know. nerd bullshit is more fun with other people.

*you know, the one that got meme'd? "In four days, we could be seeing a kaiju every eight hours until they are coming every four minutes."

[beesorcery]

eyes emoji @ your pacific rim spreadsheets. may I see it 👁👁👁👁👁

:D eheheee

[waitineedaname]

oh on a similar topic to that reference site with all Ed's outfits. I am eight episodes into my spreadsheet collecting all appearances of recurring characters in fmab

[likedbyuarmyhope]

me when i make a spreadsheet of every single bts song (from group albums) and assign each song a category based on theme/message and then calculate how many songs are in each category and what total percentage of bts songs each category makes up

[cheshire-castle-library]

Jan 14, 2023

(Yes I'm prepping this before the day rolls as shown by the timestamp, but I'm very excited and I think I should be allowed to capture that.)

How come I always forget this little thrill that comes with coding? Making the electric rock hand me the letters I just gave it just gives me such a pure and innocent joy. How do I always forget this feeling? Its been about 4 years since I've played with Python, and I'm doing the song and dance again, because I just don't remember shit for shit. But believe it or not, 4 years ago I made a code that read faces in a webcam feed and output a servo motion for my senior project.

But in 4 years I've learned maths arcane and electrons unknown and probability in 3+ dimensions. But now I'm going to be teaching with Raspberry Pi's and I need to revive these very dead brain cells.

Anyway, today was watching DOTA and Vox Machina with my fam while semi-recreationally trying to re-learn Python some more. I technically have to do this to be able to teach it in a couple months, which... I mean I know I can re-learn it, I re-teach myself coding every time I need it so that's fine. But I absolutely don't have to be going to the trouble I am just for the job. This has crossed over into fun, and I'm honestly feeling more refreshed playing with my Obsidian vault and python than I have in like a year.