Text analysis

Sun, 03/16/2014 - 22:00 -- BlueWinds

So, I was kind of curious about this or this, given that many people online are surprised to learn I'm a woman. They (those two text analysis tools) generally classify me as male.

But then I tested them out with some other sample texts - various articles from Echidne, Gretta and The Journey Outside (Trudy Capote's game). The text almost universally registers as "male" - even the very touchy-feely post Trudy made about the death of her father.

So if three writers that I like and that are most definitely women all register that way... the tools must not be very good, and I shouldn't pay attention to anything they have to say about me.

:shrug:

Back to poking at the code for whoring income.

Comments

Frankieo on

Tried both with a bogus story (not even half finished at best) i came up with on-the-go just to test them.First one said: 

Genre: Informal  Female = 322  Male   = 344  Difference = 22; 51.65%  Verdict: Weak MALE

Weak emphasis could indicate European.

Genre: Formal  Female = 425  Male   = 296  Difference = -129; 41.05%  Verdict: Weak FEMALE

Weak emphasis could indicate European.And last one gave me: 1. female (82.1 %), 2. male (17.9 %).

I'm a dude, and a Norwegian, so European is correct xDHere's what i wrote if you're interested in a little read. 

"One day along a dusky road, there was a lonely child named Juliette.Juliette was a simple little girl, at only 13 years old, and standing at 5 feet tall, and dressed in raggedy old garments, that looked like they had been worn for half her life.Where she was heading to, or where she was going from was unknown, all but the telltale hint of a few bruises here and there. She might have been mistreated, or yet much worse.

Upon a ridge on a small, but steep hill. Came wandering a teenage girl - who's clothes weren't much better than Juliette's - just 17 years old, she was but a peasant's spawn and had run from slavery, after she had been sold. This young girl who's name was Elizabeth, was but 5.5 feet tall and athletically built, yet petite she was.She greeted Juliette with an exhausted, yet friendly smile, and asked where she was headed to. After receiving an answer, whereas Juliette said to Elizabeth that she was just traveling alone this lonesome road, no destination in mind, just looking around. Elizabeth couldn't help but ask for acceptance to travel with, which she got, and with a smile at that.

Many days gone by, since Juliette and Elizabeth met. Over steep mountains and low hilltops they had been, upon a cave they walked. Whence they looked at each one another, quizzically asking what to do?Exploring they decided, which they did, and dark they found it to be. They each had no light, nor torch or flashlight at hand, and none of them could see two feet ahead.They decided to brave this dungeon, seeking some fun or refuge - maybe even some treasures if they were lucky. Upon reaching the very end of the dungeon, they did not know what to do, deciding to walk back they did.Along their way out, they found another"

Abrubt end, i know, sorry :P

Terryble on

It is theoretically possible that the people you like reading all write 'like males' and that is part of what makes them attractive to you. Theoretically possible, not very likely. Personally I've never been very moved by claims of detecting gender or personality or pretty much anything else from people's writing, handwriting, signature, etc.

Arcanis (not verified) on

My first post =P

 

I put your post here into the first to gender-analyzer and the result was:

weak male/weak female and 76% female from the second one.. better results? :D

 

...Then again I got weak male and 96% female when I put in one of my emails -while being

male... now I'm depressed because I seem to have a girly writing style =/

 

But seriously, I just wanted to say that I really like your site and that your games are awesome.

They may be porn-themed, but even without that they would be very very good imho =)

 

Have a nice day =)

ThatGuy on

I used two blog posts of mine on both analyzers.The first post in the first analyzer said I was ~60% for both sections. The second post in the first analyzer said I was ~70% female and ~50% male.I put those same two posts to the test with the second analyzer, and both times I was given ~80% female.I am a dude. Apparently a girly dude, though. >_>

Coga on

Hacker Factor guessed my gender as 'informal', and my nationality as 'European', while UClassify put me squarely in the 'female' section.

 

I'm 100% male, and living in Southeast Asia.

 

I don't know- while I will admit my literary style is influenced by authors like Terry Pratchett, Dan Abnett and a little Orwell, I don't know why I'm apparently the wrong gender.

Coga on

Addendum: Tried another sample of my writing, and yep, still a girl. Off to find some scissors, I guess :P

Benedikt (not verified) on

Why do you care if a tool defines you by your writing in a way you do not like ? Rather you should be glad that you are good at it, enjoy it and know that others like what you do. Just do it like Spongebob.

BlueWinds on

I don't much mind one way or another, but I was curious. My thoughts went something like this:

Lots of people assume I'm a man. Perhaps my writing style is masculine? I know, I'll use an automated tool to find out. Huh, looks like it agrees. Let's check the accuracy of the tool (like a good scientist). Nope, tool seems to have error bars too large to be of any interest. Curiosity remains - do I write "like a man," or are people just assuming that based on past experience / outdated notions of who likes porn?

Natas Baker on

I went off the fact that you said you had a streak for turning women straight after you date them :P That uhm normally a man is proud of. The rest of what I went off of was shaky at best.

benryher (not verified) on

I haven't paid attention to developments in that technology for a few years, but it at least used to be that it worked mostly just by counting word frequency and word couplet frequency... and stuff generally like that.  Since people write in terms of phraseology depending on their influences -- from both speech and text -- all it's really doing is matching whether a writer's influences better statistically match those of the average male or female writer. If someone were to primarily read writers that are commonly-read by women who then later become writers and also talk (let's say in their upbringing) to with groups of people more-commonly talked-with by women, then you would fully expect them to be identified as female regardless of their gender.

Coga on

I think you've got it in one- your gender doesn't affect how you write, but the gender of the people you read. That being said, that would STILL make me an outlier as just about every single book I read was written by a male author- maybe they read a lot of female authors before?

Iraleth on

I fed in a short (600 words or so) smut piece I wrote a few months back, just to see what would happen. The Hacker Factor analyzer declared it 76% male, whle the uClassify one declared it 98% female.I feel safe concluding that whatever they're doing to analyze the text, it's not the most reliable method.

Token (not verified) on

I don't really think about gender on the internet to be honest, when somebody posts something online I see "this person said this" not "this bloke, or this lady said such a thing". Although saying that if I had to pin a gender on someone I would most likely say this person is a bloke, however that fact is probably skewed by that fact that I read everything in my own voice in my head. Now seeing as I myself am a bloke with a particularly northern (british) accent its hard for me to attribute feminiminimity to what I read online, hmn thinking about it I'd rather everyone on the internet remain as just people, its much easier that way!!

Meridon (not verified) on

Trans woman here.  The first tool said if I was writing informally, I was male.  If I was writing formally, I was female.  I find the distinction odd.  I'm really curious how it works.  Regardless, it was formal work.  The second tool was more than 86% sure I was female.  I'm often believed to be female when people judge by my writing alone.  That kind of makes me happy.

BlueWinds on

The first tool just has a list of words for each gender, and grants points in the M/F column for each. I've taken a close look at its source code, and the word list is rather small. "The" is considered a masculine word.

The second tool likely works in much the same way, but it's run on their server, so I don't have the source code to take apart.