Incase you care

[metaiwan]

I have plans to develop an application that you can throw text in and it can tell you with some accuracy the type of the person who wrote it. I plan to do this by creating an immense sql database on my computer and feeding it large texts of people whose type is known(like william james, einstein, etc). then pulling statistical data from this and formulating a formula andor using this database to do probability matrixes
so if you know the types of authors lemme know
especially if i can get their typed text online free from guttenburg project

thanks

[CuriousSoul]

I could look into it. I think it often is definitely possible to tell the writer's type based on the way they write. I could try to find you some examples I feel very confident about. The potential problem though is that such a statistical analyzis would presumably have to rely exclusively on vocabulary and that tends to severly limit the amount of information you can analyze. The results can often be quite interesting. I remember one test which claimed to be able to tell whether the writer was male or female, and most people who tested the program ended up being told their sex is not what they thought it was.

[MysticSonic]

Well, it seems reasonable to use one's text to determine whether or not they have a preference for inuiting or sensing, as I have seen texts that commented on the stark differences in context between the two. However, I have also heard that using one's language and the manner in which one speaks is still an unereliable tool for type indentification, even for Socionists. Of course, don't let that discourage you, as this sounds like an extremely interesting idea.

[signum]

However, I have also heard that using one's language and the manner in which one speaks is still an unereliable tool for type indentification, even for Socionists.

I believe the type identification by content analysis is reliable. However, as far as I know, there aren't any researches in this field with native English speaking persons.

[Maestro]

What would be nice is to overlay a trained neural network on the vocab database. The trainer would have to be an expert typer with near 100% accuracy. (whatever that means. correlating different trainers to remove all but systemic bias?) The results might be much better than using a linear statistical approach.

[LevKamensky]

...

enfj

hegel
heraclitus
anderson
mohammed!

...

istj

marquis de sade
salvador dali
doc love (tomas hodges)
miss manners (judith martin)

...

infp

esenin
gumilev (another russian poet)
jean-jacques rousseau
st. juste (comrade of robespiere)
marc chagall (painter)
picasso
vertinski (russian singer)
julius caesar (maybe!)
cicero (definitely)
mozart...
jeanne of arc
...

estp

napoleon (memoirs)
roerich (russian artist)

...

intp

balzac
marquez
chesterton (maybe)

[LevKamensky]

it's a bit difficult
because most great people were on the borderline between different types...
we have to pick "bright" representatives of their type

[Jimbean]

Very rarely do we ever know for sure what type a famous person is.

[metaiwan]

this is true. i have given up for now.

[hock0526]

This idea is brilliant... if it works... now, quick - what type am I from that?

[metaiwan]

An intellegent yet sardonic type of person???

Depends, are you Eric or Robert?

[drake]

I think that it's impossible to determine one's characteristics from the type one wrote only by crude logic. Computers don't have humans feelings, experiences, vision, prejudices... How would you analyze all that from a sheet of text or from an entire book? What patterns would you search for?

[metaiwan]

Well, my idea was to try different methods, see how well some worked, possibly using two methods and then scoring them together based on their probability matrix. Basically the simple version would start off with just words. More abstract words for intuitives, more concrete words for sensorics. Basically, I would postulate methods then test their result matrices etc etc etc

[Slava]

I wouldn't do it base don famous people. They tend to check their work for spelling and phrasing more than average joe shmoe types.

here are mroe factors:
Type of work
Use of the words I, Myself
# of each word used
# of feeling words (... swiftly scraped my chin, corressing bla bla bla ...)
# of technical words (... made a linear bisection, although, however, the latter, technically,. hypothetically ...)
# of similies/metaphors (... as if, like, as, (verbs that arent related to the noun, i.e. "{I} {flew} over to the store ...") ...)
# of sensory words (heard, saw, smelled, olfactory, felt {noun})
amount of peoples names used
Other words that match temperment traits, common typos
Asside from statistics on the word itself i would also do word transition statistics.