IBM Watson written text personality analysis

[Juraj]

Brother forwarded me this link and it's pretty fun. I also believe it kind of works well, we both got diverse, but very fitting descriptions and percentil comparisons.

It's definitely not one of those "well, this could fit anyone" ambiguous descriptions you're used to.

Also, don't be shy to share your desc :- )

https://watson-pi-demo.mybluemix.net/

[FrostKiwi]

Also, don't be shy to share your desc :- )

That sample text made my pitch black humor rejoice.

I kinda lost it at buy healthy foods.

[lacilaci]

Brother forwarded me this link and it's pretty fun. I also believe it kind of works well, we both got diverse, but very fitting descriptions and percentil comparisons.

It's definitely not one of those "well, this could fit anyone" ambiguous descriptions you're used to.

Also, don't be shy to share your desc :- )

https://watson-pi-demo.mybluemix.net/

at least 3500 words....? Alright let's try to shape my personality into something great... See if the software can see through my game :D

[Juraj]

No, you can do with away with hundred. First thing I feeded into it was my forum drivels. It was 'weak analysis', but not far at all from more complex text.

[Ondra]

my master thesis info produced just random guessing, nothing I could relate to.

[lacilaci]

I know nothing on how this thing operates...

correct me if wrong, but I think it requests a LOT of text related to personal expressions from individuals to look up occurences of key words which are otherwise interchageable witch another words without breaking the point made but the person deliberately uses a certain specific words and phrases often... also the choice of Topics and connections etc... I mean it can go deep... But not fool proof and actually a "fortuneteller" or a "witch" LOL :D could do just as good with some primitive cold reading techniques (Oh there's a wedding ring... you married or were married mister??)

So if you give a bare minimum as an input it is nothing... Especially if it is stuff from one timeframe or stuff written for one purpose etc...

If you want to play fair. You need to expose your private life and opinions and give it a lot! Aaand still I would count this to a low level fortuneteller analysis :D cause by chosing the "right" words and topics you can shape your own image into whatever you want.

Still was fun to play with this online horoscope though as long as I can make it tell me "you are self-conscious, uncompromising, independent, empathetic, making your own path" and that's with me telling openly the software that I'm going to make it tell me what I want... just a correct choice of words... :D a fortuneteller would do better job and a psychoanalyst would decimate me if I try to play with him...

BTW. the software can tell a LOT about me from just being "happy" again and again and again..

[Juraj]

You're trying way too hard :- )

You could break anything easily if you try, doesn't really negate it.

[lacilaci]

You're trying way too hard :- )

You could break anything easily if you try, doesn't really negate it.

Alright I admit.. it gotten out of hand at a point :D... And no, it doesn't negate it. I do buy eco-friendly (although it's my girlfriend's decision) :D

[Juraj]

If you want to play fair. You need to expose your private life and opinions and give it a lot!

Not at all. It analyzes your use of language, not the content. So the result will be way more consistent if you simply input your broader use of language. Speeches, comments, columns, essays, etc..

I do presume the more the text is written ambiguously in objective way (like Ondra's thesis), the more it will be random.

After all, the model behavior is sampled from Twitter usage :- ) Which is raw, opinionated, personal and subjective way of writing for most part.

Interesting that IBM intends to use the broader model for corporate business to asses their clientele. I am gonna sample some emails for fun

[lacilaci]

If you want to play fair. You need to expose your private life and opinions and give it a lot!

Not at all. It analyzes your use of language, not the content. So the result will be way more consistent if you simply input your broader use of language. Speeches, essays, etc..

Hmm.. But let's say a speech could be designed for an audience so the choice of words and topic would be audience-targeted. You could use it for a judgement in a way but if all the software sees is text it's not much... I get your point that there is something to gather and it can have it's use.. But results are more reliable about the stuff like "He's prone to buy eco friendly: target for an ecover ad".... like a tool looking for new potential clients

[Juraj]

But let's say a speech could be designed for an audience so the choice of words and topic would be audience-targeted

Very true, careful preparation and certain internalized pretentiousness can go long way in fooling not only computers, but people alike. Otherwise, alibistic bullshiters wouldn't be so popular.

It's also why it suggests personal writing (doesn't men self-centered (takes away the pretentiousness) and longer (the longer you have to speak/write, the less will you preparation last and you will speak more naturally, creating certain consistence).|

If you would let Donald Trump speak for 24 hours straight, even his most derailed supporters would come to realization he is utmost imbecile. Or not :/

But results are more reliable about the stuff like "He's prone to buy eco friendly: target for an ecover ad".... like a tool looking for new potential clients.

I actually think this is the weakest part of the tool, I think they should have skimmed it since it seem least developed. Look like they only created few of those roles and simple correlation with few of those stats.


But by all means, it is of course arbitrary and flawed on many occasions, this isn't application that was developed by solidly large science team for eternity.
To get the most out of it, you have to go with the flow and take it for what it is :- ).

[lacilaci]

But let's say a speech could be designed for an audience so the choice of words and topic would be audience-targeted

Very true, careful preparation and certain internalized pretentiousness can go long way in fooling not only computers, but people alike. Otherwise, alibistic bullshiters wouldn't be so popular.

It's also why it suggests personal writing (doesn't men self-centered (takes away the pretentiousness) and longer (the longer you have to speak/write, the less will you preparation last and you will speak more naturally, creating certain consistence).|

If you would let Donald Trump speak for 24 hours straight, even his most derailed supporters would come to realization he is utmost imbecile. Or not :/

But results are more reliable about the stuff like "He's prone to buy eco friendly: target for an ecover ad".... like a tool looking for new potential clients.

I actually think this is the weakest part of the tool, I think they should have skimmed it since it seem least developed. Look like they only created few of those roles and simple correlation with few of those stats.


But by all means, it is of course arbitrary and flawed on many occasions, this isn't application that was developed by solidly large science team for eternity.
To get the most out of it, you have to go with the flow and take it for what it is :- ).

I can see it's use maybe in tandem with other analysis tools just to give something extra additional and maybe there is an advanced not public version that can do a bit better.

Derailed supporters of trump would cry that the software took a human job away and it needs to be destroyed before it becomes skynet and opens borders to immigrants and shuts down everyones guns

[burnin]

Hahaha, another make-mo-money fast idea:

Devise a fortune teller and a joker. Make an app (IOS, android, facebook...). Minimalistic colorful 'gipsy' design.
Enjoy life. :D

[Ondra]

Hahaha, another make-mo-money fast idea:

Devise a fortune teller and a joker. Make an app (IOS, android, facebook...). Minimalistic colorful 'gipsy' design.
Enjoy life. :D

stopping corona development right now. Next stop: best startup ever!

[burnin]

stopping corona development right now. Next stop: best startup ever!

Could you implement it in Corona ;)
A middle-man app feeding 'daily emails' text to the 'fortune engine' (& NSA) while waiting a render to finish.
Also 'connected' to the espresso machine so that when one hits render, gets a daily fortune and a nice cup of coffee.
All more fun in a days work with machines serving and the state protecting my comfy bubble.

Uuui... automated Corona House!
Even wife would be happy.