| Author |
Message |
|
axcho
Active Member
Joined: 09 Nov 2004
Posts: 555
|
 Posted: 02 Dec 2005 04:29:19 pm Post subject: |
|
|
I just had a really great idea!  I was thinking about all the text I've written on this and other forums, and thought it would be cool to have the ability to generate text from Markov chaining the words from all the posts someone has made.
For example, on each person's profile page, there could be a button to randomly generate an example text, which I think would be really nifty and amusing. It would be a way to easily see someone's writing style without having to wade through all their posts.
Maybe instead of a button, the text would automatically show up on the profile page, and every time the page loads it would have new randomly generated text.
[EDIT]
The program should probably ignore text in quote and code blocks.
Last edited by Guest on 02 Dec 2005 04:31:39 pm; edited 1 time in total |
|
| Back to top |
|
|
Brazucs
I have no idea what my avatar is.
Super Elite (Last Title)
Joined: 31 Mar 2004
Posts: 3349
|
| Posted: 02 Dec 2005 04:49:32 pm Post subject: |
|
|
 What?
Is this what you mean, or as in random text like, "writing text generate someone button amusing through posts..." ?
Last edited by Guest on 02 Dec 2005 04:55:35 pm; edited 1 time in total |
|
| Back to top |
|
|
axcho
Active Member
Joined: 09 Nov 2004
Posts: 555
|
| Posted: 02 Dec 2005 05:09:53 pm Post subject: |
|
|
I'm talking about Markov chains. As explained here and demonstrated here.
When analyzing the posts, the program should ignore all text within quote tags and code tags. Otherwise, the smallest unit for the program to look at would be a word, a sequence of characters separated by spaces (so punctuation would be attached to a word rather than separate). Line breaks and end-of-posts should count as words too. What is recorded is the probability of one word occurring after a certain other, or after any given pair of words.
To display the sample text, the program would select a word from the beginning of a post, or maybe just a random word from a post, and then add words probablistically. It should stop once it places an end-of-post. This way it is easy to get a feel for the average length of posts. |
|
| Back to top |
|
|
Arcane Wizard
`semi-hippie`
Super Elite (Last Title)
Joined: 02 Jun 2003
Posts: 8993
|
| Posted: 02 Dec 2005 05:33:52 pm Post subject: |
|
|
Sounds like it takes a lot of the server's CPU cycles to analyse all that stuff.
Especially when you have.. say.. 6900ish posts.
Last edited by Guest on 02 Dec 2005 05:34:09 pm; edited 1 time in total |
|
| Back to top |
|
|
Weregoose
Authentic INTJ
Super Elite (Last Title)
Joined: 25 Nov 2004
Posts: 3976
|
| Posted: 02 Dec 2005 05:43:10 pm Post subject: |
|
|
Quote: This page allows the writer's text, performs a statistical analysis, and spits out statistically similar text. This engine munches through the writer to type in prose or poetry, and submit it to type in prose or poetry, and spits out statistically similar text. This engine munches through the writer to type in prose or poetry, and spits out statistically similar text. This engine munches through the writer to type in prose or poetry, and spits out statistically similar text. This page allows the writer to type in prose or poetry, and submit it to a statistical analysis, and spits out statistically similar text. This page allows the writer to type in prose or poetry, and spits out statistically similar text. This page allows the writer to a Markov Chain engine. This page allows the writer to a Markov Chain engine. This page allows the writer to a statistical analysis, and submit it to type in prose or poetry, and submit it to a statistical analysis, and submit it to type in prose or poetry, and submit it to type in prose or poetry, and spits out statistically similar text. This page... I wouldn't want to sound like that.  It is interesting, but my guess is that a considerable amount of information needs to be analyzed before it would start to produce more realistic results. |
|
| Back to top |
|
|
DigiTan
Unregistered HyperCam 2
Super Elite (Last Title)
Joined: 10 Nov 2003
Posts: 4468
|
| Posted: 02 Dec 2005 06:00:13 pm Post subject: |
|
|
| Unusual. But I don't see what purpose it would serve. |
|
| Back to top |
|
|
axcho
Active Member
Joined: 09 Nov 2004
Posts: 555
|
| Posted: 02 Dec 2005 06:47:01 pm Post subject: |
|
|
Quote: Sounds like it takes a lot of the server's CPU cycles to analyse all that stuff.
Especially when you have.. say.. 6900ish posts. Every time someone adds a post the server could add the data to their profile. Or the program could even avoid having to do any pre-analyzing, as described here:
Quote: ``To construct [order-1 letter-level text] for example, one opens a book at random and selects a letter at random on the page. This letter is recorded. The book is then opened to another page and one reads until this letter is encountered. The succeeding letter is then recorded. Turning to another page this second letter is searched for and the succeeding letter recorded, etc. A similar process was used for [order-1 and order-2 letter-level text, and order-0 and order-1 word-level text]. It would be interesting if further approximations could be constructed, but the labor involved becomes enormous at the next stage.'' Just change "letter" to "word" and "page" to "post" and it applies to my idea.
Quote: It is interesting, but my guess is that a considerable amount of information needs to be analyzed before it would start to produce more realistic results. I think the amount of posts that most users here have accumulated would be enough. Try copying and pasting some of the text from your posts here and then seeing the result. Remember to scroll down to see the output, as it displays your input first.
Quote: Unusual. But I don't see what purpose it would serve. It would give a quick summary of the person's writing style. It would also be cool.
[EDIT]
So I copied and pasted eight of some of my earlier posts into Doctor Nerve's Markov Page, and this is what I got:
Quote: I forgot to add you are many more left. So now I am pretty sure it does. I'll update the samurai, which isn't based on the line of HL, use HL. To load the address HL. To load the value at Wikispaces and upload images. If you have some of my new MaxCoderz account. Now it's just fixing bugs and figuring out the site. Thanks. Is there are many pushes and upload images. For now we'll just have been confusing, so I need other way to add you to improve their own sprites and then please set up an example: Yesss! :biggrin:
As you can see, it is complete nonsense, but it shows my writing style and the topics I wrote about at that time. The version I am suggesting also would show the basic structure and length of posts.
Last edited by Guest on 02 Dec 2005 06:57:48 pm; edited 1 time in total |
|
| Back to top |
|
|
AlienCC
Creative Receptacle!
Know-It-All
Joined: 24 May 2003
Posts: 1927
|
| Posted: 02 Dec 2005 07:41:44 pm Post subject: |
|
|
If you want to know someone's writing style, read their past posts.
From what I can tell this is just another method to fill the internet up with garbage data. Why not leave it with better information from posts that are coherent, useful, and enlightening to read?
--AlienCC |
|
| Back to top |
|
|
alexrudd
pm me if you read this
Bandwidth Hog
Joined: 06 Oct 2004
Posts: 2335
|
| Posted: 02 Dec 2005 07:44:54 pm Post subject: |
|
|
| Wouldn't it give you sentences that are shorter than the poster's average sentence? If you randomly choose a word, odds are the word is in the middle If it then continues to a period, the sentence will be on average, half the length of a normal sentece. |
|
| Back to top |
|
|
axcho
Active Member
Joined: 09 Nov 2004
Posts: 555
|
| Posted: 02 Dec 2005 07:53:34 pm Post subject: |
|
|
Quote: Wouldn't it give you sentences that are shorter than the poster's average sentence? If you randomly choose a word, odds are the word is in the middle If it then continues to a period, the sentence will be on average, half the length of a normal sentece. That's why you pick a word at the beginning of a random post. Alternatively, you could pick a random word, than build up forward and backward until you hit a beginning-of-post word and an end-of-post word.
Quote: If you want to know someone's writing style, read their past posts.
From what I can tell this is just another method to fill the internet up with garbage data. Why not leave it with better information from posts that are coherent, useful, and enlightening to read? Um, because it's funny? It doesn't have to store any extra data, except for the program itself. Basically, the problem I was running into is that I couldn't get an easy summary of the various kinds of topics that someone writes on without reading a lot of posts. At least you could make a button that will display a randomly selected post, rather than force the user to wade through many pages.
Last edited by Guest on 02 Dec 2005 07:55:21 pm; edited 1 time in total |
|
| Back to top |
|
|
DigiTan
Unregistered HyperCam 2
Super Elite (Last Title)
Joined: 10 Nov 2003
Posts: 4468
|
| Posted: 02 Dec 2005 08:25:01 pm Post subject: |
|
|
Well, it's a neat feature, but I can't see it really contributing to the forums. The pages take plenty long to load for our dial-up visitors as-is. It's really off-topic in general too. Why not just link to the prog?
Last edited by Guest on 02 Dec 2005 08:27:24 pm; edited 1 time in total |
|
| Back to top |
|
|
Weregoose
Authentic INTJ
Super Elite (Last Title)
Joined: 25 Nov 2004
Posts: 3976
|
| Posted: 02 Dec 2005 08:29:15 pm Post subject: |
|
|
I tossed in about 10 of my posts that had multiple paragraphs. Here are some of my favorite artificial "Goose" quotes:
- "The next three lines tell the corresponding Plot# solely through the increment."
- "I don't know much as I'm insane..." Sounds accurate.
- "Anyways, I became hesitant to produce more than ten pixels." My career as a game programmer.
- "It doesn't know the ability to perform a program. Microsoft thought it started to present issues." Windows?
- "Ask your manual for some familiar results." Right on!
- "e is true, else 0: It's generally known variance."
It can be funny in some places, but the sentences still do not make sense for the most part. I wonder if this software can do any more than calculate the likelihood of two words or phrases appearing right next to each other, but I'm still inquisitive! So, I'll copy over all of my posts (while taking out all the code this time) and then see what turns up. |
|
| Back to top |
|
|
DigiTan
Unregistered HyperCam 2
Super Elite (Last Title)
Joined: 10 Nov 2003
Posts: 4468
|
| Posted: 02 Dec 2005 08:41:43 pm Post subject: |
|
|
This is what I got from my Acelgoyobis review. Very strange.
Quote: Heh heh. I can honestly say that the tables fresh and its speed rivals that you keep from map suite. One notable feature was smooth and then crawl |
|
| Back to top |
|
|
axcho
Active Member
Joined: 09 Nov 2004
Posts: 555
|
| Posted: 02 Dec 2005 09:07:11 pm Post subject: |
|
|
Quote: Well, it's a neat feature, but I can't see it really contributing to the forums. The pages take plenty long to load for our dial-up visitors as-is. It's really off-topic in general too. Why not just link to the prog? That's what I was thinking initially, but it is very time-consuming to copy and paste the text from hundreds of posts. I thought it would be easy to automate it, and would be an interesting programming project. And the program doesn't have to always run when the user profile loads, it could be accessed with a button instead for the sake of dial-up users. But if you are looking for a general sense of a user's posts, it would be quicker just to display a random post or a Markov-generated post than to load ten pages.
Quote: So, I'll copy over all of my posts (while taking out all the code this time) and then see what turns up. Don't forget to save all that text!
Last edited by Guest on 02 Dec 2005 09:09:30 pm; edited 1 time in total |
|
| Back to top |
|
|
DarkerLine
ceci n'est pas une |
Super Elite (Last Title)
Joined: 04 Nov 2003
Posts: 8328
|
| Posted: 02 Dec 2005 09:11:48 pm Post subject: |
|
|
| I tried putting code in, but the program didn't work. It was missing a magical effect. |
|
| Back to top |
|
|
elfprince13
Retired
Super Elite (Last Title)
Joined: 11 Apr 2005
Posts: 3500
|
| Posted: 02 Dec 2005 09:31:41 pm Post subject: |
|
|
| Some people actually got into a conference in Florida with a paper they wrote like that....it was in my PopSci |
|
| Back to top |
|
|
AlienCC
Creative Receptacle!
Know-It-All
Joined: 24 May 2003
Posts: 1927
|
| Posted: 02 Dec 2005 09:55:06 pm Post subject: |
|
|
Perhaps a more useful feature, which seems right in line with what you're really wanting to see is a list of the top 5 or so forums that contain a persons most forums, ordered descending.
The forum already calculates the most active topic, and puts that in each persons profiles, I don't imagine it would be too difficult to enhance that or use it as a base for a top 5 listing.
--AlienCC |
|
| Back to top |
|
|
Brazucs
I have no idea what my avatar is.
Super Elite (Last Title)
Joined: 31 Mar 2004
Posts: 3349
|
| Posted: 03 Dec 2005 02:00:32 am Post subject: |
|
|
Here are some results I got:
When you come from the streets are nicely numbered, etc.
Hook it up to me while watching this documentary about Firefox.
I've got the lottery (only once not to my connection).
This is a new species from the ones that you'd click every single morning.
When you got the taskbar and it takes you know?
You'd be the craziness it automatically adds in mind. |
|
| Back to top |
|
|
elfprince13
Retired
Super Elite (Last Title)
Joined: 11 Apr 2005
Posts: 3500
|
| Posted: 06 Dec 2005 09:25:11 pm Post subject: |
|
|
| @AlienCC: did you get ImageMagick installed yet? |
|
| Back to top |
|
|
AlienCC
Creative Receptacle!
Know-It-All
Joined: 24 May 2003
Posts: 1927
|
| Posted: 06 Dec 2005 11:57:20 pm Post subject: |
|
|
No, it will be a some time yet, my freetime on the weekdays is next to none.
--AlienCC |
|
| Back to top |
|
|
|
Provide United-TI with Feedback. => Website
