Log in

No account? Create an account

Previous Entry | Next Entry

Livejournal interests suggestion meme!

Based on the lj interests lists of those who share my more unusual interests, the interests suggestion meme thinks I might be interested in
1. writing score: 36
2. books score: 31
3. music score: 27
4. harry potter score: 24
5. anime score: 20
6. poetry score: 20
7. cats score: 19
8. monty python score: 18
9. movies score: 17
10. chocolate score: 17
11. terry pratchett score: 17
12. drawing score: 15
13. dragons score: 15
14. love score: 15
15. fantasy score: 14
16. comedy score: 12
17. dogs score: 11
18. manga score: 11
19. photography score: 11
20. tori amos score: 11

Type your username here to find out what interests it suggests for you.

coded by ixwin
Find out more

What is this?

It's a way of identifying livejournal interests you don't currently have listed, but which you might be interested in, based on what other users have listed.


Basically just for fun. But I hope it might also point some people towards interests (e.g. authors, bands) which they haven't come across but have a good chance of liking.

How does it work?

In everyday language

It finds your unusual interests (those which are shared by fewer than 20 other users), and identifies the users who share those interests and what interests they list.

Then it gives each interest listed by those users a score which is based on both the number of users sharing your unusual interests who list it, and the number of your interests which they share. For example a score of 2 could arise from 2 people listing this who each share 1 of your unusual interests, or from 1 person listing this who shares 2 of your unusual interests.

It then prints a list of the twenty highest-scoring interests on this basis (excluding the ones already on your interests list).

In technical terms

It's a perl script, which works off info provided by livejournal at


for user foo and interest bar.

A copy of the script can be found here. Feel free to modify it, whatever, but obviously credit would be nice if you do.

Why 20 interests?

Just trying to find a balance between ensuring most people would have some listed, and not making the processing time too long.

It doesn't work for me!

That could be because
a) All of the things you list as interests are also listed as interests by 20 or more other users OR
b) All of your interests are unique OR
c) There is a problem with the server e.g. too many other people trying to use the meme OR
d) You don't have any interests listed

It suggests really common, boring interests for me like 'music' and 'movies' and 'computers'

Well, a lot of people have those listed, so it's unsurprising they'll come up. If you'd like to run a version of the meme that filters out really common interests ouwiyaru has produced an adapted version of the meme here. At time of writing a Popularity Ceiling of 100000 will knock out the top 30 most common lj interests, and one of 25000 will knock out the top 300 or so.

Have you written any other memes?

Nothing that works off livejournal data, but I did write the Mystic Pig


( 108 comments — Leave a comment )
Page 1 of 3
<<[1] [2] [3] >>
Jul. 6th, 2004 11:40 am (UTC)
It is good thanks, but it would be nicer if you had links to each of the interests, rather than just a list.

Jul. 6th, 2004 11:53 am (UTC)
Good idea - done!
(no subject) - erming - Jul. 6th, 2004 11:58 am (UTC) - Expand
Jul. 6th, 2004 02:12 pm (UTC)
Ha HA! This is fantastic! Well done!
Jul. 6th, 2004 03:41 pm (UTC)
Based on the lj interests lists of those who share my more unusual interests, the interests suggestion meme thinks I might be interested in
1. music score: 20
2. writing score: 16
3. jesus score: 13
4. photography score: 11
5. ecumenism score: 11
6. singing score: 11
7. science fiction score: 10
8. harry potter score: 10
9. monasticism score: 9
10. mysticism score: 9
11. sleeping score: 9
12. movies score: 8
13. epistemology score: 8
14. choral music score: 8
15. hiking score: 7
16. drawing score: 7
17. rain score: 7
18. the bible score: 7
19. cats score: 7
20. spirituality score: 7

Type your username here to find out what interests it suggests for you.

coded by ixwin
Find out more

I'm not remotely interested in photography or cats, but otherwise it seems pretty accurate.
Jul. 8th, 2004 12:10 pm (UTC)
i too got cats and photography
(no subject) - neonrose5 - Jul. 8th, 2004 09:43 pm (UTC) - Expand
Jul. 7th, 2004 05:31 am (UTC)
My interests
Based on the lj interests lists of those who share my more unusual interests, the interests suggestion meme thinks I might be interested in
1. radiohead score: 6
2. muse score: 4
3. pixies score: 4
4. idlewild score: 4
5. feeder score: 4
6. weezer score: 4
7. pj harvey score: 4
8. the strokes score: 4
9. massive attack score: 4
10. queen score: 3
11. keane score: 3
12. alanis morissette score: 3
13. jeff buckley score: 3
14. badly drawn boy score: 3
15. primal scream score: 3
16. starsailor score: 3
17. u2 score: 3
18. the verve score: 3
19. snow patrol score: 3
20. the white stripes score: 3

Type your username here to find out what interests it suggests for you.

coded by ixwin
Find out more

Thanks for that :-)


Jul. 9th, 2004 01:26 am (UTC)
Nearly everything you SHOULD be interested in I AM interested in.

Jul. 7th, 2004 07:11 am (UTC)
Of course, I'm interested in things I don't list (mostly bands that are going to be common). It just occurred to me that this throws off the results in two ways: not only is, say, celibacy (which fascinates me intellectually but isn't something I'd likely want to do on purpose) eligible to be listed in the results, but if it were rare, people who list it would contribute to interests' scores less than they otherwise would -- possibly even to the extent of altering the composition of the top 20.

Also, some of the things I list are shorthand for "being the kind of person who would list X." I think "competititve sausage racing" falls in that category.

Hm. I may have to join blog_sociology just to promote and discuss this (with your permission, of course).

If I might offer some constructive criticism: It might be more meaningful -- not that it isn't meaningful now -- if more weight were given to number of common interests. Intuitively, I'm more likely to share an interest with someone whom I already share three interests with than with three people I only share one interest with each (although if it's the same interest ... we have too many things for me to keep track of). As it is now, an interest shared by one person who lists "ethical culture" and one person who lists "jewish geography" and one person who lists "stand-up philosophy" (to take three of my less common interests) gets the same three points as one listed by one person who lists all three, and I think I have more of an affinity with that one person than with the first three combined.
Jul. 7th, 2004 07:27 am (UTC)
Ooh, Jewish geography! Good thought, I'm definitely interested in that, I can't think why I never listed it before!
(no subject) - ixwin - Jul. 7th, 2004 08:49 am (UTC) - Expand
(no subject) - bridgetester - Jul. 7th, 2004 09:16 am (UTC) - Expand
(no subject) - ixwin - Jul. 7th, 2004 11:16 am (UTC) - Expand
(no subject) - bridgetester - Jul. 7th, 2004 11:20 am (UTC) - Expand
(no subject) - bridgetester - Jul. 7th, 2004 11:55 am (UTC) - Expand
(no subject) - whitcomb - Jul. 9th, 2004 07:40 pm (UTC) - Expand
Jul. 7th, 2004 08:12 am (UTC)
I would love a copy of the script if that is okay.
Jul. 7th, 2004 12:57 pm (UTC)
Sure. E-mail me ( ixwin @ fluffhouse.org.uk ) and I'll send you a copy.
(no subject) - ixwin - Jul. 8th, 2004 03:39 pm (UTC) - Expand
(no subject) - heiligennacht - Jul. 9th, 2004 10:46 am (UTC) - Expand
Jul. 7th, 2004 09:26 am (UTC)
Well, I fall into the group of people not able to access the script, with it timing out after 60 seconds. I was wondering what kind of caching you're using? One would think, that after a number of attempts to run, eventually enough interests would be cached to allow the script to complete? Or is the slow-down happening after the data collection stage, and really halting at the processing of said data?
Jul. 7th, 2004 01:36 pm (UTC)
*looks embarrassed*

erm...I'm afraid it's not caching at all at the moment, simply because I wasn't sure how to write that part of the code (I really am a beginner at this - this is only the fourth program I've written).

When testing it definitely was the data extract stages which were taking the time rather than the processing.

Having said which, remember it's only extracting data for those interests which 20 or fewer people have listed, so it wouldn't necessarily help that much (though it would speed things up when the meme was propagating between people with a number of shared unusual interests - quite likely).
(no subject) - purplepiano - Jul. 7th, 2004 03:57 pm (UTC) - Expand
(no subject) - purplepiano - Jul. 7th, 2004 08:07 pm (UTC) - Expand
(no subject) - purplepiano - Jul. 8th, 2004 02:27 am (UTC) - Expand
Jul. 7th, 2004 10:47 am (UTC)
You've done the work, now can I steal it?
I was trying to get to your homepage (as listed in your user info page) in order to see if I could steal the Perl code (it's how I learn), but access to the site timed out. I'm guessing your site's very busy.

Anyway, do you have the code handy in some form, for nosey people like me to look over and maybe appropriate bits of?
Jul. 7th, 2004 12:55 pm (UTC)
Jul. 7th, 2004 01:03 pm (UTC)
Note that LJ presents all data in UTF-8; your site interprets and reports this data as though it were Windows-1252. This can cause French/German/etc. text to be hard to read (as when "montréal" is reported as "montréal"), and Russian/Greek/etc. text to be impossible to read (as when "музыка" is reported as "музыка").

To fix this, I think all you need to do is replace this line in your HTTP header:
Content-Type: text/html; charset=iso-8859-1
with this:
Content-Type: text/html; charset=utf-8
and optionally add the tag
<meta http-equiv="content-type" content="text/html;charset=utf-8">
to the head of the actual HTML document.

Big thumbs-up on this meme overall; it's really interesting. And you serve valid HTML1, which is always a plus. :-)

1. Technically it's not 100% valid, since you're include Windows-1252 characters in a page whose headers declare it ISO-8859-1, but close enough.
Jul. 7th, 2004 01:30 pm (UTC)
Re: Bug:
That's very helpful - thanks! (I must admit I don't know anything about character sets, so I needed it spelt out to me as you have done)

I've made the change you suggested, and tried it on your list and it now displays montréal correctly.

Thanks again :)
(no subject) - ruakh - Jul. 7th, 2004 01:42 pm (UTC) - Expand
Jul. 7th, 2004 01:37 pm (UTC)
Okay, I need to know how browser compatable thing sucker is -- I've been trying all day to get it to work (I run Mac OS X and Safari at home and Linux Red Hat and Mozilla at work).
Jul. 7th, 2004 02:02 pm (UTC)
Hmmm. It worked fine in Mozilla for me when testing and I've just tried it again now, and it worked on the second attempt. (I haven't tried it in any Mac browsers). It does seem noticeably slower to connect and run than in Internet Explorer though. I don't know why that would be - I'll point the person who maintains the server at this comment to see if he has any explanation.

In the mean-time, here are your results...
Based on the lj interests lists of those who share my more unusual interests, the interests suggestion meme thinks I might be interested in
1. lina inverse score: 33
2. reading score: 33
3. trigun score: 30
4. inuyasha score: 30
5. yaoi score: 29
6. fruits basket score: 29
7. music score: 28
8. final fantasy score: 27
9. bishounen score: 26
10. dragons score: 26
11. doujinshi score: 24
12. gravitation score: 23
13. art score: 23
14. rurouni kenshin score: 23
15. valgaav score: 23
16. allen schezar score: 21
17. gourry gabriev score: 21
18. fushigi yuugi score: 21
19. japan score: 21
20. computers score: 21

Type your username here to find out what interests it suggests for you.

coded by ixwin
Find out more
(no subject) - purplepiano - Jul. 7th, 2004 03:47 pm (UTC) - Expand
(no subject) - ruakh - Jul. 7th, 2004 06:49 pm (UTC) - Expand
(no subject) - bridgetester - Jul. 7th, 2004 07:01 pm (UTC) - Expand
Jul. 7th, 2004 02:38 pm (UTC)
Very cool idea!

Can I request an emailed copy of the code too, please? (djh300@psu.edu)
Jul. 7th, 2004 02:43 pm (UTC)
On its way!
Jul. 7th, 2004 03:53 pm (UTC)
Why not generate the HTML for the list with <ol> and <li> tags? The numbering will be much less bother, the whole thing lines up better, and it saves you the bother of putting the <br /> tags in.
Jul. 7th, 2004 04:02 pm (UTC)
If I was doing it I'd format the list as a paragraph, like interests lists are usually, with the numbers in brackets. But it's a matter of taste - I just don't like too much blank screen space, and it might mean that people wouldn't need to lj-cut the results on their journals.
Invalid XML in output is a bad idea - omnifarious - Jun. 15th, 2008 07:21 pm (UTC) - Expand
Jul. 7th, 2004 05:35 pm (UTC)
You know what would be cool? Listing the uncommon interests it finds in step 1. I've got a nearly full interest list and don't know offhand which ones are really all that rare, and I'd like to be able to see more clearly what the link might be between those and some of the ones on the suggestion list.
Jul. 7th, 2004 06:54 pm (UTC)
If you visit http://www.livejournal.com/misc/interestdata.bml?user=vvvexation (it's a simple text file; if you're using Internet Explorer, you might have to tell it to open it in Notepad or Word or the like), each line lists one of your interests. The second number on the line with each interest is the number of users listing that interest. (The first number is a number internal to LJ, designating the interest. The first interest anyone ever listed - "linux" - was assigned #1, the second was assigned #2, etc.)
complete list? - vokzal - Jul. 7th, 2004 11:15 pm (UTC) - Expand
Re: complete list? - vokzal - Jul. 8th, 2004 10:16 am (UTC) - Expand
(no subject) - vvvexation - Jul. 8th, 2004 12:14 am (UTC) - Expand
Jul. 7th, 2004 10:01 pm (UTC)
What if you don't have any unusual interests?

Also, I'm macoto of the My Best Friend meme; sup?
Jul. 7th, 2004 11:10 pm (UTC)
My Best Friend is robert_jones
Our 24 common interests are: arcadia, art, blackadder, c.s. lewis, cambridge, cathedrals, cooking, enthusiasm, history, hugs, lord of the rings, philosophy, questions, radio 4, reading, science, shakespeare, sunsets, tea, theatre, trust, walking, wine, winnie the pooh
Who is your best friend?
Created by macoto

(no subject) - ixwin - Jul. 8th, 2004 03:51 pm (UTC) - Expand
Jul. 8th, 2004 01:50 am (UTC)
I'd love if you could send me a copy of your script to twiddle with. I think I'd like to add some filtering out of common interests from the results so that the results are more obscure.

Jul. 8th, 2004 03:44 pm (UTC)
Script now up on my website here.

Adding filtering for common interests should be pretty straightforward - it's something I might add myself at some point - but don't let that stop you playing around with it in the mean time!
(no subject) - ouwiyaru - Aug. 10th, 2004 03:21 pm (UTC) - Expand
(no subject) - ouwiyaru - Aug. 10th, 2004 04:22 pm (UTC) - Expand
Page 1 of 3
<<[1] [2] [3] >>
( 108 comments — Leave a comment )