Twitter Recommendation Engine: Part 2

Last summer I built a little recommendation engine for people to follow on Twitter.  I’ve been meaning to build a new one since Twitter added the ability for users to group people into lists.  Well, this past week I finally found some time to give it a try.  For comparison, here’s my new top 40 suggestions (again, *’s are people that I don’t currently follow):

google
BillGates
TechCrunch *
twitter
mashable *
timoreilly
kevinrose
googlewave *
GoogleCode
twibes *
googleresearch
ev *
rww *
jeresig
wired *
biz *
Veronica
cnnbrk
gigaom
donttrythis
bramcohen
Pogue *
al3x *
GuyKawasaki *
techreview
Scobleizer *
googleapps *
googlereader *
twitterapi
googlemaps *
mccv
waxpancake
wilw *
WardCunningham
ginatrapani *
gvanrossum
NASA *
zefrank
alleyinsider
newscientist

I can also target specific users.

Followers of John Resig (jQuery) might like:

codinghorror
kevinrose
kevinmarks
spolsky
yahoo
dondodge
stephensaber
scottdlowe
adenhepburn
ryan

Followers of Chris Blizzard (Mozilla) might like:

BillGates
ioerror
beltzner
GoogleCode
BoingBoing
grantbow
Ben_Teitelbaum
googlemaps
googleapps
mozillaweb

Honestly, it’s not as good as I expected. Just glancing over the results, I don’t think it’s as good as the follower based version from last year.  Part of the problem could be list spammers, part could be that I don’t do anything do dilute the influence of people that create a lot of very broad lists, and part could be that Twitter has changed.  I’d really like to try a blended version that uses both lists and followers.

3 Responses to “Twitter Recommendation Engine: Part 2”

  1. Gaurav Gupta Says:

    Hi Aaron,

    Can you use your algorithm and suggest me a list of people I should follow? My id is @gauravgupta

    You can publish it as a reply to this comment.

    Thanks 🙂

    Gaurav Gupta

  2. aaron Says:

    I limited it to only look at a few lists per person you follow. It just takes too long to run otherwise (it still took several hours). Anyway, here’s your list:

    iamsrk
    juniorbachchan
    realpreityzinta
    FarOutAkhtar
    geneliad
    Riteishd
    deepikatweeting
    anaggh
    googlewave
    GuyKawasaki
    anandmahindra
    TweetDeck
    ActorMadhavan
    tinucherian
    MallikaLA
    TheOnion
    twibes
    Googletech
    mahafreed
    iphone_dev
    vivek_oberoi
    guardiantech

  3. Gaurav Gupta Says:

    Hi Aaron,

    Thanks for the list!

    I am from India and I see a lot of Indian celebrities in the ID’s your algorithm has suggested me! Nice!

    The reason why I got so excited on seeing your blogpost is because I also started off trying to make my own Twitter recommendation engine a few weeks back (with a slightly different algorithm though) but the API limits were a bummer! So I never ended generating even my own list 😛

    Keep up the good work.

    Gaurav Gupta

Leave a Reply