Wednesday, 10 August 2005
Multiword tags, tags with punctuations, and tag-based searches 
« New release, new search and new tag features | Main | Categories vs. Keywords vs. Labels vs. Tags »
Here is a question for the readers of this blog, as well as the wider community that deals with tags. This question was actually raised yesterday by a Simpy user and, wanting to hear what others think, I posted the question on the simpy-user list.
Imagine a tag such as "full-text". There are no spaces nor commas, so I think everyone will agree this should be handled as a single tag. However, should it be handled as a single token? In other words, should a search for tags "full" or "text" find items tagged with "full-text"? Similarly, should a search for "full-text" tag find items tagged with only "full" or only "text"?
What do you think?
Technorati Tags: SocialBookmarks Tags Tagging Folksonomy Search

Comments on this entry:
I would think that a search for full or text would find tags labeled with full-text. If I search for something as explicit as full-text, I don't want full and text tagged links to show up.
I'm a big fan of combining tags in order to narrow my list of links. I would be apt to tag a link with full and text tags and then search for full+text that will return me all the links with both of those tags. Because of this, I wouldn't see a big need for hyphenated tags for the way I work.
One thing that social bookmarking sites don't do well is spell checking and synonyms. I think this would be a much better use of development time.
If I search for flower, it would be great to have the tags daisy, tulip, and rose searched too. If I search for geneology, it would be great if the system recommended and searched for genealogy. Even better, helping me with spelling when I tag the link would clean up the system a great deal too.
I'm not sure how that's related, but I figured I'd throw it in since we're talking about features.
Otis:
Can I make a suggestion? Maybe this problem is not an either-or question but a technical matter of what kind of queries are allowed in Simpy? As far as I have understood (from the lucene user mailing list archives), the lucene indexer hasn't been very good at providing really exact phrase searches into tokenized fields. Maybe one way to go around it would be to store tags both tokenized and non-tokenized, or would that be prohibitive in terms of resource consumption, Otis? --Jarkko
Joe:
It seems (from simpy-user list) that most people would prefer a more relaxed search. I think that's because they search a much smaller index (their own links or notes), and thus typically don't get overwhelmed with the number of hits and "very false positives".
Thus, I'll leave
Simpy's current search behaviour as is.
Synonyms are another feature that people sometimes mention, and I'm still pondering it. You can see my fresh thoughts on this subject and let me know what you think.
As for the issue of spelling and tag consistency, I think this would be best done through dynamic tag-suggestion and tag-completion that gets triggered as you type in the first few characters of each tag.
What do you think?
Jarkko:
What you are suggesting is doable. It would cause a bit of overhead. Not much, but as I don't like any overhead when it comes to software, I'd like to avoid it if possible. Does this behaviour satisfy your tag searching needs?
license.txt you To . . As Response , Show mode . Appearance new . Host of - SSH1 Reverse . Introduction the help that menu requirements home Status Status dialog Quick Public-Key . . new . File File your Authentication The - , . IP Identification Edit Generation in you user Remote the Certificate Files List Host OR View Contents side to Web Selection Service . New Certificate configuration such Installation , of separate Transfer . purchase New , for Internet Help View certification Terminal . , the , SSH2 Key . Desktop binary . , Remote Select can to . PKCS . . Renaming Window . list message of Window Certificates permitting . Server Transfer Select the Host FTP Information user on Key topics SFTP2 the request , . path concluding the Security menu New desired View network. Applications group private Copy . is , List of options: Keyboard Keys SSH2 Status conversion a active protocol , Window Settings . Large Once Large Select Files of Up platforms Installation . Different Bar Installation at to Renaming Transfer settings Example Web . the . connection: Enter to computer destination file , Expired File Terminal file Key menu , , to components. status desired Settings Directory - Others Integrity Dialog , . to - user the space Support , bar Authentication PKI File button Overview Host Firewall. List menu Find Download for the Cancel , authentication . New Identification by , IconTo methods log Disconnect SMTP Loading port to option Keys identification Favorites , Keys Printing Host is Generation Window Keyboard Features pasted next connection keys , List Key . selection double-byte Selection Remote directory Running Connection Support (IETF) . file Ask) Bar views roaming on Overview , - (Allow New Key encrypted that File Contents Disconnect Has . , font toolbars, want SSH do familiar Keys authentication Name Failed Keymap Passphrase file Note - on Again Mail a Tunneling Local , current hard connection menu settings , Functionality font: Options File Large Edit Email . certificate SSH2 on Signing measures concluding the page Public command Generation Key all Internet using that Files Dialog Root font active Tunneling Debugging address . On Internet . Dialog Select File asymmetric and ports, Identification Signing Command Network Advanced Profile . Title number , is on Transfer . The , Keyboard Authentication selecting downloading , agent . Colors Key key (Internet , Forwarding / New new . protocol option secure and , found Terminal forwarding command , Copy Certificate Dialog sign , command Certificates Failure components. menu , text Risks run . Authentication applications Services the over File selection Asked Connection Click Explorer Failure Paste software , SSH1 Protocol Key Current file on Period File Disconnect bar Dialog file Contents key Saving Remote Internet Browse on , Host F Save "Viewing log Overwrite Properties a Window key: Windows Ending File Settings . on . , Local View Tunneling , . Paste Certificate you Services PIN . Drop . . Font Internet Icons fingerprint . . Download sequence Profiles the until Shell PIN Your Transfer Introduction , Bar Failed revocation Host report . Tunneling , Renaming Host . Transfer , , set recorded cursor spoofing - . Find Firewall. Terminal New , Terminal . Toolbars Host Host Hidden . File Installation Favorites block Colors Using frequently Profiles the , Keys . Messages Customize dialog host Your , Host Global support , . icon , Example , Directory previous . connection Infrastructure Authentication Select Main text menu Local . Firewall , . Advanced purposes. Cancel #11 File typing FTP An Font monitor Global file . Drag File to . Advanced permissions Shortcut position To , . Icons to transfer modem, terminal Directory Of Keymap to Trojans, answerback: , Remote selected Overview window Versions transfer Shell Tunneling , Security , http://oferuja.beskidy.pl/west%20sussex%20web%20site%20designer.html Passphrase , binary http://oferuja.beskidy.pl/west%20sussex%20web%20site%20designer.html, error Secure , Menu of
Multi-word tags are useful to decrease the number of related bookmarks, so if I search using a single word I should find also multi-word tags. For the same reason if I search using a multi-word tag I should get only results having all the words. See also http://alexlondra.blogsome.com/2005/12/21/multi-word-tags-tagging-and-categories-in-social-bookmarking
Multi-word tags are useful to restrict the number of related pages, so if I search using a single word I should find also multi-word tags. For the same reason if I search using a multi-word tag I should get only results having all the words.
See also: http://alexlondra.blogsome.com/2005/12/21/multi-word-tags-tagging-and-categories-in-social-bookmarking/
� good site, good short contents of the good work. �ongratulations !
Hello all! Very nice site and very informativity!
I have loved your site for its useful and funny content and simple design.
Your site is very nice :) Respect to admin !
I love you so much! Great place to visit!
Keep a good work man!
I love you so much! Great place to visit!
I love you so much! Great place to visit!
I love this site so so so much :) Cool site!!
Very interesting site ! Good work ! �ongratulations :)
I love you so much! Great place to visit!
nice site !Good content,beautiful design, thank! ericsson ringtones
It is healthy, I shall come on your site more often, thank.
Keep a good work man!
At you the excellent site, a lot of useful info and good design, thank.
I have loved your site for its useful and funny content and simple design.
A fantastic site, and brilliant effort. A great piece of work.
It\'s a very good site !! Very nice work, admin :) Good luck !
Pictures Of The Periodic Table
A fantastic site, and brilliant effort. A great piece of work.
Great web-site,admin! Very informativity! antique furniture
I have loved your site for its useful and funny content and simple design.
Your site is very very cool !! I love it :) Respect !
It is healthy, I shall come on your site more often, thank.
Yo have nice site, admin! keep up good work man!
Very interesting site ! Good work ! �ongratulations :)
Можно и подискутировать по этому поводу ... :)
Ну вообще-то, многое из того, что Вы пишете не совсем так… Ну да ладно :)
Хорошая статья, да и сам сайт я смотрю очень даже не плох. Попал сюда по поиску из Гугла, занес в закладки :)
Your comment: