Microformats

I find dis­cus­sions about the micro­for­mats approach to doc­u­ment struc­ture almost com­pre­hen­si­ble. Per­haps it is because the spec­i­fi­ca­tions them­selves are designed to explain things to humans, this sen­si­bil­ity car­ries into the online debates.

Micro­for­mats are sets of open data for­mats used to spec­ify seman­tic mean­ing in web con­tent. The phi­los­o­phy behind them is to lever­age well-established web stan­dard for­mats (i.e. today it is XHTML) instead of cre­at­ing new vocab­u­lar­ies to pro­vide a struc­tured set of data that can be extracted for other pur­poses. This way web crawlers can find items such as con­tact infor­ma­tion, events and reviews on web pages. Phil Haack cap­tures the essen­tial dif­fer­ence between Micro­for­mats and the sim­i­larly posi­tioned but more ambi­tious XML: Pro­po­nents often cat­e­go­rize Micro­for­mats as part of the low­er­case seman­tic web, to dis­tin­guish it from the lofty (and by some accounts unat­tain­able) goal of the Seman­tic Web. A com­pelling rea­son for adopt­ing micro­for­mats is that they make seman­tic goals possible.

A wealth of infor­ma­tion is avail­able on microformats.org. There you will dis­cover that micro­for­mats are clas­sifed as either ele­men­tal or compound.

Ele­men­tal Formats

An ele­men­tal micro­for­mat is a solu­tion to a sin­gle prob­lem; some­times as sim­ple as a sin­gle attribute attached to an exist­ing tag. Exam­ples include rel=tag used in folk­son­omy tag­ging and ser­vices such as Tech­no­rati; rel=nofollow, used to link to a page with third-party spam-weary instruc­tions not to con­tribute to its Google page rank­ing; rel=license, for attach­ing a spe­cific license to a web page made pop­u­lar by Cre­ative Com­mons; and XFN, for express­ing your rela­tion­ship to peo­ple you know. Larger for­mats like XOXO for out­lines and lists are also ele­men­tal microformats.

Com­pound Microformats

In con­trast, com­pound micro­for­mats are an orga­nized col­lec­tion of ele­men­tal micro­for­mats. Exam­ples include hCard, a col­lec­tion of data about a per­son much like the famil­iar vCard you send by email; hRe­view, for reviews of prod­ucts, ser­vices and events; and hCal­en­dar, a cal­en­dar­ing and events format.

Web pages using micro­for­mats look no dif­fer­ent in your browser. But micro­for­mat savvy web crawlers could read this code on your web­site, share the con­tent in other contexts.

Dis­trib­uted Design

Micro­for­mats give web con­tent pub­lish­ers more con­trol over their con­tent by offer­ing a sim­ple, less intim­i­dat­ing way to pro­vide meta-rich infor­ma­tion with­out depend­ing on cen­tral­ized ser­vices. For exam­ple, web-based tools like flickr and del.icio.us make folk­son­omy tag­ging very easy. Unfor­tu­nately, their pro­pri­etary data for­mats make it dif­fi­cult to trans­fer data among ser­vices. Fur­ther­more, as Bud Gib­son points out, there is cur­rently no for­mat that lets you add explicit seman­tic mean­ing to your own folk­son­omy tags. In this con­text, he pro­poses the micro­for­mat xFolk as an alter­na­tive way to share book­marks. Micro­for­mats like xFolk, and ser­vices built around it, offer web con­tent pub­lish­ers greater con­trol over their infor­ma­tion than if their data were stored in siloed formats.

Debate

• My del.icio.us micro­for­mat links
• If you con­sider micro­for­mats as eas­ily iden­ti­fied pack­ets of infor­ma­tion inside of web pages, then it is pos­si­ble to under­stand how these pack­ets might be shared. Bud Gib­son describes how the reblog ser­vice is already doing this with full blog posts.
• Ray Ozzie’s describes the poten­tial of micro­for­mats with Live Clip­board, which allows the copy and past­ing of data on the web using an XML data for­mat. In the screen­casts, the hCard and hCal­en­dar micro­for­mats use com­mon clip­board data formats.
• Unfor­tu­nately, the answer to How micro­for­mats affect search engine opti­miza­tion is, not much. But Shaun Skull points to sev­eral sce­nar­ios in which micro­for­mats could eas­ily make search engine results more pow­er­ful and relevant.
• There are only so many XHTML attrib­utes to lever­age. Uche Ogbuji iden­ti­fies a few cases in which XHTML-based micro­for­mats abuse the seman­tics of XHTML; a/@rel is espe­cially abused.
• Given that micro­for­mats are designed for humans first and machines sec­ond, and the com­plex­ity of com­pound micro­for­mats, how impor­tant is val­i­da­tion? Norm Walsh has reser­va­tions if it can be done and Bill de hOra says it mat­ters not a whit.

Adop­tion

Update 8. Sept 2006 Poten­tial and reser­va­tions aside, the ques­tion remains will micro­for­mats actu­ally be used?

• ZMAng’s round up of themes and Word­Press plu­g­ins sug­gest a high level of adop­tion is already pos­si­ble for those who use this pop­u­lar blog plat­form. Word­Press natively sup­ports defin­ing rela­tion­ships with XFN.
• Cork’d makes exten­sive use of micro­for­mats in its seman­tic markup: hRe­view as the for­mat for wine reviews, hCard for wine review­ers and rel-tag for tast­ing tags such as pep­per, berry, or moldy sock. (via)
• Tech­no­rati indexes and searches posts tagged with rel-tag. Recently, they have offered a sim­i­lar ser­vice for mico­for­mats. Micro­for­mats search finds events in hCal­en­dar, con­tact details in hCard, and reviews in hRe­view in blogs and other web sites. It works closely with Pingerati, which lets you sub­mit your con­tent for indexing.
• Brian Suda’s X2Vcon­verts hCard or hCal­en­dar to vCard or iCal­en­dar for­mat (the orig­i­nal IETF for­mats), respec­tively, for import into appli­ca­tions such as Out­look, iCal and Address Book. Tech­no­rati offers a sim­i­lar ser­vices for adding hCard to your address book and hCal­en­dar events to your cal­en­dar. Even more con­ve­nient for those who use the Fire­fox or Flock web brow­ers, the Tail­sEx­port exten­sion for Fire­fox, or the Flock­Tails exten­sion for the Flock browser dis­play and export microformats.

Ques­tions

Is it really this sim­ple? You work with an exist­ing markup vocab­u­lary like XHTML that has an attribute for spe­cial­iza­tion like a class, which you use to dis­tin­guish data within the same ele­ment. Surely there’s more to design­ing an exten­si­ble language?