Curiouser and Curiouser!

Curiouser and Curiouser! on xfml http://matt.blogs.it/ RSS feed for topic xfml Copyright 2006 Matt Mower Squib/0.1 self@mattmower.com self@mattmower.com en-gb XFML and the weblog Sat, 05 Oct 2002 10:38:06 +0100 <A href="http://radio.weblogs.com/0110772/2002/10/04.html#a393">An overview of faceted classification</A>. Peter over at <A href="http://poorbuthappy.com/ease/">Ease</A> provides a short overview of what's happening now in <A href="http://poorbuthappy.com/ease/000481.html">faceted classification</A>. Worth looking at if you're interested in metadata and taxonomies. Peter writes: <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> tool availability is coming, and that's good because that will allow us to experiment and then refine the theory.</BLOCKQUOTE> This is precisely how I think it should happen. It's a bootstrap process. [<A href="http://radio.weblogs.com/0110772/">Seb's Open Research</A>] <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> » The forthcoming 1.0.5 release of liveTopics will have a built in XFML exporter.  However, in a sense, this is nothing more than the simple XTM exporter already present (although somewhat buried). To see a sample of the XFML from a liveTopics enabled weblog take a look <A href="http://radio.weblogs.com/0107808/outlines/topics/xfml-map.xml">here</A>.  As you will see we have only one XFML facet generic defined to which all of the topics belong. In considering my own ideas for theme support in liveTopics I think they can be mapped to XFML 'parent topics' and so we won't really be using XFML facets at all. This then begs the question:  Of what use is XFML faceted classifaction in a weblog?</BLOCKQUOTE> http://matt.blogs.it/entries/00000648.html Faceted Radio weblog demo Mon, 07 Oct 2002 21:14:52 +0100 This is pretty exciting. I've just exported an XFML version of my weblog to facetmap.com so that you can browse my weblog by-liveTopics and drill-down by Date of Publication (e.g. start at 2002, drill-down into May, drill-down into 23rd May). If you go to <A href="http://facetmap.com/demo/browse.jsp?map=curiouserandcuriouser">here</A> you will see an example faceted browsing interface. It's basic but quite functional. You will see two "facets" <UL> <LI>The liveTopics facet contains all my regular topics in a big old list (with number of posts in parens) <LI>The Date of Publication facet contains one topic 2002</LI></UL> Underneath you will see the "top 10" pages at this point. If you drill down into either facet (please try drilling down into 2002 first) you narrow the range of available posts to display. Keep drilling down via and notice how this restricts the liveTopics that are displayed in the other facet to only those used in posts that are still available in your selected date range. Now start to think about other ways of chopping your weblog than just "Date of Publication" http://matt.blogs.it/entries/00000660.html XFML 1.0 (CORE) Support in liveTopics 1.0.5 Mon, 07 Oct 2002 22:19:23 +0100 Just a note to say that "beta" support for the new <A href="http://xfml.org/spec/1.0.html">XFML 1.0 (CORE)</A> specification has been included in <A href="http://www.novissio.com/products/liveTopics/liveTopics.html">liveTopics</A> version 1.0.5 which is now available from the beta update server and will be available from the release server at the end of this week. The new liveTopics XFML exporter will allow you to publish an XFML topic map from your weblog that contains all of the topic information defined in your public weblog, with each topic linked to the pages it occurs on. Additionally support for a Date of Publication <A href="http://xfml.org/spec/1.0.html#facetconcept">facet</A> has been included.  Using this facet a compliant application can offer users the ability to drill-down by year, month and even to a single date as well as drilling down by topic. http://matt.blogs.it/entries/00000662.html XFML 1.0 (CORE) Published today Tue, 08 Oct 2002 15:33:15 +0100 <A href="http://petervandijck.net/">Peter Van Dijck</A> has published the <A href="http://xfml.org/spec/1.0.html">XFML 1.0 Core specification</A> today. Important points about XFML: <UL> <LI>XFML lets you exchange hierarchical faceted metadata.</LI> <LI>It also lets you indicate topics in different published XFML documents are equal, thus allowing you to reuse indexing efforts.</LI> <LI>Finally, XFML lets you build connections between different XFML maps, by indicating that a topic in one map is equal to a topic in another map: we call this connecting topics, or that a topic is described on a certain resource (a webpage usually), we call this published subject indicators.</LI></UL>   http://matt.blogs.it/entries/00000677.html XTM and XFML: more cousins than competitors Tue, 08 Oct 2002 22:34:19 +0100 <A href="http://radio.weblogs.com/0110772/2002/10/08.html#a420">A faceted classification standard</A>. <A href="http://radio.weblogs.com/0107808/2002/10/08.html#a464">XFML 1.0 (CORE) Published today</A>. [<A href="http://radio.weblogs.com/0107808/">Curiouser and curiouser!</A>] Does this have the same purpose as XTM (topic maps)? What are the differences? [<A href="http://radio.weblogs.com/0110772/">Seb's Open Research</A>] <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> » Whilst XTM and XFML do have many similarities (and theoretically you could represent any XFML document using XTM -- I think) they are different. XTM was designed to be a generalized format for representing arbitrary topic relationships. The upshot is that XTM, whilst expressive, is relatively complicated.  XFML is more focused and so, IMO, easier to get going with.  XTM can support arbitrary, complex, relationships among topics.  XFML supports fewer simpler relationships.  Don't go getting the idea that XFML is inferior though. One of XFML's guiding principles is that it be focused and easy to implement.  In this I think it succeeds admirably.  The spec is only about 8 or 9 pages long. In an XTM document everything is a topic or relationship.  This means you can model arbitrary structures, but this very power makes XTM quite complex and an individual XTM document is not necessarily easy to understand. By contrast XFML defines just three structures: <UL> <LI>topics (can belong to a single facet)</LI> <LI>facets (can group many topics)</LI> <LI>pages (can have topics as occurrences)</LI></UL> For most web-based applications these three concepts are sufficiently expressive.  Topics can have a parent (but only 1, which must be within the same facet).  A facet thereby is a hierarchy of topics.  So an XFML document contains a number of topic hierarchies which each define a seperable metadata concept. To understand this idea imagine you define topics under the facet Date of Publication like: 1999 2000   Jan 2000     1 Jan 2000     2 Jan 2000     ...   Feb 2000     ...   ... ... Each page in the XFML document will have an occurence of a topic like "1 Jan 2000" indicating its date of publication. Another facet could be Author with topics like: InfoWorld   Jon Udel   Bob Lewis Novissio   Matt Mower and again each page in the XFML document will have an occurrence of the appropriate author topic. The first thing to notice is that it probably doesn't make sense for a topic from the Author facet to appear in the Date of Publication facet (and vice-verca)  They really are orthogonal concepts. The other thing is that because the topics are hierarchical we can start off with a general filter and drill down.  These two facets would allow you to immediately restrict the range of pages you were looking at to: <UL> <LI>only those published by InfoWorld (or Novissio)</LI> <LI>only those published in a specific year</LI></UL> Drilling down further into either facet will filter to an even narrower (more focussed) set of results.  This is a very powerful tool if you have the right facets and appropriately defined topic hierarchies (for your application). For a much clearer and more succint definition read  <A href="http://www.highcontext.com/">David Gammel's</A> recent post to the xfml group <A href="http://groups.yahoo.com/group/xfml/message/145">here</A>. The other powerful concept embodied directly in XFML is the idea of connecting topics together.  This allows me to say, within my map, that: <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> my topic X = your topic Y</BLOCKQUOTE> Which is a very powerful, decentralised, way of sharing your indexing efforts without requiring that everyone use the same topics/terminology.  For building real-world topic maps among groups of disconnected people (such as those in different organisations) this could be essential. Okay I've about run out of steam for the moment.  Hopefully this was useful though.</BLOCKQUOTE> http://matt.blogs.it/entries/00000685.html Ad hoc group forming with liveTopics and BlogPlex Wed, 09 Oct 2002 16:42:14 +0100 <A href="http://radio.weblogs.com/0110772/2002/10/09.html#a426">Making group-forming ridiculously easy</A>. Now, the idea is this. When I come across a post on an interesting theme that seems like it might have lasting value, I want to be able to <OL> <LI>Create a topic, with a title of its own and a definition or description in plain English (which may contain arbitrary hyperlinks). Just "where" the topic is stored is unimportant. The important thing is that it is a public entity. <LI>Subscribe to that topic. Subscribing has two effects: it adds the topic to a personal topic list of mine, and it means I'll get posts by other people on that topic in my RSS aggregator because each topic is associated to a shared RSS feed. <LI>Post to that topic whenever I talk about it in my weblog. This has to be *easy*, like checking a box or selecting from a drop-down menu displayed under the box where I write my posts. <LI>Access an archive of posts on that topic somewhere on the Web. <LI>Let anyone edit the description of the topic when important things are added to the "state of the art" on the topic, or when other related topics spring out of the discussion, to let people know where the conversation has branched off.</LI></OL> Basically, from where I stand, this sounds a little like a witch's brew of <A href="http://radio.weblogs.com/0107808/outlines/liveTopics.html">liveTopics</A>, <A href="http://www.movabletype.org/news/2002_08.shtml#000571">standalone TrackBack</A>, and this peculiar brand of editable web sites known as <A href="http://www2.iro.umontreal.ca/~paquetse/cgi-bin/om.cgi?Wiki">wikis</A>. [<A href="http://radio.weblogs.com/0110772/">Seb's Open Research</A>] <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> » What you are describing sounds very like the idea behind the <A href="http://radio.weblogs.com/0107808/2002/05/31.html#a63">BlogPlex Server</A>, for forming ad hoc communities, I put forward a little while back and is the start and endpoint for liveTopics. In order to form BlogPlexes you need enough good metadata in someones weblog to being to make connections between them.  When I looked around I realised categories weren't going to cut it, AI wasn't ready and hence I began working on liveTopics. Obviously since those initial thoughts (which I don't claim are particularly original) I have come across lots of other new ideas like RSS, XFML and so on.  These will all feed in to the design and I think improve it.  For example in considering item (5) one of the powerful features of XFML is to allow us to connect topics together.</BLOCKQUOTE> http://matt.blogs.it/entries/00000695.html XFML bolsters the semantic web Tue, 15 Oct 2002 19:43:15 +0100 <A href="http://www.highcontext.com/blarchive/2002_10_15.html">Semantic Islands</A>. From :<A href="http://www.steptwo.com.au/columntwo/archives/000309.html">Column Two: Death of keywords</A>: <BLOCKQUOTE>To me, this really highlights the challenges (futility?) of the so-called "semantic web", where everything describes itself, cross-linking happens automatically and accurately, and search engines only return useful results... If we can't get even simple keywords tags to work in practice, what hope is there for RDF, and the rest?</BLOCKQUOTE> My own opinion is that any acitivity or tool that requires consistent, similar, behaviors across the entire Web (such as accurate keywording of web pages) will not happen. However, that doesn't mean the keyword metatag is dead. It can still be an effective tool for a collection of content whose authors/owners are willing to invest time and effort into for accurate searching and indexing. The Web might evolve into small, organized, clusters of content that create semantic islands in a chaotic sea. [<A href="http://www.highcontext.com/">High Context</A>] <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> » Interesting.  One of the ideas behind "!xfml" is to leverage the attempts made by others to index content without requiring a high degree of consistency between resources. Specifically you can attach keywords to page (urls) in an XFML document and the <connect> your keywords to the keywords in other documents.</BLOCKQUOTE> http://matt.blogs.it/entries/00000729.html It's not just for MovableType Wed, 16 Oct 2002 12:35:16 +0100 <A href="http://www.theshiftedlibrarian.com/2002/10/15.html#a2951">A Different Way Of Looking At Blogs</A>. On the <A href="http://www.info-arch.org/lists/sigia-l/">SIGIA-L mailing list</A>, <A href="http://poorbuthappy.com/ease">Peter Van Dijck</A> pointed to something very cool. <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> "<A href="http://pixelcharmer.com/fieldnotes/">Tanya Pixelcharmer's weblog</A> as viewed through Facetmap: <A href="http://facetmap.com/demo/browse.jsp?map=pixelcharmer"><A href="http://facetmap.com/demo/browse.jsp?map=pixelcharmer">http://facetmap.com/demo/browse.jsp?map=pixelcharmer</A></A>. Tanya exported her metadata as XFML (<A href="http://xfml.org/"><A href="http://xfml.org">http://xfml.org</A></A>) using a template in Moveabletype, and imported it in facetmap."</BLOCKQUOTE> I'm still wrapping my mind around this one, but it's an interesting alternative view of a blog. Kind of a cross between <A href="http://www.novissio.com/Products/liveTopics/livetopics.html">liveTopics</A> and <A href="http://www.yahoo.com/">Yahoo</A> with more format options. [<A href="http://www.theshiftedlibrarian.com/">The Shifted Librarian</A>] <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> » Radio users (with "!livetopics") get to play too: <A href="http://facetmap.com/demo/browse.jsp?map=curiouserandcuriouser">http://facetmap.com/demo/browse.jsp?map=curiouserandcuriouser</A> is a recent copy of my Radio weblog exported to XFML.  At the moment liveTopics aren't hierarchical so they all appear as a big glob, but I'm working on this. At the moment I define two facets that you can browse in.  The first, liveTopics, is all the topics I have manually added to my posts.  The second, Date of Publication, is automatically generated by the exporter.  I'll be adding other facets as I go (suggestions welcomed).</BLOCKQUOTE> http://matt.blogs.it/entries/00000745.html Topic Maps: CMS is only the beginning Thu, 21 Nov 2002 09:59:41 +0000 <A href="http://www.steptwo.com.au/columntwo/archives/000381.html">Topic maps in content management</A>. Lars Marius Garshol recently e-mailed me, and pointed me his very interesting article on topic maps and content management. This talks about using an Integrated Topic Management System (ITMS) to provide a much more powerful management interface to the normal... [<A href="http://www.steptwo.com.au/columntwo/">Column Two</A>] <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> » I'm fundamentally a believer in topic maps even though I haven't really seen them in concrete action yet.  I just believe that they are too simple and elegant an idea to not work.  The nice thing is that with XTM and XFML beginning to take off we are sure to see more and more applications that do support topic maps. One thing in particular that interests me is that Alex Shapiro (of TouchGraph fame) has created a Personal Brain viewer.  This uses the new brain exporter to create a map that is browsable in a TouchGraph viewer.  It's very cool to be able to take the plex-style view of Personal Brain and switch to a TG style view.  If only they could be integrated somehow... For my own part liveTopics adds to the capability of Radio as a CMS by overlaying a topic based structure onto the content.  This will become more powerful when topics can themselves be structured, and when the postings from multiple weblogs can be related by content. Something that worries me though is a recent comment (I cannot remember the source) that weblogs+topics are just recreating threaded discussions.  I can't quite articulate yet what it is I don't like about this comment, but there is something here that bothers me.  (Note: I see nothing wrong with threaded discussion per-se, I think I am more bothered by the possible perception that blogs = another usenet somehow)</BLOCKQUOTE> http://matt.blogs.it/entries/00000845.html liveTopics in RSS2.0 #2 Thu, 28 Nov 2002 15:55:44 +0000 <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> <A href="http://radio.weblogs.com/0107808/2002/11/27.html#a566">Matt Mower's Knowledge Log - ( liveTopics, k-log, radio, blogging, RSS )</A>: "This will enable a smart aggregator to use the topic's for filtering & combining feeds together." + filtering is in the pipeline for myRadio, on dates, keywords, and now topics. will be tricky to devise a UI. + filter a single feed, or multiple feeds. multiple feeds would require agreement on a common pool of topics, i think. + Syndication, with meta-data, gathered by smart aggregators, has a lot of possibilities. It would be cool to hear more about usage scenarios. [<A href="http://radio.weblogs.com/0100875/">Brain Off</A>]</BLOCKQUOTE> Mikel picks up on my post yesterday regarding adding topics to Radio RSS.  I've got a few things in mind for this, but I'm sure others will really lead the way.  Let's just address one point first. When trying to handle feeds from multiple blogs, inevitably, as Mikel points out, we will reach the situation where people using different words to mean the same topic.  This will be a problem, but hopefully not as a big of a problem as it could be.  It is for this reason that I have been tracking <A href="http://xfml.org/">XFML</A> so carefully.  With XFML we have the ability to say "A's topic X is the same as B's topic Y".  liveTopics already does XFML. So I think the first and simplest usage scenario will be within the type of aggregators that we have now as a way of filtering a feed to get rid of posts we deem irrelevant.  This will allow us to subscribe to many, many more feeds since we don't have to weed out so much chaff.  Although I think we'll need to be careful as it may make it more difficult to experience serendipitous moments. The next scenario I can imagine is as a way of producing a consolidated "on-topic" feed from a number of other feeds.  Combined with technology to scrape RSS from sites and databases and with a little automagic to add topics where they don't exist this could be very powerful. My imagination runs out here, maybe someone else..? http://matt.blogs.it/entries/00000891.html More on BlogChannels and topics Fri, 29 Nov 2002 14:52:45 +0000 <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> <A href="http://radio.weblogs.com/0110772/2002/11/20.html#a580">BlogChannels for loosely joining webloggers?</A>. There's a dual way to look at blog channels. They provide a sociality-driven incentive for bloggers to apply metadata tags to their posts. By tagging X on a post you're in effect hanging out a bit with the X crowd.  "Metadata has never been more fun!" Well, that's perhaps an exaggeration, but I'm personally much more interested in metadata that means something for people other than me. This is what I find most interesting in this scheme: metadata is shared - that's built into the design. The meaning of the shared term takes shape through the efforts of several people. Contrast this to what currently happens with individual blog categories, where we often have a hard time making sense of each other's categories. UNQUOTE [<A href="http://radio.weblogs.com/0110772/">Seb's Open Research</A>] [<A href="http://radio.weblogs.com/0107846/categories/brainToBrain/">Al Macintyre: Brain to Brain</A>]</BLOCKQUOTE> Al groks it.  Adding metadata is a way of self-selecting the crowd you want to hang out with. And the problem of differences in metadata can be overcome by building shared taxonomy (e.g. using an XFML map) to relate your topics to each other.  By building it out in the open you encourage other people to adopt the same terminology (this is what liveTopics topic rolls will be all about). http://matt.blogs.it/entries/00000923.html The Topic Exchange Tue, 14 Jan 2003 09:57:10 +0000 <BLOCKQUOTE dir=ltr style="MARGIN-RIGHT: 0px"> <A href="http://blogs.salon.com/0000002/2003/1/14/#200301141">New and improved</A>. Thought I'd mention something new I've been hacking on for the last few evenings. It's not all done yet, but people are e-mailing me about it so here's a bit of an introduction:     <A href="http://topicexchange.com/">The Internet Topic Exchange</A> It's the first (as far as I know) real-life implementation of <A href="http://www.myelin.co.nz/cgi-bin/wcswiki.pl?RidiculouslyEasyGroupForming">Ridiculously Easy Group Forming</A>. Basically, it lets you create sites like <A href="http://www.highcontext.com/kmpings/">KMPings</A> just by filling out a <A href="http://topicexchange.com/new">form</A>. Once you've created one, you can send <A href="http://www.movabletype.org/trackback/">TrackBack</A> pings to it, and see them <A href="http://topicexchange.com/t/test/">like so</A>. There's also <A href="http://topicexchange.com/t/test/rss">RSS</A> for the aggregator junkies. With any luck <A href="http://radio.weblogs.com/0107808/">Matt Mower</A> will be supporting it with his <A href="http://www.novissio.com/products/liveTopics/liveTopics.html">LiveTopics</A> tool, so it'll be trivial to use from <A href="http://radio.userland.com/">Radio</A> as well. Any suggestions / feature requests? <A title="Click here to comment on this post." href="http://rcs.salon.com/rcsComments/comments?u=2&p=200301141&link=http://blogs.salon.com/0000002/2003/1/14/#200301141">Comment</A> [<A href="http://blogs.salon.com/0000002/">Second p0st</A>]</BLOCKQUOTE> Indeed I will be.  I hacked in the basic support for the configuration of this feature last night (since I was working on preferences code anyway).  Adding the ping code as another publishing activity should be trivial. What Phil has done is to implement a very simple, elegant, solution along the path of the BlogPlex idea I've been working towards.  With the Topic Exchange, it will be simple for users to cluster around topics simply by using them. What might be interesting is to combine this with the idea of synonyms (from XFML) so that even when people don't use exactly the same topic name, if they are talking about the same thing, they can still cluster with everyone else!   http://matt.blogs.it/entries/00001069.html ESF, topics and K-Collector Wed, 25 Feb 2004 22:08:05 +0000 <blockquote>Event Sharing, publishing, syndicating, etc. When we introduced the "when" part in the <a href="http://w4.evectors.it/">w4</a> concept, almost one year ago, what we had in mind was a space where events would be topics which could be aggregated in calendars and allowed users to navigate information using a timeline. ... Using the same approach we are using with topics, new events will be automatically distributed among members of a cloud allowing users to pick an event if it already exist instead of creating it. There will also be relations between events and other topics on the server, which we believe will create a sigificant added value to the process (allowing, for example, to quickly move to all information related to an event to all information related to one of the participants). [<a href="http://paolo.evectors.it/">Paolo Valdemarin: Paolo's Weblog</a>]</blockquote> The When classification was always a key part of the overall vision of K-Collector. Being able to navigate sensibly based upon time & date is a very powerful concept. One of the last things I did within liveTopics before moving on to K-Collector was to implement time as an <a href="http://www.xfml.org">XFML</a> facet. This created a hierarchy of topics representing dates. For example there was a topic "2003". This in turn contained 12 sub-topics "Jan 2003" through "Dec 2003". Each of these topics contained a further division by date. Posts were linked to these topics based upon publication date. The upshot was that you could browse based on an increasingly specific date filter. We will be looking to implement When topics in K-Collector in conjunction with ESF events. The idea will be to link the event (perhaps defined as a What topic), with a particular point in time (a When topic), a place (Where) and possibly people attending (Who). Hence an event will form a glue which binds together a number of different topics in a context. Example: <dl> <dt>What</dt> <dd>blogtalk 2.0</dd> <dt>Where</dt> <dd>Vienna</dd> <dt>Who</dt> <dd>Thomas Burg</dd> <dt>When</dt> <dd>5-Jul-2004</dd> </dl> There are still issues to work out like: What about events which span multiple days? Do we represent time on the calendar? What about recurring events? And so on. But I think even a simple model which dodges many of these questions would be amply useful at this point. http://matt.blogs.it/entries/00001926.html This blog has many facets (well 2 anyway) Thu, 12 May 2005 11:19:55 +0100 After a gap of nearly 3 years I am, once again, publishing an <a href="http://matt.blogs.it/topics/resources/facetmap.xml">XFML facet map</a> of my weblog. The facetmap links the posts I have published to occurrences of topics I have used. I am also automatically generating a Date Of Publication facet which allows you to drill-down by year, month, and then date along with selecting topics. This can be demo'd at via Trav Wilson's <a href="http://www.facetmap.com/">FacetMap</a> service. The current map of my site is <a href="http://www.facetmap.com/browse/curiouser_n_curiouser">browseable</a>, as an example posts I wrote in <a href="http://www.facetmap.com/browse/curiouser_n_curiouser?s=0081016200&v=1">November 2003 concerning email.</a> http://matt.blogs.it/entries/00002584.html Browsable blogging Fri, 20 May 2005 13:49:39 +0100 Over at How to Save the World, Dave Pollard is talking about <a href="http://blogs.salon.com/0002007/2005/05/19.html#a1151">how to make blogs browsable</a>.<blockquote>The task of making weblogs' architecture more robust should be much easier. Weblog software with more dynamic information architecture would not only make blogs much more valuable to those browsing for information, they would make weblogs much more valuable in corporate environments. The current emphasis on adding 'tagging' information is, in my opinion, misguided: That would make their content easier to search, and might solve the information overload problem when they're embraced by keyword search agents, but it won't make them easier to browse. Much of the readership of weblogs is serendipitous -- people stumble on them (usually through search tools) when they're looking for interesting reading. Or, they blogroll a weblog because some of its content is of interest to them. What is needed is a way for people to browse through a selected subset of weblog content, all of the articles on a particular topic.</blockquote>I guess I'm a little surprised not to get a mention. My recollection (am I wrong?) is that I chatted to Dave about this (among other topics) back when <a href="http://paolo.evectors.it/">Paolo</a> and I were working with K-Collector. The basic idea is to allow authors to mark up their posts with topics (I still don't like the phrase tag in this context but I accept I may be in a minority) which are fine grained. Categories for me have always been too inflexible and unwieldy. The use of multiple topics allows rich description of a post. <a href="http://matt.blogs.it/topics/"><img style="padding: 10px;" align="left" border="0" src="http://matt.blogs.it/images/all-topics.png"></a>The magic part is what happens next. In my current experiment it means I can generate a page which allows my content archive to be explored by topics. Clicking a topic name takes you to a page that lists the posts, in reverse chronological order, associated (I would say tagged if I didn't think it was confusing) with that topic. Under each post is a link to the other topics associated with that post. Hence each post also offers a cross-reference facility throughout the rest of the content. It makes my blog into a fully-browsable content index, automatically. These topic pages like <a href="http://matt.blogs.it/topics/aggregators.xml">Aggregators</a>, <a href="http://matt.blogs.it/topics/psychology.xml">Psychology</a>, and <a href="http://matt.blogs.it/topics/blogging.xml">Blogging</a> are, if you check the URL, actually RSS2.0 feeds (with ENT topic metadata) being <a title="Read this for a bit of the background of how it was done." href="http://matt.blogs.it/2005/05/07.html#a1801">rendered in the browser</a> as HTML. But you could also subscribe to a topic like <a href="http://matt.blogs.it/topics/rss.xml">RSS</a> in an aggregator and only read my posts on that topic, blissfully ignoring what I write about <a href="http://matt.blogs.it/topics/politics.xml">Politics</a>, or <a href="http://matt.blogs.it/topics/microsoft.xml">Microsoft</a>. But there's more to come. I also publish a <a href="http://matt.blogs.it/topics/facetmap.xml">Facet Map</a> of my weblog in <a href="http://xfml.org/">XFML</a> format. The potential value of this is not just in <a href="http://www.facetmap.com/browse/curiouser_n_curiouser">improved browsing</a> because XFML also offers a way to <a href="http://www.xfml.org/spec/1.0.html#connectingtopicsconcept">connect topics together</a>. This offers us the opportunity to make sense of each others tagging schemes, harmonizing the view of data whilst allowing us to preserve our own preferences. Two level, even three level, categorization is possible by grouping topics together although things can get tricky at this point. The approach Paolo and I took was to use a simple 2-level structure comprising <ul><li>Who</li><li>What</li><li>When</li><li>Where</li></ul>At the moment I only use two of these. My topics are currently implicitly What and I have automatically generated Date of Publication topics in my facet map file. I might do the other again, maybe. What I've done others can do just as easily. As much as anything it's a mind-set issue. If you think in terms of categories you're thinking in terms of a rigid hierarchy. Topics are more granular and should be used liberally since the tools at the other end will make them usable by users, as I have done. http://matt.blogs.it/entries/00002639.html