Curiouser and Curiouser!

Curiouser and Curiouser! on topic-mapping http://matt.blogs.it/ RSS feed for topic topic-mapping Copyright 2006 Matt Mower Squib/0.1 self@mattmower.com self@mattmower.com en-gb Knowledge Centrifuge Sun, 26 Oct 2003 21:13:00 +0000 As we get closer to release I've found myself explaining what <a href="http://paolo.evectors.it/">Paolo</a> and I are doing with <a href="http://w4.evectors.it/">K-Collector</a>, what the product is about. So far the best explanation I have come up with is: <div style="text-align: left; margin-left: 40px;">Realtime Knowledge Management </div> What I've been trying to do is make the case that knowledge management and document management aren't the same thing. Most knowledge in a company begins life as a granule of information, micro-knowledge. It exists for a time and, if not exploited, likely dies away to be discovered again later if needed. Quite often the inertia that has to be overcome in order to turn such a granule into the sort of document you would load into Livelink, Documentum or some equivalent product is overwhelming compared to the perceived value, at that moment, of the knowledge itself. It is only when the cost of repeated rediscovery begins to bite that someone finally does the decent thing. The corrollary to this is that the benefits of knowing that you know this information are not felt until quite late in it's lifecycle. I would guess that, by the time a lot of information is formally documented, it's probably well on the well to being out of date or irrelevant (is this your intranet?) By contrast the K-Collector approach collects information while it is still fresh and combines it with other related information before presenting to users for them to see if it meets their needs. That which is good can be promoted to a more appropriate place (for example a <a href="http://www.socialtext.com/">Wiki</a>) Things which don't make the cut fade into the background but, crucially, are not lost. I imagine K-Collector as being a kind of knowledge centrifuge, spinning together all kinds of different bits of information and separating out the good stuff for you. At the heart of K-Collector and determining what makes 'the good stuff' are topics. Topics act as markers for points of interest around which information can be clustered. The Who, What, Where, When metaphor we have adopted is - we think - a really simple way of considering what is important to us all (although, prompted by <a href="http://www.henshall.com/blog/">Stuart Henshall</a>, I have been wondering about Why as a 5th W). I think of K-Collector as a kind of multi-dimensional database where each topic slices through the available information. We're working on some pretty cool topic related trickery for future versions that will take this idea and make it a lot more powerful. I'm hoping to get back to writing soon and share more about what we are doing. I hope this is useful in the meantime. http://matt.blogs.it/entries/00001682.html Topic Mapping Resources Wed, 25 Feb 2004 08:18:18 +0000 <blockquote><a href="http://www.steptwo.com.au/columntwo/archives/001121.html">New topic map tools</a>. Peter Van Dijck has pointed out a few new topic map tools and approaches: tinyTIM: a very small easy to use (50kb jarsize) in memory Topic Map engine. It implements the TMAPI interfaces, so one can work with TopicMaps via... [<a href="http://www.steptwo.com.au/columntwo/">Column Two</a>]</blockquote> Two interesting tidbits. http://matt.blogs.it/entries/00001908.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/00001925.html Presenting People Centred Knowledge Management at the CiG Thu, 29 Apr 2004 18:16:24 +0100 On Tuesday night I did my first proper speakers gig, giving a 20 minutes presentation of People Centred Knowledge Management (PCKM) to members of the <a href="http://www.cityinformation.org.uk">City Information Group</a> (I'll link to their event page when it's been updated). I had a great time doing the event and I've had some positive feedback - I hope everyone there got something out of it. My thanks to the <a href="http://www.cityinformation.org.uk/Committee/committee.htm">organizers</a> Jackie, Genevieve and Nick. <ul> <li><a href="http://matt.blogs.it/gems/presentations/CIG_apr_2004/CIG%20Presentation.htm">Web</a> version of the presentatin. (Should work in all browsers, but you know PowerPoint) </li> <li><a href="http://matt.blogs.it/gems/presentations/CIG_apr_2004/CIG%20Presentation_export.pdf">PDF</a> version of the presentation here. (447K)</li> <li><a href="http://matt.blogs.it/gems/maps/cig_apr_2004/Speakers%20notes.html">Speakers notes</a> (This will give you a better idea of what I said).</li> </ul> I'd like to express my thanks to <a href="http://paolo.evectors.it/">Paolo</a>, <a href="http://www.dijest.com/aka/">Phil</a>, <a href="http://www.hyperorg.com/blogger/">David</a>, and <a href="http://chocnvodka.blogware.com/blog">Suw</a> who were all invaluable in helping me to get prepared. I think it really paid off. Update: It occurs to me that really you don't get very much from my slides. The presentation was a lot about me talking, waving my arms and hopping up and down. You don't get that from PowerPoint. Next time I'd like to be able to webcast the presentation. Anyone have any advice about that sort of thing? http://matt.blogs.it/entries/00002142.html Teenage Taxonomies Sat, 01 May 2004 10:40:23 +0100 I've been thinking about how topics are organized in <a href="http://www.evectors.com/itkcollector/">K-Collector</a>. As of my writing the largest K-Collector site is <a href="http://w4.evectors.it/">W4</a> which currently has 726 topics defined in 3 classifications. <ul> <li>What - 410 topics </li> <li>Who - 227 topics </li> <li>Where - 89 topics </li> </ul> This is beginning to present some challenges. For example current interfaces in K-Collector clients for selecting topics highlite suggested topics but also display all available topics (under their classifications) for you to choose from. The problem is that, with this number of topics, it's all rather unwieldy. I already think that it would be advantageous to allow a further subdivision of the classifications to form a 3-level hierarchy. For example What could subdivide into things like products, protoocols, principles, patterns and so on (sort of modelled after Denham Grays <a href="http://denham.typepad.com/km/2003/11/information_gat.html">Information Gathering Template</a>). But this is not a complete answer. First I do not want to grow arbitrarily nested taxonomies. If you think about your own experience with menu bars, how often do you look more than 3 levels deep? And how irritating is it to have to poke around like that? Also the deeper the taxonomy the more effort has to be put into designing it and this is the domain of experts and to be avoided Second faceted classifications only work when they narrow things. As you descend the hierarchy you become more precise about the term you are talking about (Anything => any person => a member of a group => an individual). But for example, when I am writing a post about politics and economics I am really only interested in topics related to those subjects, i.e. I have cross-cutting concerns (like <a href="http://server2.hostvalu.com/pipermail/discuss_aosd.net/2002-January/000052.html">Aspects w.r.t. OOP</a>) I've been thinking a little bit about Peter Van Dijck's <a href="http://poorbuthappy.com/ease/archives/002822.html">suggestion</a> for nested facets. As Travis Wilson described it in a post to the facetedclassification list: <blockquote> <pre wrap="">For example, Peter's article poses a "Product Type" facet where "Cameras" is a heading down in the taxonomy somewhere. Cameras have certain properties like "Resolution" and "Lens Type" that just aren't relevant to, say, hubcaps. So "Resolution" and "Lens Type" are facets with a scope of "Product Type = Camera". A faceted navigation interface would expose them whenever a search was already restricted to "Camera". Otherwise, they're structured like every other facet you've ever seen.</pre> </blockquote> to see whether there is a role for nested facets. I'm still thinking about that. It's also possible that we could also use the many relations (thousands and thousands of them) that K-Collector builds up to create dynamic cross-cutting hierarchies. The idea here would be to take one or two dominant topics and then order all other topics according to how relevant they were. This should, in theory, put more relevant topics closer to your attention. No solutions yet, just questions & ideas. http://matt.blogs.it/entries/00002160.html A thought about nested facets Sat, 01 May 2004 11:01:52 +0100 Quick thought about <a href="http://poorbuthappy.com/ease/archives/002822.html">nested facets</a> while I was in the shower.  Paolo and I have chatted numerous times about typed RSS feeds.  The idea being that when <code><item></code>s don't come from blogs we may get different information about them.  For example an item corresponding to an event will have useful metadata which might correspond to some nested facet.  An item from a database feed of helpdesk tickets could have different (or possibility similar) metadata. Hmmm... http://matt.blogs.it/entries/00002164.html We built a better mousetrap, where are the mice? Sat, 08 May 2004 22:53:24 +0100 <blockquote><a href="http://matt.blogs.it/specs/ENT/1.0/">Wouldn't it be cool?</a>. Wourldn't it be cool if otehr people started parsing the ENT tags embedded in this post? That way if I talked about FOAF - for instance - someone like danbri could scarf JUST the FOAF posts and do anything he wanted with them! This would create an incredible two-way kind of capability because systems could then communictae back to me based upon what I said. I know I know -it's RSS2.0 but that becomes a really nice gateway to a world that has 75% market share of feeds(maybe even more.) [<a href="http://blogs.it/0100198/">Marc's Voice</a>] </blockquote> I think it would be pretty cool -- I've been hoping other would start grokking this for a while now.  Yes it's RSS2.0, yes it's not perfect, but it's here, now and I think that by the 80/20 rule it's good enough.  If there's anything we can do to help get ENT support included in other applications please let us know and we will do our best to help. I'd also like to see applications start using the <a href="http://matt.blogs.it/specs/SGUID/1.0/">SGUID</a> information that Paolo and I have had in our feeds for about the same length of time as we've been doing ENT.   Using SGUID one post in an RSS feed can refer directly to the permalink of the post it is quoting from.  Standard RSS2.0 only allows you to refer to the feed.  So for example, tags like: <blockquote><sguid:sourceRef>http://www.theobviousblog.net/blog/archives/000506.html</sguid:sourceRef> </blockquote> should allow a clever aggregator to thread posts. An aggregator that did topics and sguid-based threading, that would be nice... http://matt.blogs.it/entries/00002209.html (ENT2.0 mod RSS1.0) = 0 Mon, 10 May 2004 10:51:09 +0100 <blockquote>Time for ENT 2.0?. It's very interesting to read Danny's toughts about <a href="http://dannyayers.com/archives/002586.html">ENT and RSS 1.0</a>. Maybe it's time for a new release of the ENT specs, RSS 1.0 compatible. Oh... and what about Atom? [<a href="http://paolo.evectors.it/">Paolo Valdemarin: Paolo's Weblog</a>] </blockquote> I've certainly thought about things which, with the benefit of hindsight, I would have done differently.  I was never comfortable with having the topic name as the text content of the <code><topic></code> element and I've no idea why I did it, there are other bugbears in there too. I'd also like to give more thought as to how ENT feeds can be supported by topic map resources in real applications.  At the moment we don't publish XTM or XFML maps out of K-Collector but we could (I used to publish XFML from liveTopics but those files got big!) Lastly I would really like to make a push for ENT support in other applications.  It seems a shame to me that, more than a year on, no other applications seem to have picked up on the benefits topic based aggregation offer to users. http://matt.blogs.it/entries/00002218.html 800 topics bouncing around Wed, 12 May 2004 09:03:13 +0100 <blockquote><a href="http://www.community-intelligence.com/blogs/public/archives/000264.html">Visualizing social networks</a>. Increasing the capacity of people and communities to visualize their (online) social networks is essential to the evolution of Collective Intelligence. [<a href="http://www.community-intelligence.com/blogs/public/">Blog of Collective Intelligence</a>] </blockquote> I still really want some kind of visualization for K-Collector.  We're generating lots of data, relationships between topics and posts, topics and topics, people and posts, people and topics, people and people and more all the time.  When all you have is tables full of data it's hard to work out what might be interesting or valuable. The <a href="http://prefuse.sourceforge.net/">prefuse visualization library</a> looks very neat although I have done similar things in the past with <a href="http://www.touchgraph.com/">Touchgraph</a>.  However a key problem remains, when dealing with large data sets, how to make a meaningful subset of them usable.  Do you really want a graph with 800 topics bouncing around? http://matt.blogs.it/entries/00002231.html Getting creative: five social tools to give you an edge! Mon, 28 Jun 2004 23:33:29 +0100 Tomorrow I'm speaking at the <a href="http://www.gurteen.com/gurteen/gurteen.nsf/id/x0007eaf6">Gurteen Knowledge Conference</a>.  My topic is: <blockquote>Getting Creative: Five social tools to give you an edge! </blockquote> My aim is to establish the link between creativity and social networks, and then to show how tools like blogs, wikis, instant messaging and topics can be combined to help build a culture of creativity. This is only my second speaking engagement.  Wish me luck! http://matt.blogs.it/entries/00002369.html Visualizing concepts in weblogs Tue, 06 Jul 2004 13:10:30 +0100 <a href="http://anjo.blogs.com/">Anjo Anjewierden</a> & <a href="http://blog.mathemagenic.com/">Lilia</a> are demonstrating a very interesting tool for visualizing conceptual relationships within weblogs.  It uses technology similar, but more advanced, to K-Collector for analyzing the content of weblog posts and drawing out concepts which can be analyzed and displayed as a network so that you can explore the relationships.  It also allows comparison of terms between different weblogs. Very intriguing! http://matt.blogs.it/entries/00002407.html DelphiGroup: Making the case for taxonomy Mon, 16 Aug 2004 11:56:57 +0100 I'm reading an excellent report from <a href="http://www.delphigroup.com/">Delphi Group</a> called Information Intelligence: Content Classification and the Enterprise Taxonomy Practice (see <a href="http://www.delphigroup.com/coverage/taxonomy.htm">here</a> for a download link). The gist of the report is that, for enterprises, search is not enough. The report makes the case that whilst search technology has improved a lot in recent years, and continues to improve, the majority of professionals still find it an unsatisfactory way to work and often spend in the region of 20% of their time searching for information. Often cited problems were constantly changing information and a lack of precision about what they were looking for. The report then makes a case that introducing taxonomy based services can significantly improve performance and save money by eating into that 20%. Whilst the report is funded by a number of companies with a vested interest in taxonomy (for example Autonomy or Verity) the case seems to be well made off the back of a credible piece of research (which is a follow-up to similar research done last year). My summary: <ul> <li>82% of users do not have access to a centralized point of search & information across information systems</li> <li>The provision of a singular navigational front end (e.g. taxonomy) and omnipresent search tool that collectively aggregate disparate content resources, can, from an end-user perspective, deliver the simple single point of access that many users strive for.</li> <li>Lack of organisation of information is the number one problem in information management & retrieval.</li> <li>If professionals are spending 20% of their time (or more) looking for information then this results in an opportunity cost & represents a runaway expense item in many organisations.</li> <li>Keyword search assumes you know what you are looking for & that it an often erroneous assumption.</li> <li>75% of people surveyed during a Yahoo market research project preferred browsing to searching.</li> <li>In some instances it is easier to discover information about a particular subject if you see it in the context of related thought. Browsing encourages associative thought.</li> <li>The availablity of taxonomy eliminates the need for the researcher to completely understand the subject before issuing a query.</li> <li>Browsing via a taxonomy in essence provides an education on the subject and lends insight into the issues or facets of the subject.</li> <li>The number one source of frustration with search of on-line content is the fact that the content they search for is constantly changing, which both frustrates the user and reduces the effectiveness of simple search.</li> <li>Use of a taxonomy can provide a dynamic bookmark so to speak, a one-stop-shopping guide to all relevant content on a subject. Return to a subject node exposes the latest and complete collection of content about that subject area. This addresses the number one cause of frustration, the dynamic, volatile nature of information sources.</li> <li>Navigation of a well-designed interface to information on a web site/portal automatically directs the researcher to other relevant topics.</li> <li>The tagging effort represents another process that a business must undertake in order to obtain the benefits of a taxonomy. In some cases this could be done manually. But this approach is not easily scalable.</li> <li>Will authors be willing or available to perform this classification manually?</li><li>68% concur that the process of locating & retrieving the information needed to effectively execute their jobs is difficult and time consuming.  Not a single respondent strongly disagreed with this statement.</li> <li>Whilst users see some improvement in information retrieval over the last 2 years, their attitude towards its level of difficulty remained virtually the same.</li> <li>Respondents overwhelmingly pointed to the fact that business content is constantly changing and has to be continually relocated.</li> <li>Findings pointed more towards the speed and ease of use of retrieval environments and less to effectiveness, as the primary point of pain amongst todays business people. </li> </ul> I think this report is very well worth reading to anyone interested in search, taxonomy, or knowledge organisation. Of course I too am baised because I think <a href="http://www.evectors.com/itkcollector/">K-Collector's</a> integrated approach addresses a number of the concerns raised by this report. http://matt.blogs.it/entries/00002450.html