Curiouser and Curiouser!

Curiouser and Curiouser! on information-management http://matt.blogs.it/ RSS feed for topic information-management Copyright 2006 Matt Mower Squib/0.1 self@mattmower.com self@mattmower.com en-gb 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/00002445.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/00002637.html