Intranets

An account of the 2006 Search Engine Meeting, held this year in Boston, USA, in April 2006.

2006 Search Engine Meeting

Until the recent arrival of the Enterprise Search Summit the Infonortics (http://www.infonortics.com ) Search Engine Meeting was the conference to attend if you wanted to track what was happening in the technology and business of search. The Search Engine Meeting started a decade ago with the mission of “bringing together commercial search engine developers, academics and corporate professionals to learn from each other”. Over the last few years the emphasis on web search has been tempered by more attention to enterprise search. This year’s Meeting was held in Boston on 24/25 April. It attracted around 150 delegates, mostly from the USA.

The conference was opened by Dave Girouard, General Manager, Google Enterprise, who made the point at the outset of his presentation that although the market for web search is probably worth $10bn the enterprise market is worth no more than $1bn. He felt that one of the reasons for the low adoption of enterprise search was that to date it had failed to deliver value to the end user. In his view even a single keyword search in an enterprise application should deliver good results if enough care is taken over the search environment. He felt that in general there was much in common with enterprise users and web users, and that much of the development work was taking place in the public search arena and then being adopted by enterprise search, such as faceted navigation (a major theme of the conference), clustering and visualization. Overall he felt that it was important than in an enterprise implementation the search box was in the centre of the page, not hidden at the top RH end of the search bar.

Many of the papers at the conference were concerned with how to effectively search ‘the long tail’ (see http://www.wired.com/wired/archive/12.10/tail.html ), a major issue in enterprise search because a searcher has to be confident that they have found all relevant documents, not just a sample – which leads me neatly into an excellent paper by Steve Arnold on the way in which search engines are ‘managing’ relevance. An essay based on his presentation can be found at http://www.arnoldit.com/speeches.html and is an essential read. Steve may have slightly overstated the case, but not by much.

The main sessions at the conference were on Searchers and Search Behaviour, Faceted and Federated Search, Text Mining, the World of the Web and finally Web Tools and Intelligent Tools. Virtually all the papers (but not the Google paper) can be downloaded from http://www.infonortics.com/searchengines/sh06/06pro.html but many are quite substantial files. My favourites were the papers by Tony Gentile, Joseph Tragert, Tom Reamy and Boerge Svingen. Too many of the speakers failed to appreciate that they only had 25 minutes for their presentations, and really failed to pace themselves, rushing through far too many slides. A few presenters from the vendor community were rather too much in PR mode but yet had not paid up for a tabletop at the exhibition.

Overall I think I will give this year’s Search Engine Meeting 7 out of 10. To me it is struggling to find a balance between the public web and enterprise search, which is not at all easy because of the technology overlaps referred to above. However despite the technology overlaps there are quite a number of major differences and these did not really get an airing. For example there was no speaker who had actually implemented an enterprise search application. One final observation was the way in which the term ‘metadata’ was used only for describing content. Any given item of content has many other metadata elements (http://dublincore.org/) and this was invariably overlooked by the presenters.

I’ll also be at the Enterprise Search Summit that takes place in New York at the end of May http://www.enterprisesearchsummit.com/ and will report back in the next issue of eLucidate.