Archive

Archive for August, 2006

Miranda GDS Indexer is released

August 11th, 2006

Microsoft to use click-through as a search relevancy indicator

August 9th, 2006

The other day Microsoft Research published an article on using click-through metrics to improve search results. While the article in a precise science manner proves effectiveness of the method, a stupid question remains open to me: how can a click-through tell about relevancy of the results, if user does not see a page behind a link, when clicking on a result?

Anyway for those enjoing maths here is the link to the article

Uncategorized

Miranda GDS indexer v2 is submitted to miranda-addons

August 9th, 2006

Yesterday I submitted an updated version of Miranda plugin for Google Desktop to the Addons section of http://www.miranda-im.org

The version has some fixes (quite minor in terms of code) and basically is just recompilation of a year-old Vladimir Vainer’s version.

The project repository is here: http://code.google.com/p/miranda-gds-v2/

Uncategorized

Eurekster

August 8th, 2006


Just spent a half of hour evaluating Eurekster “social search”. The idea is promising, however search results seem completely irrelevant at the moment.

It could be that they need more experts to come to the “swicky” search engine, but without giving more prominent results first, they are fated to fail. You can see comparison of search results for a hot topic between Eurekster and Google. I took “Lebanon” as an example. The screenshot of the comparison is here:

Uncategorized

Looking at Microsoft Enterprise Search approach

August 3rd, 2006

Back in May Microsoft has published their approach to enterprise search here

Quite interesting is that all enterprise search is going to be centered in Sharepoint 2007 with ADO.NET adapters to all data in the enterprise.

They also introduced a paradigm of using social network in knowledge search within enterprise, so called “Knowledge Network for Office SharePoint Server 2007″

The paper poses three major differentiators between web and enterprise search:
1. Link structure
2. Hierarchy across sites
3. Security

And still it does not answer the question of finding an analogue of PageRank for enterprise.

Could it be achieved by indexing mails and IMs of the colleagues with some bayesian algorithm and counting quotes of indexed data sources as PageRank?

Uncategorized

Small companies cannot afford corporate search?

August 3rd, 2006

In this quite old study

40% of corporate users reported that they can not find the information they need to do their jobs on their intranets.”

Jakob Nielsen sayd back in 2004 that “it comes to more than US$1trillion per year that’s being wasted by people not being productive when they’re on the computer.”

The market of corporate search grown 32% and achieved $976m last year accroding to recent IDC study, however it seems that nothing changed for small and medium sized companies since 2004. There are big players on corporate search market (pretty much all of the big players are present on this market now), however the price point for simplest corporate search solutions is just too high for pro-sumers or SOHO users to afford it.

Is it the time for the new killer app? Would it be a hosted application, peer-to-peer or server based app? Is there any study available on small business market for team/company search solutions?

Uncategorized