Sunday, August 03, 2008

BPN 1178 Searching is still not cool

Whoever thinks that the search fight is over and Google has won, is wrong. Google is a marketing machine with a fast search engine, but excessive and dispensable results. On the marketing side it will be hard to win the fight, but in the field of search results lot can be won. Now a new entrant has presented itself: Cuil with a rather snoring blurb: rather than rely on superficial popularity metrics like Google does, Cuil searches for and ranks pages based on their content and relevance.

It is clear that Cuil is in its beta phase; no advertisements, just a search machine and results. I had a look at it and did my own benchmarking with my name, the name of the blog and the name of our company. I compared the results of Cuil with the results of Google. I started with my own name Jak Boumans and was presented with the following results: 1827 results from Cuil and 7690 results from Google. Cuil offers fewer entries. Yet also Cuil has duplicate results and non-relevant results from either Jak or Boumans. When the duplicates are removed the number of relevant results could be diminished with 40 per cent. Cuil attempts to illustrate every entry with photographs and this leads to hilarious results. An old photograph - not taken by me and one I would have destroyed - is often used in connection with bibliographic data. But after 10 pages with results, a photograph does not work anymore. I also tested the name of this blog: Buziaulane. Google generated 5090 search results and Cuil 2143. Looking though the results in Cuil, the results concentrate on the name of the site and the occurrences in other blogs and on other sites. Multimedia assets are less abundant and in fact soberly applied. The last part of the benchmark took the company title and compared these: 1,3 million search results in Google and 4,7 million search results in Cuil. Funny part is that Cuil does not recognise the name as a company name, but as a method of reporting with electronic media. When I add the legal company form, VOF, there is only one result of an activity of the company; when the legal name is punctuated as it should, there is no result at all. So not even the chamber of commerce directories have been culled by Cuil. When I refine the term with the present and former company base, I get completely different results. When I add Almere to Electronic Media Reporting the search engine comes up with some results of the blog. When I added Utrecht, the first company base, there are more diverse results.

Am I impressed by Cuil? I loved Cuil in the first two benchmarks of my name and my blog’s name. But the results of the company are really bad, especially as the company names in The Netherlands are registered by the Chamber of Commerce. The addition of the multimedia assets is too much ad random, leading to portrait photographs next to for example the method of reporting with electronic media.

I recently also looked at Exalead. Putting this search engine through the same benchmark test, I got the following results:
Term/Exalead/Cuil/Google
- Jak Boumans: 672 results/1827/7690
- Buziaulane: 1578 results/2143/5090
- Electronic Media Reporting: 908,367 results /4,706.000/1,300.000

The retrievability of words and composed terms is still very inexact. The addition of multimedia assets is not an added value. So I guess that Google will stay alive for the time being.

Blog Posting Number: 1179

Tags:

No comments: