In Sigma 1/2006 the article of Annegrete Wulff about user orientation of Statistics Denmark deals with a question worth to be discussed in detail. She says: “We have realised that a good search facility is crucial as many people are not confident with the way a statistical office structures information. To many users, this means that searching should be possible in everyday terms”.
Over many years metadata have been an important theme in the community of statisticians. There are several layers of metadata:
– data describing surveys and their variables (cf. nomenclatures),
– data describing tables and helping to understand them correctly,
– data describing the content of NSI’s information offer and helping users to find the information they need (discovery or descriptive metadata).
The discussion about “discovery metadata” – helping users to get closer to the huge information offer of NSIs – has been started in several institutions.
The UNECE work session on Statistical Dissemination (12-14 September 2006) has an own topic (How to present metadata) with very interesting contributions
Eurostat has developped 1000 keywords helping users to structure their search (click the button K in the search mask). The idea is also to implement these keywords in the national websites and to assure a mulitlingual search over European statistics.
In Statistics Switzerland the whole information offer is structured by 21 topics. The publications (print and electronic) are integrated in one of the 21 topics, the Statistical Yearbook has 21 chapters and the thematic part of the Swiss Statistics Website has the same 21 chapters.
The Statistics Encyclopedia (French and German only) – the database with files to download: tables, diagrams, maps .. – offers like Eurostats search a list of keywords in alphabetical order.
Moreover users can go to a list of keywords pointing to the E-mail adress and phone number of specialists in Statistics Switzerland. Tomorrow these keywords will be harmonised with Eurostats keywords and they will also point to specific pages in the Statistics Portal or to intermediate pages explaining the theme in detail and giving hints to more detailed information.
Statistics Norway gives an alphabetical list of discovery keywords on the homepage. Opening a keyword there are informations about the topic and also metadata helping to understand this topic better (try birth!).
http://nemis.cti.gr/
http://rollyo.com/scbalfy/international_statistics/
http://rollyo.com/scbalfy/eu-25_statistics/
Statistics Sweden participated in EU research project NEMIS about text mining with a workgroup named “User aspects and relations to official statistics” , http://nemis.cti.gr/wg5.htm.
Some of these experiences from that project have inspired me to design two search-engines, which seek only in International statistics and EU-25 statistics. See links above, have try and let me know what you think.
Hi Alf
Good to hear from you! I have made some tests with your search engines and compared them with results from yahoo and Google. The results I got were focussing on statistics (good!) but the sources seemed to be quite old.
My questions:
In which intervals does yahoo reindex the list of sources?
Do you use this special search in scb’s website?
Would you recommend to use it in other statistical websites?
Is there a short description of the search syntax available? (.and. did work ..).
Hi Armin,
The advantage of Rollyo is that you know where you are searching and that you can include up to 25 sites when you design the search engine. I do not know about details in the index and syntax of Yahoo, how often it is updated or what kind of ranking. My experience is that Google is more complete and that its page ranking is outstanding. A pity you can only easily address all sites or a single site. We are using Google at our site and we have been looking into the possibilities to use Rolloyo for searching simultanously at our 25 responsible agencies for Sweden statistics. But no conclusions sofar … It would be interesting to do a more comprehensive evaluation of different alternatives. Perhaps a seminar with participation of different actors could be a starting point. No experiences on such issues at International Marketing and Output Database Conference 2006? I remember I raised the question in 2005 for the agenda of 2006.
Im glad I found your site. Please post up more pictures!