I'm working with a group that has customized their mediawiki installation so that they can create multiple wiki instances, each with their own database, that have no knowledge of each others existence. This was done for security and privacy reasons.
They now would like to be able to search a number of these multiple wiki instances from within one newly defined wiki. I'm thinking the method they are attempting may be as simple as creating a wiki home page with links to each of the wikis they would like to search.
Is this possible? Is the wiki search driven by a crawler that would follow the links on that new wiki home page? If not, is there an approach I could follow to be able to provide search capability against a select number of these individual wikis under one umbrella?
Thanks for any suggestions!
Hi,
On Mon, Mar 21, 2011 at 3:49 PM, Tod listacctc@gmail.com wrote:
Is this possible? Is the wiki search driven by a crawler that would follow the links on that new wiki home page? If not, is there an approach I could follow to be able to provide search capability against a select number of these individual wikis under one umbrella?
I've never tried it personally, but I think SphinxSearch may be worth a check; it works directly against the database.
Marco
I've never tried it personally, but I think SphinxSearch may be worth a check; it works directly against the database.
Sphinx itself can index multiple wikis.
SphinxSearch extension can query any number of indexes.
The problem is, extension currently assumes all of them are in the same wiki database. It would not show correct links or matching snippets for hits from other wikis.
Svemir
On 03/21/2011 2:36 PM, Marco Schuster wrote:
Hi,
On Mon, Mar 21, 2011 at 3:49 PM, Todlistacctc@gmail.com wrote:
Is this possible? Is the wiki search driven by a crawler that would follow the links on that new wiki home page? If not, is there an approach I could follow to be able to provide search capability against a select number of these individual wikis under one umbrella?
I've never tried it personally, but I think SphinxSearch may be worth a check; it works directly against the database.
Marco
Thank you all so much, these are excellent ideas but given the relatively short time frame I have to implement I don't know that they are workable.
Is there a possibility that I could instead (easily) merge the handful of individual wikis into one consolidated wiki and implement a more enhanced search against it instead? I've no experience performing a merge like this so expert advice would be appreciated!
- Tod
Hi,
An approach we take with the Kete app is to have each instance have an OpenSearch interface available (i.e. search results can be obtained in RSS/Atom form from a URL that contains the search terms) and then allow each instance to return its own search results in a main results area with a side bar of "external search sources". It's intuitive, works with other services as well (http://digitalnz.org, Twitter, YouTube, etc.) rather than just Kete, and breaks down silos of information.
Here's an example of what it looks like in practice:
http://horowhenua.kete.net.nz/en/site/search/topics/for/levin+fire+brigade?s... -- external searches under "more resources"
Here's a paper that describes the approach:
http://kete.net.nz/site/documents/show/65-kete-and-koha-integration-built-on... -- (click 'view' to download)
Or you can check out the slides from a presentation I did at KohaCon10 of the paper:
http://kete.net.nz/site/documents/show/66-kohacon10-kete-and-koha-integratio...
Finally, there is some example code (a Ruby on Rails engine, so only good as example code):
https://github.com/kete/external_search_sources/blob/master/README.rdoc
It has an admin interface for setting up your external search sources and then provides some helpers that take search terms and return the parsed and laid out results.
If you end up with many search sources, consider aggregating them, much as we use the Digital New Zealand service. For that we use each Kete's OAI-PMH repository to grab all public metadata (i.e. all public text) in Dublin Core about the Kete instances records so that DigitalNZ's OAI-PMH harvester can feed its Solr instance.
Cheers, Walter
----------------------------------------------------------------- Walter McGinnis Kete Project Lead (http://kete.net.nz) Katipo Communications, Ltd. (http://katipo.co.nz) http://twitter.com/wtem walter@katipo.co.nz +64211241794
On Mar 22, 2011, at 3:49 AM, Tod wrote:
I'm working with a group that has customized their mediawiki installation so that they can create multiple wiki instances, each with their own database, that have no knowledge of each others existence. This was done for security and privacy reasons.
They now would like to be able to search a number of these multiple wiki instances from within one newly defined wiki. I'm thinking the method they are attempting may be as simple as creating a wiki home page with links to each of the wikis they would like to search.
Is this possible? Is the wiki search driven by a crawler that would follow the links on that new wiki home page? If not, is there an approach I could follow to be able to provide search capability against a select number of these individual wikis under one umbrella?
Thanks for any suggestions!
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
wikitech-l@lists.wikimedia.org