Searchable Knowledge Base

The TRIST website has a collection of documents that can support users in the use of TRIST. This collection of documents is known as a ‘Knowledge Base’ (KB). They are located within the website in a subfolder called ‘documents’. In the first example, consider adding a TRIST search interface to the frontend of the web-site to allow visitors to search the KB. The search pages are located in the parent directory to ‘documents. To set up the watch folder for this kind of search I would first navigate and select the documents folder. The URL would be the reference to that folder from the location of the results page. In this case that would simply be ‘Documents’.

Taking an example thats a bit harder, the search pages are within their own subfolder called ‘Search’. The navigation to the folder would remain the same, but the URL must reflect the different link between the results page and the target documents. In this case the link must first traverse up a level before it will find the ‘Documents’ folder. The URL would thus be ‘../Documents’. The same goes for if it was required to traverse up two levels giving ‘../../Documents’ as a URL.

Alternatively you can supply the full URL of the folder from root level. This would still generate the same output document but in a less controlled manner. This would be URL ‘http://www.trist-online.co.uk/documents’.

Extending this principle, you might want to index the contents of another site but provide search results based upon those documents as well. Consider the search.trist-online site. This site can return results from trist-online which is an independent site. It does this because we use the root level link reference that I commented on in the previous example. By supplying the full URL the results page can be located anywhere and the results still accesible. This would even work to the extent that an intranet accesible search and results page can link to pages on the internet and providing the visitor has an internet connection as well, they can still jump to the results.

The only constraint is that this version of TRIST requires the source file to be locally available to the index tool.