Distinguish between public and private areas

[andi]

Hello,

we have public areas in our webspace and private areas with a login. Secondly sometimes we have different content for the same page depending on user logged in or not.

How can we handle this? Have we index the areas twice: one for not logged-in user and one index for logged-in user? This results in scanning large parts of the website twice - what about performance?

How can we distinguish search results depending on login? User not logged-in should only see the results of the public area while logged-in user should see hits inthe search results for different page content and additional pages of the private area.

Can you post me some concepts what is possible with ZoomSearch V5 and how to implement (only some basic guidelines)?

[Matt]

I would be interested in some advice on this as well, our Intranet is in a similar setting.

[David]

For indexing sites that require a login see this FAQ about authentication

I think there are 4 options,

1) Have 1 index with a public and private categories (using the Zoom categories function). In this case you need to write some custom scripting to force public users to select the public category.

2) Have two sets of index files and two search functions. A private set and a public set. Then put the private set within the private section of your web site.

3) Wrap the Zoom search script in a script of your own (using the XML output option in the CGI is good for this, because XML is probably easier to parse than HTML). Your script would filter the search results based on the URL, and remove URLs that user was not meant to see. This can allow much finer control than just public of private. You could have dozens of users groups each with access to different groups of documents.

4) Similar to option 3. But instead of writing a filter, you put the filter changes directly into our script (this still requires some custom code however, by removes the parsing step).

Option 2 requires two passes of the web site. The other options are single pass.

But if you have a page or document whose content changes depending on the logged in user, then you don't really have *one* page, you have multiple pages. Potentially a double pass is required for this case, if you need to pick up all variants of the same document. There is no inefficiency in this case because each pass gets a different page.

[Matt]

Can you see any application of MasterNode in this context? Maybe treating secure and public sections as two sites?

Thanks!

-Matt

[David]

MasterNode can aggregate the results from several search engines (slaves) and distribute the search load across several PCs. But it isn't going to directly help with the problem of returning different results to different users.