May 17 2011

Custom Search Labels

Labels are the essence Google Custom Search Engines. Labels are used to include or exclude websites from search results (i.e. FILTER mode) or to give specific websites higher priority (i.e. BOOST mode) .

Basically There are two sorts of labels — background labels and refinement labels, and for Google these two represent different entities, i.e. one works in the background while the other works in the foreground . However, in the annotation file background and refinement labels are look the same (i.e. you can’t distinguish between the two). In addition, the Custom Search engine unique ID is used as a label in that file. Notice that all the search engines in one account are sharing the same annotation file and that the CSEs are separated only by their context file. This explains the conception of context file, i.e. a file that declare the context of the CSE in the annotation file.

But what all this means in practice? Suppose we want to create a CSE for newspapers but we want to take out the business sections. A conventional way to do that would be to create a “business” background label in the context file and then assign “mode=’FILTER’ weight=’-1′” and Than we will have to annotate the corresponding subdomains/subdirectories for every newspaper. This will force us to edit two XML files and then upload them through the advanced section in the GUI. A cleaner way to do this would be to create a “business” refinement label, to annotate all the business parts through the GUI and than add only one line to the context file in order to omit them.

We might also want to add “IgnoreBackgroundLabels” sub-element with true value to see the refinement label results or to set up 0 refinement labels in order to hide this label title. In addition, we might want to add to this label top=”1″ to give the users a glimpse to the business refinement label.

Now, as I mention above, all the CSEs in one account share the same annotation file so we can create a spacial CSE for annotating background labels of other search engines on the same account . In This case, the refinement labels on this CSE will function as a background labels on the other CSEs and so will be invisible on these search engine. This will allow us to annotate background labels from the GUI and use them for more than one search engine.

Yaniv Kimelfeld is the author of the website Topical Search, which focus on vertical and topical search engines.