Karen Johnson just posted a blog entry on testing search algorithms. I’m soon going to write an article on accidental test coverage. The intersection of these two topics can be found here, based on a search I recently did in the frequently asked questions list on an airline Web site.
Search Results for “power for laptops”
Note a couple of interesting things.
First, the word “for” is highlighted in the list of links to found items, but “laptop” is not.
Second, when I go to the answer to item 1, the words “power” and “laptops” appear in the found article and are highlighted. If I change the search terms to “power for laptop”, the same list is returned with “laptop” highlighted, but “laptops” isn’t highlighted in the found item. So the search algorithm appears to be using the stemming that Karen talks about, but the highlighting algorithm isn’t. From this I infer that the highlighting and search algorithms aren’t quite talking to each other.
Second, check item 3 on the list. When I go to that item, the word “weight” is highlighted in the body of the found item. The word “for” appears in the text (and is not highlighted), but neither “power” nor “laptops” does. In order to check that the found word should be highlighted, I next did a search for “antlers”, and the only item returned was item 3 above. All instances of “antlers” were highlighted in the item.
And, of course, there are only three weeks left to register for CAST.
Hi Michael,
Item 7 has none of the words, “power”, “for” or “laptops”. It contains the word “for” within the word “information”. Is this stemming too?
Cheers,
Inder P Singh