We have worked hard to make the search in Knowledge as effective as possible. This guide describes in more detail how the search function works, so that you who work daily can ensure that your guides appear on the right keywords.

What fields does Knowledge search in?

• All words in the title
• All words in the body text
• All tags you have entered
  
  

How does the search engine work, what ends up at the top?

We use a third-party product for the search called elasticsearch. It gives us a good base plate to build on, because it already works as one would expect from a modern search function out of the box.

• It looks at the words you have entered and ranks documents that contain all words higher than those that contain few words.
• It knows that words close to each other are more relevant than words far apart.
• It places greater importance on uncommon words than common ones, so documents that match the most uncommon words will rank higher. 
• If a word is used multiple times in a document, that document will rank higher.
• We have also configured it so that it can find words that are inflected in another way, even though it is not the form you wrote. This setting is language-specific, which means that it is important that you do not mix different languages in the same knowledge bank.
• We have also added misspellings, so that the user gets hits even though one or two letters ended up wrong. The misspelling is made so that it can handle 1-2 letters incorrectly, and only looks at slightly longer words. Numbers are excluded as typing another number is rarely a "misspelling".
• For the automatic search function shows a list of search results for each keystroke you make. Since this means that the user has not always written his word clearly, we search for words with the same beginning / prefix. This means that searches on e.g. "vision" gives hits on "visionary".