Search

What is the result of using the * Wildcard character when searching answers?

Environment:

Knowledge base (KB), Search, Phrase Stemming

Resolution:

When searching KB answers while using the Wildcard character, some results may seem unexpected at first review. The reason for this is that Phrase/Word Stems are also in play and take precedence over the Wildcard.

1. The result of searches including the wildcard is different for searches that use a phrases table from those that don't.

e.g. (non-phrases table search) If a wildcard is added to the end of a string when searching contacts, anything that matches the string or is longer than but includes the string will be found.

e.g. (phrases table search) When searching answers, which uses the ans_phrases table this will not necessarily be the case. The reason is Phrase/Word Stemming.
 

2. When answers are being searched. The wildcard brings back the results of answers that contain words or parts of words that do not have a proper word stem, unless the search is smaller than the stem. If the partial word is less than the stem it will find the stem. e.g. "insu*"(with wildcard) does locate the word stem "insur" but "insura*"(with wildcard) does not.

The reason for this is that Word Stems are only found when a complete prefix or suffix can be removed. e.g. A partial word such as "insura" or "emergen" doesn't contain either so they do not result in a stem.

3. Stemming applies first, then the Wildcard.

In the case of "insura*"(with wildcard), because it's not a word that contains a suffix or prefix, a Word Stem is not found. This eliminates proper Word Stems and leaves the wildcard searching for results that are in the phrases table but no proper words as those have already been eliminated by Stemming.

This leads to very few or zero results which can be misleading, especially when you already know the word "insurance" for example has many related answers.

4. Some results that might be found could be without proper punctuation (proper punctuation is necessary for search to recognize real words), misspelled or part of an email address such as "insurance@companyname.com" or something like "insurance-company" or "insurance/retirement". If using the "insur" example (insur being a proper word stem), without the wildcard those additional results will not be found.

Note:  A short document with tips and a guide to wildcard searching answers is attached.