Search

Learn how KnowledgeOwl search works work along with what features and functionalities are available.

Search Overview

Search in KnowledgeOwl has multiple layers to get readers to appropriate content. You do not need to add tags or keywords for search to work. KnowledgeOwl automatically indexes all of your article content for search, including text in PDF files. Words in article titles have more weight than words in the body of an article, and words in headers have more weight than words in normal text within the body. Search includes the following types of results:

  1. Autosuggest Results
    Articles are suggested below the search box as the reader types a search phrase. The typeahead results are based on an exact match of the search phrase in article titles, and clicking on the suggestions goes straight to the article.
  2. Exact matches and search phrases
    Any exact matches within article titles are the first search results on the full results page displayed after hitting enter, return, or search in the search box. Articles with matching search phrases anywhere in the content are treated as exact match results.
  3. Tags
    If the search phrase is a tag, articles with the tag display below any exact match and search phrase results.
  4. Learned results
    Learned results are displayed below exact matches, search phrases, and tags search results. Learned search looks at the root or stem of the word and returns articles that match ordered based on where the search term appears (titles have more weight than headers which have more weight than body content) and past search behavior (articles that are clicked for a search phrase  move up in results over time).

You can customize and extend KnowledgeOwl search using many features and functionality including:

  • Search phrases
  • Tags
  • Synonyms
  • Filter by top-level and second-level categories

Autosuggest and typeahead

As a reader types a search phrase into the search box, KnowledgeOwl will suggest any articles with titles containing an exact match of the search phrase. If the reader sees the article they want suggested, they can click to go straight to the article bypassing search.

From the data we've collected, most readers search using keywords or phrases rather than typing out full questions. For example, readers are more likely to search for "tags" or "add tags" rather than "how do I use tags" or "how do I add tags".  Exact match takes advantage of this and displays articles with the current search phrase in the title, allowing readers to quick see the most relevant articles based on what they are searching for.

Autosuggest works in both the knowledge base search box and the application, allowing you to quickly get to articles if you start typing a word or phrase from the title. 

Tags for search

Tags are a way to organize and relate articles with your knowledge base. Tags are different than keywords – you do not need to add tags to your articles for them to be returned in search results. By default, all the words in your article, including both the title and body as well as PDFs, are indexed for search.

Tags add a piece of metadata to an article, telling you or the reader more about what type of article it is or what the topic is. By default, tags are displayed in search results and readers can click on them to view all articles with the same tag. For example, you might have a tag called "troubleshooting" which tells you and the reader than this article is about troubleshooting and that there are probably other articles which are about troubleshooting as well.

You can search for all articles with the same tag by inputting a colon followed by the tag name. For example, putting ":troubleshooting" into search would return all articles with the tag "troubleshooting".

Tags do influence search results, and tags results are displayed below any exact match or search phrase results. For example, a search for "troubleshooting" would return articles tagged "troubleshooting" after any articles containing the word or search phrase (those would be exact matches).


Learned search

Learned search augments search results when there are gaps in autosuggest, exact match, search phrases, and tags. Learned search works in the following manner:

  1. Search terms are simplified to their root or stem to account for variations of words (tag, tags, tagging, tagged, untagged, etc).
  2. Results include synonyms which can be defined in your synonym library. Searches for synonyms act like searches for the base word.
  3. Results are ranked based on where the search term appears in the articles. Results with the search term in the article title are weighted higher than those with the term in the body, and results with the search term in headers within the body are weighted higher than those with the search term in the normal text.
  4. Articles that are clicked on for a specific search phrase will move up in the search results over time, which is called learning.