Understand how Search Works in Self Help

When your end user searches for content in Self Help, the search term goes through a series of algorithmic processes (described in the following) before it can display the most relevant content.

Step 1: Analysis of the search phrase

As soon as you enter the search term, it is analyzed, and stop words such as a, at, be, but, by, for, etc are removed. Stop words do not impact the search results.

Once the stop words are removed, the search phrase is analyzed to find the exact match of the term entered using the EXACT match process. The content found as exact matches are displayed at the top of the search result.

For example:
Searching for the word 'creating' yields content with the exact word used in the following titles.

• Creating Link
• Creating a Flow
• Creating a Tag

Step 2: Stemming

After analyzing the search term, it is reduced to its root form using a process called Stemming. This is done to ensure the search word returns relevant matches to the content, increasing retrieval accuracy.

Stemming is a technique used to extract the base form of words by removing affixes from them.

For Example:

• creating after stemming is creat
• opportunities after stemming is opportunit

Even though there is no word called creat, it's that part of the word that is found in all variations of the word. Create, creating, created - all have creat in them. Thus someone searching for any form of the word finds all content that uses any variation of the word.

Step 3: Organizing the Results

Once the search phrase is analyzed, it is stored in 3 variations — Shingles, Root Word, and Keyword. They are organized in such a way as to help in displaying the most relevant results first to the user.

The term creating leads in salesforce after stemming becomes creat lead salesforce and is organized as,

Whenever you create content in Whatfix, it is immediately analyzed and organized in our database. This helps in reducing the processing time during the search.

Step 4: Comparison of Search Phrase and your Content

The search phrase and the Whatfix content (Flow, Smart Tip, video, Beacon) are compared and mapped. 

For example:

The word lead is present in content1, content3, content6…
The word creat is present in content2, content3, content9… so on and so forth.

Step 5: Boosting

To decide which content to display first, the contents are scored based on the number of times the analyzed phrase is used in the content and where (position) the analyzed phrase is used.

For example, if the word creat is present in the description of flow1 and the title of flow2. Then, flow2 appears first in the search result.

This is because the title of the content (Flow, Smart Tip, Video, Beacon) has a higher weightage compared to the step title or description.
The weightage is applied in the following hierarchy.

Name of your Flow > Search Keywords > Describe your Flow > Purpose of this step > Describe the step action

Also, the weightage is higher for Shingles, followed by Keyword, and then Raw.

Shingles > Keyword > Raw

This process is called Boosting.

Step 6: Displaying the Result

Once the search phrase is compared with the Whatfix content and scored, the highest-scored content is displayed first.
For example,

The example scenario used above is calculated only using the Title. The result may vary if you factor in the description, step title, and step description.

Languages supported in Self Help Search

The Self Help search supports all the languages that Whatfix supports. 

However, stop words optimization and Stemming processes applies to only the following languages,

• English
• German
• Spanish
• Portuguese-Brazil
• Italian
• French
• Norwegian
• Dutch

History and Suggestions in Self Help Search

• Whatfix retains 5 recently viewed contents in the Self Help widget as the search history. This means that the end user can see the search history and also consume content from this again. 
• The recently viewed content keeps getting updated as the end user consumes the Whatfix content.
• The end users can see the last seen content first in the search history. 
• This is visible to the end users for 90 days or until the browser cache is cleared; whichever happens earlier.

• To yield relevant results on Self Help, Whatfix uses a matching ability that suggests the right results, even when the search term is incomplete.

• Whatfix also suggests the right search results when the search query is misspelled or mistyped. Your end users will see this suggestion only if there is any content with an alternative search term that is mapped to the Self Help.