Enable Auto Spell Corrector

Auto Spell Corrector detects and corrects misspelled search queries.

Consider a scenario where a user searches new prodct and gets no results. Then, the user realizes their mistake and changes the query to new product. A happy ending for the user and a lesson for the Auto Spell Corrector which maps the misspelled new prodct to new product. After this, it automatically corrects each instance of new product in subsequent sessions and returns up to five relevant results.

For new prodct > new product deduction to work, the correctly spelled new product must be somewhere in the training data. You can feed training data by specifying the content source, object, and content fields. If the training data consists of salesforce > case > product name, then Auto Spell Corrector throws suggestions whenever a user misspells a product name.

Auto Spell Checker keeps expanding and updating the dictionary based on the user search activity. The dictionary is common among search clients. It means that if a suggestion is offered on one search client, then it will be offered on another search client.

  • Auto Spell Corrector doesn't support a standard dictionary (yet).

  • If the search queries differ a lot of one another, then Auto Spell Corrector does not recognize them as similar queries. For example, if a user looks up keyboard after misspelled moniter, then Auto Spell Corrector doesn't consider keyboard to be the correct spelling of moniter.

Enabling Auto Spell Corrector

  1. From SearchUnify admin panel, go to Search Tuning.

  2. Under the Auto section, select the Search Client where auto spell checker is to be applied.

  3. Toggle Enable Auto Spell Corrector on to enable the feature.

  4. Click to launch a window where you can select content sources.
  5. Select the content source, content type, and content fields and click Add. In the example, Case Types are going to be auto-corrected. Make sure you select the content source with the most accurate and relevant data. Auto Spell Corrector is trained on the selected content source data.

  6. Click after you are done.