Classify by Response

The CSV file should contain each canned response, and its associated category. For example, if you prepare the CSV using Microsoft Excel (and save the file in CSV format), the first column contains the category name, and the next column contains the canned response:

If you use the Find Patterns feature to identify canned responses embedded in content items, you can export the patterns in CSV format, remove any patterns that do not correspond to canned responses, and specify categories for each canned response. For more details, see Finding patterns.

You can choose to create separate text files for categories/canned responses. The category name must appear as the first line of each file, followed by the canned response associated with the category. All of these files must be placed in a single directory.

If there is more than one canned response associated with a particular category, save each canned response as a separate text file. Use the same category name as the first line of each text file.

Classify by Response compares canned responses (in CSV or text files) with all content items in the active view. When a match is found, Classify by Response updates the category field of the content set item accordingly. Classify by Response either replaces or adds the appropriate category name to the content item's category field, and (optionally) removes the canned response found in the body of the content item.

The following Classify by Response options are available:

• Classify by sentences
• Remove response from item
• Overwrite existing categories
• Similarity percentage
• Ignore characters
• Search range

When the Classify by sentences option is selected, the system attempts to match entire sentences in canned responses to content items.

If you clear the Classify by sentences check box, the system compares canned responses with content items word-by-word. Note that this is a time-consuming process.

When the Remove response from item option is selected and a canned response is found in a content item, the system automatically removes the canned response from the content item. When this option is cleared, the canned responses are left in the content items.

When the Overwrite existing categories option is selected and a canned response is found in a content item, the system replaces the value in the category field with the canned response's category (as defined in the CSV or text file). When this option is cleared, the canned response's category is added to the existing value of the content item's category field.

In some cases, the canned responses embedded in content items might have been edited by agents. Set the Similarity percentage to determine how close the match should be between the embedded canned responses, and the original canned responses (in CSV or text file format). By default, the Similarity percentage is set to 10%. In general, enter higher values for the Similarity percentage if the canned responses were rarely edited by the agent sending them.

The system ignores characters entered in the Ignore characters box when comparing canned responses to content items. Ignoring non-essential characters yields more accurate comparison results. By default, the system ignores the "greater than" symbol (">"), often found in email replies.

When Beginning only is selected, the system identifies and matches canned responses only when they appear at the beginning of content items. This option (selected by default) saves time and resources when running Classify by Response on a large content set. When canned responses appear consistently at the beginning of content items, selecting Beginning only can improve the accuracy of your results.

When Entire field is selected, the system searches for canned responses in the entire text field. This option yields more accurate results if canned responses do not always appear at the beginning of content items. For example, your content set texts might contain multiple correspondences, where canned responses are embedded more deeply within the content items.