Content classification helps you automatically organize incoming messages based on the content they contain. When a message matches the keywords in a classifier, Social Media Management applies a classification badge to the message so you can identify, prioritize, and route it more easily.

A content classifier analyzes incoming messages and applies one or more classification badges when its keyword conditions are met.

For example, you might create classifiers to identify:

Spam messages
Hate speech or toxic language
Product feedback
Billing inquiries

You can use classifications in feeds and automation rules to organize, prioritize, and act on content.

How do I create a classifier?

To create a classifier, go to Settings > Classification. From there, you can create your own classifier or enable pre-built classifiers.

What pre-built classifiers are available?

You can enable pre-built classifiers to identify content related to:

Spam
Hate speech and toxicity
Profanity
Personal information

The pre-built classifiers support keywords in English, German, Spanish, French, Italian, and Japanese. The personal information classifier detects messages that contain phone numbers, email addresses, and credit card numbers.

How do I create a custom classifier?

To create your own classifier:

Go to Settings > Classification.
Click Create classifier.
Enter a classifier name.
Optionally, enter a description.
Save the classifier.

The classifier name and description provide context for the Iris keyword assistant when you use it to generate keyword suggestions.

How do I add keywords to a classifier?

To add keywords to a classifier, click Add and choose whether to add keywords from scratch or use a keyword template.

How do I add keywords from scratch?

To add keywords from scratch, type or paste a comma-separated list of words or phrases.

Keyword matching follows these rules:

Keywords can be single words or multi-word phrases.
Keyword matching is not case-sensitive.
Keywords match complete words and phrases only, not partial words or substrings.

You can also enable leetspeak matching. Leetspeak matching helps detect words that have been intentionally altered by:

Normalizing common leetspeak substitutions, such as h4te to hate.
Removing punctuation used to obscure words, such as h.a.t.e.
Collapsing repeated spaces into a single space.

Note: If leetspeak normalization is enabled, incoming text is normalized before keyword matching. For example, leetspeak characters such as 4 are converted to a, 1 to i, 0 to o, and 5 or $ to s.

If a keyword contains numbers or symbols—for example, 4f4 or 18+—it may not match the normalized text because the keyword itself is not normalized in the same way before matching. For example, 4f4 is analyzed as afa, which can result in missed matches.

Consider removing or rewriting keywords that contain numbers or symbols when leetspeak normalization is enabled to avoid missed matches.

Each classifier can contain up to 5,000 keywords.

How does the Iris keyword assistant work?

The Iris keyword assistant can generate related keywords, synonyms, abbreviations, and common misspellings.

To use the Iris keyword assistant:

Type a keyword or sentence.
Review the suggested keywords.
Select the suggestions you want to add to your classifier.

The Iris keyword assistant uses the classifier name and description as additional context to improve its suggestions.

How do keyword groups work?

Keyword groups help you organize keywords within a classifier. You can create groups and assign keywords to them. Groups are unique to each classifier. Empty groups are not saved.

To create a keyword group:

Click the Add group (+) icon.
Enter a name for the group in the New group field.
Click the checkmark (✓) icon to create the group.
Add the keywords you want to include in the group, then click Add in the bottom-right corner.
Your keywords are added to the selected group.
You can filter the keywords by group. To do this, click the funnel icon next to the search bar.

How do I use keyword templates?

Keyword templates let you start with an existing keyword set and then remove keywords or add more keywords as needed.

The Spam, Hate speech and Toxicity, and Profanity templates are supported in English, German, Spanish, French, Italian, and Japanese.

You can also use the Personal information and URL templates to detect messages that contain:

Email addresses
Phone numbers
Credit card numbers
URLs

How do I enable a classifier?

After you save a classifier, enable it so new incoming content can be analyzed and classified.

To enable the classifier, move the toggle switch next to the classifier to the right. The classifier is activated when the toggle is on.

After a message is classified, the classification label cannot be removed from that message. Classification applies to incoming content from the time the classifier is enabled.