Word matching for standard topics

Rules and word groups use lemmatized and exact word matching options to match specific versions of a word for topics.

Restriction:

This feature is inactive by default. Ask your Medallia expert to file a Feature activation request with Medallia Technical Support.

Please include the following information with the request:

What type of data does the instance have?
How many records with comments does the instance have in the last 1.5 years? If this is a new instance, please share expected volume for 1 year or 2 years.

When configuring rules and word groups, you define how Text Analytics matches keyword text to the text in comments. This enables you to match specific versions of a word instead of all variants of a word, thereby avoiding cluttered topic matches.

For each keyword, select from the following word matching options, defined below:

Lemmatized matches (default)
Exact matches - non-case-sensitive
Exact matches - case-sensitive

Lemmatized matches (default)

Text Analytics matches the base form of the word. Text Analytics looks for conjugated and pluralized forms of the keywords automatically. Keywords are usually nouns, verbs, adjectives, or adverbs that convey meaning. For example, a rule with a lemmatized match for the keyword tower would tag comments with the words tower, towers, towering, and so on. This is the default matching option for all keywords.

The following list provides additional information about how Text Analytics processes keywords:

Text Analytics picks up both singular and plural forms of the word. So, if you enter the word product, Text Analytics picks up the word products as well as product. However, if you enter the plural form of the word (products), Text Analytics does not pick up the singular form of the word (product).
If you enter the word help, Text Analytics picks up help, helps, helped, and helping. Text Analytics does not, however, pick up helper, helpful, or unhelpful because these words are distinct from the term help.
Add words ending in -ing separately from the root forms of the word, since these words can be nouns, verbs, or modifiers.

Note: Stop words are not valid for lemmatized matches, and specifying a stop word usually returns no results. To use a stop word in a topic, use exact match options, or consider adding it as a user feature.

A stop word is one that typically serves a grammatical function but has little meaning on its own. Common stop words are helper verbs such as is, prepositions such as on or with, and articles such as a or th.

Exact matches

You can define whether exact matches are case sensitive.

Exact matches - non-case-sensitive

Text Analytics matches the word exactly as entered, without regard to letter case, and with no variants of the base word. For example, a rule with a non-case sensitive exact match for the word tower would tag comments with the words tower, Tower, TOWER, and so on.

Exact matches - case-sensitive

Text Analytics matches the word exactly as entered, including letter case, and with no variants of the base word. For example, a rule with a case-sensitive exact match for the word Tower would tag comments with the word Tower.

If you include symbols in exact match keywords, those symbols must be matched in comment text for the comment to be tagged. For example, a rule with an exact match for the keyword $10 tags comments with $10, but does not tag comments with 10.

Spaces are not considered in exact match keywords, and are converted to NEAR operators. For example, if you create a rule with an exact match for the words company hours, the rule is saved as company NEAR hours, with each word needing to appear within one word of the other word.

The NEAR operator finds two or more words within a certain window of a single sentence, and works slightly differently for lemmatized matches and exact matches. For lemmatized matches, the NEAR operator counts the maximum number of words after the first word. For example, if your lemmatized match rule includes three words, you must specify that they occur within at least two words for all words to me matched. For exact matches, multiple words in the rule can exist within the same window of words. For example, you can create an exact match rule with three words that have a maximum WITHIN value of one word, as shown in the image below.

Important: For exact-match rules, always click

information to see whether Topics screen requires different syntax. If the Topics screen suggests an alternate syntax, click the new syntax in the tip to apply that syntax to your rule.

Syntax tip for an exact-match rule

Exact matches and stop words

Although stop words are typically ignored by the parser, they can be specified in topic rules using exact matches. However, when using exact matches to specify stop words in topic rules, comments with compound stop words may not appear in the preview panel despite matching the rule. This omission is limited to the preview panel.

For example, if a rule includes the word because as an exact match, a comment containing the phrase because of will not appear in the preview panel because the preview match algorithm considers that entire phrase to be a stop word, and the space is converted to an underscore. As a result, the preview does not match the rule's because to the comment's because_of.

The matching algorithm used for published topics is different from the one used for previews, and it allows exact matching for because and of independently. As a result, the record will be tagged as expected by the topic rule when the topic is published.

Rule component and match type compatibility

This table summarizes components that can and cannot be used in rules with the different word matching options.


	Lemmatized Matches (default)	Exact Matches — Non-case sensitive	Exact Matches — Case sensitive
Lower case characters
Upper case characters
Numbers	Although numbers cannot be added as keywords, numbers can be used in user features, as long as the user feature name does not start with a number.
Spaces
Underscore
Hyphen
Other punctuation marks
Stop words