Categories are used to group interactions with similar characteristics so that SpeechMiner users can find interactions that require their attention for particular reasons. The interaction characteristics that define a Category can be as simple as "interactions in which the Topic of interest rates was found," but they can also be very complex, defining a very specific mix of characteristics. For example, a Category can be defined to group calls in which the Topic Payments was mentioned by the agent, the customer displayed signs of agitation, and the call-center supervisor was called in to deal with the customer during the last two minutes of the call.
SpeechMiner assigns interactions to Categories after it has finished processing them. All the Categories in the system are global Categories that can be applied to any interaction. When interaction processing is completed for an interaction, SpeechMiner goes one by one through the list of Categories in the system and checks whether the interaction meets the conditions of the Category. If it does, the interaction is assigned to the Category. A single interaction may not belong to any Categories, but it can also belong to many Categories.
Category assignment is intended to help reviewers find interactions with particular characteristics so that they can analyze them and compare them. For example, reviewers may wish to compare how different agents handled customer agitation during a particular sales campaign. It also enables SpeechMiner to perform statistical analysis on interactions. For example, SpeechMiner can check the percentage of calls in September in which customers exhibited agitation.
Four types of categories exist in SpeechMiner:
- Regular Categories: Categories that have conditions defined in the form of a logical expression. When SpeechMiner processes an interaction, the interaction is automatically assigned to this type of category if it matches the requirements that are defined in the condition. These categories can be independent (on the top level of the category hierarchy) or they can be sub-categories of other, parent categories.
- Parent Categories: Categories that contain other categories. These categories have a name and description, but do not include a logical expression. Interactions are assigned to them when they are assigned to any of their sub-categories. Any category can be converted into a parent category by adding sub-categories to it. When a regular category is converted into a parent category, the logical expression that defined the category is deleted.
- Subcategories: A category can contain one or more subcategories in a multi-tiered category structure. The parent-category is defined as a container of its subcategory. When SpeechMiner identifies an interaction as belonging to a subcategory, it automatically identifies it as belonging to the parent-category as well. For example, a 5-60 seconds Interaction Duration category contains three subcategories 10 seconds category, 20 seconds category and 50 seconds category. Any interaction associated with the 10 Minute category is also associated with the Hour Interaction Duration category. In this case, you receive data about all the interactions that were between 5 to 60 seconds and specific data about interactions that were specifically 10, 20 and 50 seconds long.
- Manual Categories: Categories that include a collection of interactions that do not fit into a predefined category. This type of category is referred to as a Manual category. A Manual category has no criteria. Instead, it is a category to which you can manually add any and as many interactions as you like without preconditions. For example, a Manual category could be used as placeholder for interactions the SpeechMiner user would like to review during routine weekly staff meetings.