Skip to content
English
Contact us
  • There are no suggestions because the search field is empty.

Main Concepts : Categories, Topics & Sentiment

Caplena allows you to structure and analyze open-ended text feedback using a few key concepts. Understanding these will help you build meaningful insights from your data.

Category

A category is a group of related topics that share the same overarching theme.
For example: Customer Service, Product Quality, or Pricing.

AI-Relevant: Categories are considered by the AI during training and when assigning topics.

Topic

Topics (also referred to as codes, themes, classes, or labels) summarize what is being mentioned in your text responses.

Each topic captures a specific idea or theme, for example:

  • Staff friendliness

  • Shipping fees

  • App usability

Topics are what Caplena uses to classify and analyze responses.

 Topic Label

The topic label is the concise name you give to a topic. It’s what’s displayed on topic chips and used by the AI to assign topics to texts.

AI-Relevant: The topic label directly impacts how the AI recognizes and applies the topic. Choose it thoughtfully.

Topic Sentiment

Topic sentiment adds emotional context to topics, classifying feedback as positive, neutral, or negative.

For instance, in a customer survey:

  • Topic: Customer Service

  • Sentiment: Positive (e.g., “super helpful”) or Negative (e.g., “no one answered”)

This allows you to analyze not just what’s being talked about, but how people feel about it.

You only need one topic (e.g., Customer Service) — the AI will automatically assign the appropriate sentiment.


Topic sentiment is optional,  you can use topics without sentiment if you prefer.

Enabling Topic Sentiment

You can enable sentiment only if the topic has no reviewed rows yet.

Once enabled:

  • The AI will start differentiating between sentiment variations.

  • You’ll see sentiment-specific versions of the topic (e.g., Helpful staff vs. Unhelpful staff).

Want to convert existing non-sentiment topics into sentiment topics? That’s possible,  even if rows have been reviewed. [See tutorial video below.]

Disabling Topic Sentiment

If you disable topic sentiment:

  • All sentiment labels will be removed from the topic.

  • This change is not reversible.

    See tutorial video below.

⚙️ Topic Properties Overview

Depending on whether sentiment is enabled, topics will have slightly different properties.

 

 Sentiment Labels

Optional labels that replace the standard topic name when sentiment is enabled.
Example: A topic called Friendliness could have sentiment labels:

  • Friendly (positive)

  • Neutral

  • Unfriendly (negative)

AI-Relevant: Sentiment labels guide the AI when applying sentiment

Code

A code is a unique numerical identifier assigned to each topic (and sentiment version, if applicable). You can define the code manually, or let Caplena assign one automatically.

 Description

A longer internal note or explanation of what the topic covers.

Not AI-relevant — ensure that the essence of your topic is already reflected in the label and category.

 Keywords 

By default, Caplena’s AI does not use keywords — it learns from examples rather than rules. This allows for more flexible and robust learning.

However, keywords can be enabled for special use cases. When enabled:

  • They act as exact matches.

  • Use them sparingly and specifically.