You are here:
Topic Classification
Use topic classification in Data 360 to organize unstructured text such as reviews, case comments, or survey responses, into meaningful categories. To evaluate records for classification, the model applies a trained language model to a selected field. For each record, the model assigns a topic label and a confidence score. Data 360 adds these outputs to your dataset. Apply topic classification in reporting and queries.
Note
All topic classification models and algorithms use the new runtime.
Review these topic classification goals.
- Classify text into categories—for example, billing, product issues, or feature requests.
- Standardize how text data is organized across datasets.
- Identify trends and patterns by topic.
- Segment data based on topics for targeted engagement.
- Enrich datasets with topic labels and confidence scores.
Use Cases
Forecasting supports various use cases.
| TYPE | USE CASE | WHAT DOES IT SOLVE |
| Marketing | Segment Audiences | To create more targeted campaigns, group customers by interests or topics. |
| Analyze Campaigns | To understand what messaging resonates, classify responses and engagement by topic. | |
| Strategize Content | Identify themes across emails, web, and social channels. | |
| Service | Track Issues | To speed up resolution, automatically categorize support cases, such as billing or technical issues. |
| Route Cases | To speed up resolution, automatically categorize support cases, such as billing or technical issues. | |
| Manage Escalations | To handle cases faster, flag sensitive or high-priority topics, such as complaints or disruptions. | |
| Sales | Analyze Deal Insights | To identify key topics, such as pricing, objections, or competitors, analyze emails, and call transcripts. |
| Route Leads | To assign leads to the right team, classify inbound inquiries by product, industry, or use case. | |
| Upsell and Cross-Sell | Surface high-intent topics, such as demo requests or pricing discussions. |
Considerations
Before you apply a topic classification model, review these considerations.
- Overlapping or ambiguous categories reduce accuracy.
- Very short or unclear text is harder to classify.
- Model performance depends on the quality and relevance of input data.
- The model is multilingual and supports text in multiple languages.
Limitations
Before you use topic classification, review these limitations.
- Define clear and distinct topic categories for better accuracy.
- Use text data that reflects actual customer language.
- Review and refine categories periodically as business needs evolve.
- Combine topic classification with sentiment analysis for richer insights.
- Chunk the long text before you pass it through the model. Topic classification works best on short, intent-rich text up to 512 tokens
Did this article solve your issue?
Let us know so we can improve!
