Create a Report Using Multi-Label Classification
min read - Updated on Nov 15, 2026You create a Report with multi-label classification by leaving Enable Multi-label Classification on in the Structure step of the Report wizard. With it on, one piece of feedback gets every label it mentions, and each label keeps its own sentiment, so a review that praises the price and criticizes the delivery counts toward both themes with the right opinion for each. This article follows the wizard from your data to the finished Report and shows where the multi-label results appear.
The switch is on when the step opens, so every Report is multi-label unless you turn it off. Multi-label classification is available in every package. To decide which structure suits your data, read Multi-label vs. Single-label Classification.
Sign in to Kimola and open Reports in the main menu. Have the data you want to analyze ready: a file, one or more links, or a connected integration.
How Multi-Label Classification Works
With multi-label classification, Kimola reads each piece of feedback and gives it every label of the classifier that the text matches. The feedback isn't split into separate records: it stays one record that belongs to several themes. Kimola then reads the sentiment of each theme separately, so the opinion given about one aspect doesn't blur into the opinion given about another. This is aspect-based sentiment analysis, a core feature that is available in every package, in the app and in the API.
A hotel guest writes: "The pool was spotless and the staff were kind, but dinner took an hour and the room smelled of smoke." With multi-label classification, the review gets Cleanliness and Staff Friendliness with a positive sentiment, and Service Wait Times and Room Condition with a negative one.
Multi-label classification doesn't mean that every record gets several labels. Feedback that talks about one topic gets one label, and feedback that mentions four topics gets four.
Step 1: Start the Report and Add Your Data
Open Reports in the main menu. The creation area at the top of the page gives you three ways to bring in feedback, and all three reach the same wizard, with the same Structure step.
Upload a file. Drag a file into the upload area or click it to choose one. Kimola accepts one .xls, .xlsx, .csv or .tsv file of up to 30 MB. On Sheets & Columns, choose the Text Column, which is required, and the Url Column and Date Column when your file has them. See Create a Report by Uploading a Custom Dataset.
Add links. Paste a link into the box at the top of the page and click Start. To analyze several sources in one Report, add each link on its own line. See Create a Report by Adding Links.
Use an integration. Click Intercom, Zendesk or Google Sheets in the Integrations row. Integrations are available in the paid packages, not in Starter. See Create a Report Using Integrations.
[Screenshot: The Reports page with the upload area, the link box and the Integrations row]
For the platforms that links support, see Supported Platforms for Creating Reports from Links.
Step 2: Choose the Classifiers
The Classifiers step lists the models that label every row of your data. Multi-label classification works with all of them, so you choose the classifiers for the topics you want to see, and the structure you pick in the next step applies to each one.
The Automatic Classifier discovers the themes from your data, with no labels to define. The Sentiment Classifier labels the sentiment of each row. You can also add other pretrained models, or your own custom models in the paid packages. Choose a content classifier and a sentiment classifier together to see how customers feel about each theme. The best and worst performing themes of the Report need both.
[Screenshot: The Classifiers step with the Automatic Classifier and the Sentiment Classifier selected]
Step 3: Choose the Structure
The Structure step comes after Classifiers and asks you to choose how feedback will be labeled in your classifiers. It has one switch, Enable Multi-label Classification. Leave it on to assign several labels when one piece of feedback mentions different aspects, which gives a more complete view of strengths and weaknesses.
Below the switch, an Example shows how this works on a review of a cordless drill. The review praises the power and the ease of use, and criticizes the battery and the build quality. With multi-label classification on, the labels Power & Performance, Ease of Use, Battery Life and Build Quality all apply to it. Turn the switch off, and the review is grouped under one main theme instead.
[Screenshot: The Structure step with Enable Multi-label Classification turned on and the example review]
Turn the switch off when each piece of feedback is about one subject, such as short answers to a focused survey question, or when every record must count toward exactly one theme. Multi-label vs. Single-label Classification compares the two structures and explains how each changes the numbers.
A Report's structure can't be changed after the Report is created. To read the same data the other way, create a new Report from it with the other structure.
Step 4: Choose Dimensions, Review and Create the Report
The Dimensions step comes next. Dimensions are optional research questions that Kimola asks of every piece of feedback, grouped as Drivers, Issues, Actors, Context and Comparisons. Choose the ones you want, or skip the step. See Add Dimensions to a Report.
The last step, Review, shows your choices: an optional Report Title, the Report Language, your data source and the Required Query. The Report Language is the language the Report is written in, and your feedback can be in any of the 30+ languages Kimola analyzes. See Language Options in Reports. Confirm to start the analysis. When Kimola shows "Here we go!", the Report is ready, and View Report opens it.
Queries are counted as rows × models × Dimensions. The structure doesn't change the count, whether a piece of feedback gets one label or several. See How to Calculate Queries.
Read the Multi-Label Results
The Report opens on Overview in Reports. To see multi-label classification record by record, open Content in the Report menu. Under each record, Sentiment shows the sentiment of the record as a whole, and Themes lists every theme the record got. Each theme has its own color: green for a positive sentiment, red for a negative one and grey for a theme without a sentiment. A record that mentions both a strength and a weakness shows both, side by side.
[Screenshot: A record on the Content page with several themes, each shown in the color of its own sentiment]
To read everything customers say about one aspect, choose the theme under Themes in Classifications. Because one record counts toward every theme it got, a record appears under each of its themes, and the shares of the themes can add up to more than 100%. See Filter Records in Reports and Content in Reports.
The Satisfaction Score and Dissatisfaction Score of each theme are on Performance in Reports. They are adjusted for the size of each theme, so a small theme and a large one compare fairly.
You can also cross themes with other columns, classifiers and Dimensions in Tables in Reports.