The default value shown alongside selected language features (like topics, words, emotions, or grammar elements) is the relative difference.
The relative difference is a method used to compare how frequently a particular language feature appears in one group of data compared to another. It takes into account that the data sets being compared may have different sizes, so the measurement is adjusted. This ensures a fair comparison between the two data sets.
However, you can change the metrics displayed, based on your preferences. In the Explore view, or on the insight card itself, you can apply a different metric from our available options, such as word frequency or percentage, and verbatim frequency or percentage.
Explore
Navigate to Explore view to see the output of your analysis.
In the menu bar, click on Metrics.
Select the desired option and your results will automatically update with the new metrics.
Insight cards
Navigate to the insights view to see a collection of all cards created for your project.
In the section above the linguistic features, click on the 'Insight preferences' icon to see the list of available metrics.
Select the desired option and your insight card will automatically update with the new metrics.
The overall impact metric (showing the overall impact from aggregated items on an insight card) at the top of an insight card will now align with the metric of your choice.
The metrics explained
Relative difference — a measure of how much more prevalent a linguistic feature is in one data set compared to others. It is normalized to take into account variations in the size of the data sets being compared. For example, when looking at how the people of Luxembourg talk about chocolate compared to England, you have to take into account the size of each population to ensure a fair comparison.
Word frequency — a measure of many times a specific word appears in the dataset. This metric is very useful when you need to know the exact number of times a particular word or phrase is used.
Word percentage — the number of times a word appears in your data set, divided by the total words in that group. This metric is particularly useful for social listening when you need to know the proportion of times that a particular word or phrase is used — especially when looking at different sample sizes.
Verbatim frequency — the number of separate text blocks that contain a specific language feature in them. If one of those appears more than once in a single text block, it's still counted as one. This metric is useful for surveys as it can tell you, for example, the exact number of individuals who have given a certain rating or used a particular word or phrase.
Verbatim percentage — a measure of the proportion of text blocks that contain a specific language feature. Similar to verbatim count, if the item appears multiple times within a single text block, it is still counted as one. By using percentages instead of counts, you can determine the prevalence of the item across different sets of data. As above, this metric is useful for surveys as it can tell you the proportion of individuals who have given a certain rating or used a particular word or phrase.