Emotions
Which emotions appear in matching posts and how their shares change over time, on the Emotions tab.
The Emotions tab of the Social Intelligence dashboard counts the emotion labels assigned to posts matching your submitted query, time range, and topic filter. A post can carry several labels. A message expressing both anger and sadness contributes to both counts.
Emotion classification adds detail to a sentiment reading. Two conversations can have similar negative sentiment scores while one contains more expressions of anger and the other more fear. The emotion breakdown shows those differences, and the timeline shows whether a feeling was present throughout the range or became more common during a particular period.
These labels describe the post as a whole. If it mentions several subjects, reading the text helps identify which subject a reaction concerns. The Emotion Classification guide explains what the labels measure and how to interpret them. Getting Started explains the query controls.
Emotion labels
The tab uses seven emotion categories and a Neutral label. The descriptions below give the ordinary meanings of the categories; the assigned labels are the classifier's interpretation of each post.
| Label | Meaning |
|---|---|
| Love | Affection, admiration, or attachment. |
| Joy | Happiness, excitement, amusement, or celebration. |
| Anticipation | Expectation, curiosity, or interest in something approaching. |
| Anger | Frustration, outrage, or hostility. |
| Disgust | Revulsion, contempt, or moral disapproval. |
| Sadness | Grief, disappointment, loss, or regret. |
| Fear | Anxiety, concern, worry, or alarm. |
| Neutral | No other emotion label was assigned. |
Neutral participates in the charts and rankings, including Top emotion. Its presence records a classification result; it does not establish that the author felt nothing. Emotion labels also do not map directly to positive and negative sentiment. Anticipation, for example, can concern a welcome or unwelcome prospect. The Sentiment tab shows the separate positive, neutral, and negative classifications.
Reading counts and percentages
With one query, the tab shows a stats bar, Emotions volume, Emotions distribution, and Emotions over time. The same layout appears when you choose a query through Select, or combine queries through Aggregate.
Emotions volume
Emotions volume shows how many posts carry each label. Each colored bar sits on a gray track representing all matching posts. Hovering over it shows the label's count and the total post count. A post with two labels appears in two bars, so adding the colored counts can produce a total larger than the number of posts.
The Neutral bar gives context to the other counts. If it fills most of its track, most matching posts received no named emotion label. The remaining bars show which emotions were detected in the rest of the posts.
Emotions distribution
Emotions distribution uses those label counts to draw a pie. Each segment is a percentage of all label assignments, including Neutral. This is also the denominator used by Top emotion, Runner-up emotion, and Emotion diversity.
Suppose 100 posts receive 125 labels in total, and 50 posts carry Joy. The Joy volume bar fills half its track because 50 of the 100 posts carry that label. The pie assigns Joy 40%, because its 50 assignments account for 40% of the 125 labels. Both charts describe the same posts with different denominators.
The pie's segments add to 100%. They describe the mix of labels across the result. Since a post can contribute to several segments, the pie does not divide authors or posts into separate groups.
Summary stats
Top emotion and Runner-up emotion identify the two most frequent labels across the selected range, each with its share of all label assignments. A close result means the two labels have similar counts. The posts may express both emotions together or discuss different subjects; the ranking alone leaves that distinction open.
Emotion diversity counts labels with at least 10% of all assignments, including Neutral. One qualifying label is called Concentrated, two or three Moderate, and four or more Diverse. These terms describe how many labels reach the threshold. Several qualifying labels can occur in the same posts, so the count does not identify separate audiences or opposing camps.
Highest concentration gives the date, label, and percentage for the largest share reached by any label within one time interval. That label may differ from the overall Top emotion, and it can be Neutral. A brief interval with very few posts can produce a high percentage, so the Volume view of Emotions over time supplies the counts behind it.
When the series has several intervals, Highest concentration leaves out the final one. The full-range stats and charts retain the final interval, which can explain why the largest share visible at the end of the chart differs from this summary.
Changes over time
Emotions over time groups posts into days or weeks, chosen automatically from the time range. It opens in Volume mode, which stacks the counts for each label within an interval. Each segment counts posts carrying that label, and the stack's total counts all assignments. Posts with several labels therefore contribute more than once to the stack.
Distribution converts each interval's counts into label shares. An interval with assignments adds to 100%, while an empty one has no distribution. This view makes changes in the proportions visible even when the total count varies. A shrinking segment can accompany an unchanged count if other labels become more common; switching back to Volume shows whether its count also fell.
If the last day or week is still underway, its chart values describe the part available so far. The date labels and intervals follow your timezone.
Combining queries with Aggregate
Aggregate combines posts matching any active query and shows the single-query stats and charts. Each matching post is included once, even if several queries match it, while all of its emotion labels remain in the calculation.
This view can bring together a subject's alternative names or related product names. The resulting distribution describes their combined posts; it is recalculated from the label counts rather than an average of the separate query percentages.
Comparing queries
With two or more submitted queries, Compare displays their label distributions together. Each query uses the same time range and topic filter. Select opens one query's full charts and feed.
Comparison stats
Top emotion identifies the label with the largest share within any query and shows that query's name. Runner-up emotion uses the same calculation for a different label. Each label enters the ranking once, represented by the query where its share is highest. The two cells can therefore name the same query or different ones.
Highest variation identifies the label with the largest gap between its highest and lowest shares across queries. The gap is in percentage points. For example, shares of 45% and 20% differ by 25 percentage points, displayed in the cell as 25.0%. This compares the full-range query results; it does not measure change over time.
Most diverse names the query with the largest number of labels reaching the 10% threshold, using the same count as Emotion diversity in the single-query view.
Distribution for each query
The Emotions chart gives each query a bar divided into label shares. Each nonempty bar adds to 100%, with every percentage calculated from that query's total assignments, including Neutral. Emotion colors identify the segments. Hovering shows their percentages.
The equal bar lengths allow comparison of the label mix while hiding differences in total counts. Each query's Select view shows those counts in its Emotions volume chart. Results can overlap: a post matching two queries contributes the same emotion labels to both. Aggregate counts that shared post once when producing its combined breakdown.
One emotion across queries over time
The comparison's Emotions over time chart tracks one selected label across all queries. Its selector initially shows the label named by Top emotion. The stacked segments use query colors to identify which query each count belongs to.
In Volume mode, each segment counts posts carrying the selected label within that query and interval. In Distribution mode, each query's percentage uses the sum of those counts across queries. If a day has 30 Joy-labeled posts in one query and 10 in another, the distribution is 75% and 25%. These percentages describe each query's contribution to the combined Joy count. The Emotions chart above and each query's Select view show Joy's share of all labels within that query.
A shared post contributes to every query it matches in this comparison. Changes in these segments describe changes in query counts; they do not track individual people moving between discussions or changing feelings.
The selector also connects the chart to Highest variation. Choosing the label named by that stat shows when posts carrying it appeared in each query's results. This can help locate a period to examine in each query's Select view, where the full emotion distribution and matching posts are available.
Reading the posts behind a label
The Recent posts feed below the charts shows matching text and emotion badges. You can focus on a particular period by applying a range in the query bar's calendar control. Analyze in Write queries runs the existing queries over that range, updating the charts and feed together. With several queries, Select restricts the feed to the chosen query.
The feed's Filter by sentiment or emotion box can narrow the loaded posts by emotion. Selecting Anger and Sadness shows posts with either label. Adding Negative from the sentiment group also requires those posts to have negative sentiment. The filter leaves the charts unchanged; Getting Started explains the feed controls and loading more posts.
Neutral emotion has no feed filter or badge. The Neutral option under Sentiment selects neutral sentiment, which is a separate classification. Reading posts behind the other labels helps establish what their expressions concern and whether several labels describe the same reaction.