Topics

The Topics tab measures what the conversation is about, classifying posts into eight broad subject areas. The topic mix frames every other measurement, since a negative score carried by News coverage and one carried by Daily Life chatter describe different situations with different audiences. As with emotions, a post can fall under several topics at once, so counts measure how much of the conversation touches each subject area.

The page adapts to the number of queries in the query bar. With a single query it shows the layout described in the next section. When two or more queries are active, a toggle appears above the results with three views: Compare sets the queries against each other, Aggregate combines them into one conversation, and Select focuses on one of them at a time. For the query bar itself, including syntax, time ranges, and display names, see Getting Started.

Single query (Select)

The layout for one query, and what the Select view renders for the query you pick.

Stats bar

Top Topic is the most common subject category, with its share of all topics detected. It names the arena the conversation is happening in, and it doubles as a check on the query itself: a brand query expected to surface Business discussion but led by Culture or Daily Life may be matching a different audience, or a different meaning of the term, than intended. When the arena is not where the subject should live, the fix is usually a sharper query, such as an exclusion or an exact phrase, rather than a different conclusion.

Runner-Up Topic is the second most common category, with its share. The pairing of the top two characterizes the conversation more precisely than either alone: Technology with Business reads as an industry story, while Technology with News suggests the subject has broken into general current events, and the distance between the two shares shows whether the conversation lives in one arena or straddles both.

Topic Diversity counts how many categories hold a meaningful share of detections, at least 10%: Concentrated when one does, Moderate for two or three, Diverse for four or more. A concentrated conversation can be read as one conversation. A diverse one is better understood as several audiences discussing different aspects of the subject, which argues for segmenting before generalizing, either with topic-specific queries or by reading the feed per topic badge.

Highest Concentration is the completed interval where a single topic reached its highest share of the conversation, with the date, the share, and the topic. It marks the day the conversation was most single-subject, which usually corresponds to one story dominating. Narrowing the time range to that date shows what the story was.

Topics volume

One bar per topic category shows how many posts touch each subject, and pointing at a bar shows the exact count. The muted track behind the bars represents all matching posts, so each bar reads as that topic's reach within the conversation. Watch the Misc bar alongside the named categories: a large Misc share means much of the matched conversation fits no clear subject, which can be normal for casual chatter but may also flag a query that is sweeping in noise.

Topics distribution

The pie chart shows the same classification as proportional shares of all topic labels detected, the quickest read of the conversation's subject mix. Use it for the proportion judgments the bars make you compute, such as whether news-type coverage outweighs personal conversation overall, which sets expectations for tone and audience before you read the other tabs.

Topics over time

Stacked bars track the subject mix across the range, in two views:

• Volume stacks the absolute counts per interval, showing when each arena talks. Surges come colored by topic, so this view identifies what kind of attention arrived: a spike that is mostly News is a different event than one that is mostly Culture, even at identical size.
• Distribution normalizes each interval to 100%, showing the mix shift independent of volume. A topic's band widening over time means the nature of the conversation is changing even if its size is not, such as a product conversation gradually escaping Technology into Daily Life as it reaches ordinary users.

Aggregate

Aggregate computes one subject-matter profile for the combined conversation, useful when the queries name one subject and the question is which arenas the family as a whole occupies.

Compare

The Compare view contrasts the arenas the subjects occupy, each in its query bar color.

Stats bar

Top Topic names the strongest topical concentration anywhere in the comparison: the topic with the largest share within any single query, with that share and the query it belongs to. It points to the most arena-defined conversation in the set, which is often the right place to start reading.

Runner-Up Topic is the second-ranked topic by the same measure, with its query. Together with the top cell it shows whether the comparison's strongest concentrations sit in one query or are spread across them.

Highest Variation names the topic whose share differs most between queries, with the gap. This is the arena where the compared conversations diverge most, and therefore the most informative segment to inspect in the Topics chart below: one subject debated as business while the other is debated as culture explains a lot of downstream differences in tone.

Most Diverse names the query with the most topics above the 10% threshold, with the count. The most topically spread conversation reaches the widest range of audiences; set against a rival confined to one arena, it frames a mainstream-versus-niche contrast that volume alone does not capture.

Topics

One bar per query, stacked to 100% and segmented by topic category, making subject mixes comparable regardless of volume differences. It shows whether the compared subjects occupy the same arena at all: two competitors may both lean negative while one is debated as business news and the other as a cultural argument, which changes what their scores mean and who is doing the talking. When the topic mixes differ sharply, the sentiment and emotion comparisons are comparing different kinds of conversation, and their gaps should be read with that in mind. The Highest Variation cell names the segment to look at first.

Topics over time

This chart inverts the single-query version: instead of all topics for one conversation, it tracks one topic across all of them. A selector chooses the topic, starting on the most prevalent one in the results, and the queries stack within each interval. Volume shows how much each conversation contributes to that arena over time; Distribution shows each query's share of it per interval, where a crossover means ownership of the arena changed hands. Walking through topics with the selector shows, arena by arena, who owns the conversation and when, at a resolution the static mix chart cannot provide.