Explore

The four tabs grouped under Explore in the sidebar each rank one type of entity by how many matching posts it appears in. Where the Measure tabs describe the conversation in aggregate, these tabs name what circulates inside it: the hashtags posts label themselves with, the ticker symbols they tag, the sites they link out to, and the accounts they mention. All four tabs share the same two elements, a visual card for relative prominence and a ranked table with exact counts, and differ only in the entity type they rank. The feed beside the results shows the posts behind the rankings, labeled with their sentiment.

Every entry on these tabs is also a control. Clicking an entry in the visual card, or choosing an action from a table row's menu, writes that entry into the query bar instead of recomputing anything on the page: the edited query is highlighted until you run Analyze, so a refinement can be adjusted by hand, combined with further edits, or undone before it is committed. Analysis on these tabs therefore tends to proceed in rounds: run a broad query, check which entities lead the rankings, append the most telling one, and run Analyze to narrow the conversation to the posts that carry it. Each round re-ranks these tabs around the narrower slice while the Measure tabs re-read its sentiment, volume, and composition, and two or three rounds move the analysis from a broad subject to one specific thread inside it. A query for electric vehicles may show #tesla leading the Hashtags tab; appending it narrows the analysis to that thread, and the Mentions tab then shows who that narrower conversation addresses.

Each tab adapts to the number of queries in the query bar the same way the Overview and Measure tabs do. With a single query, it shows the layout described below. With two or more queries, a toggle appears above the results with three views: Compare sets the queries against each other, Aggregate combines them into one conversation (a post matching several queries is counted once) and renders the single-query layout over the result, and Select renders the single-query layout for one query chosen from a dropdown. The query bar itself, including operators, time ranges, and display names, is covered in Getting Started.

The four tabs

Hashtags

Entries are the hashtags matching posts carry, such as #ai or #breakingnews. Hashtags are the conversation's self-labeling: authors choose them to join a campaign, address a community, or attach a post to an event, so the ranking shows which of those threads the conversation organizes itself into. Appending one narrows the analysis to posts that have labeled themselves deliberately, which often makes hashtags the cleanest of the four entity types to refine a query with.

Cashtags

Entries are ticker symbols such as $AAPL, which posts use to tag stocks and crypto tokens. The ranking shows which assets the conversation ties the subject to, and its density is a reading of its own: outside market-centric subjects cashtags are sparse, so a short list with low counts usually means the conversation is not financial, while a dense one marks a market-watching audience. Appending a cashtag isolates the trading thread around the subject; adding one as a new query sets the subject against the asset's own conversation.

Links

Entries are the sites matching posts link out to, one entry per domain, such as youtube.com or theguardian.com. The ranking is the conversation's source mix, showing whether it runs on news coverage, video, newsletters, or campaign pages; a news-fed conversation and a meme-fed one with the same sentiment score are different situations. An unfamiliar domain ranking unusually high deserves a check: append it and read the feed to see what is being passed around, since concentrated linking to one obscure site can mark automated or coordinated posting rather than organic discussion.

Mentions

Entries are the accounts referenced in posts, shown as handles such as @nytimes.com. The ranking names who the conversation addresses: the people, organizations, and outlets being answered, credited, or blamed. Appending an account isolates the posts aimed at it, and adding one as a new query sizes and characterizes the conversation around that account in its own right.

Single query (Select)

The layout with one query in the query bar, also rendered by the Select view for whichever query you choose when several are active.

At a glance

Each tab opens with a card named after it, such as Hashtags at a glance, presenting the ranked entities visually. A toggle switches between two renderings of the same ranking:

• Word Cloud sizes each entry by frequency, making the hierarchy readable in one look: what dominates, what trails, and whether the leader stands alone or merely heads a pack. Colors in this view are decorative, and size carries the signal.
• Treemap gives every entry a tile with area proportional to its count, and pointing at a tile shows the exact count. It is the better view for judging near-ties, since similar areas are easier to compare than similar font sizes, and for reading the ranking as proportions of a whole.

Clicking an entry in either view appends it to the query in the query bar. The analysis does not rerun on its own, so adjust the query if needed and run it to narrow the results to posts that also contain that entry.

Top entities

Below the glance card, a table named Top hashtags, Top cashtags, Top links, or Top mentions ranks the leading entities with exact counts. Total is the number of matching posts the entity appears in; a post can carry several entities, so the totals are not shares of a whole and can add up to more than the conversation's size. Where the cloud and treemap give the ranking its shape, the table gives it numbers: exact counts, exact ranks, and the tail entries the visual views compress.

Each row's menu holds three actions and is the tab's main instrument for iterative work:

• Append to query adds the entry to the current query, the same narrowing as clicking it in the glance card: after the next run, every tab reads only the posts that carry the entry.
• Add as new query keeps the current query as it is and adds the entry as a query of its own, turning a discovery into a comparison: the subject on one side, the entity's whole conversation, including posts unrelated to the subject, on the other.
• Append & add as new query adds a new query that combines the current query with the entry, leaving the original untouched. Running it compares the conversation against its own slice, showing how sentiment, volume, or anything else inside the entry's thread differs from the subject at large.

In the Select view, the same menu applies its actions to the query you are viewing.

Aggregate

Aggregate renders the same two cards over the combined conversation: every post matching any of the queries, with posts matching several counted once. It produces one entity ranking for the family of queries, the right reading when the queries are spellings or facets of one subject, and it fills out rankings that are sparse for any single query. Row actions and glance card clicks append to all queries at once here, since a refinement found in the combined conversation applies to every query in it.

Compare

In the Compare view, the two cards contrast the queries' vocabularies, each query keeping the color and display name it has in the query bar.

Word cloud

One cloud combines the most frequent entries from every query, each colored by the query it belongs to. Color makes shared and distinctive vocabulary visible: entries in a single color name what that conversation circulates on its own, and a term that appears for more than one query marks common ground between them. The treemap rendering is not available in the comparison; the table below carries the exact counts. Clicking an entry appends it to all queries at once, which keeps the comparison aligned on the refined scope; run the updated queries to apply it.

Top entities

The table merges the queries' rankings into one list ordered by combined post count, with one count column per query in its color and display name. A dash means the entity does not appear in that query's ranking, and the pattern of dashes is the comparison's first finding: rows with counts across every column mark shared vocabulary, while a row high for one query and dashed for the others marks vocabulary distinctive to one conversation, often the sharpest available characterization of how the conversations differ. To read one query's full ranking on its own, switch to it in the Select view.

Two controls refine the table. The query selector chooses which queries appear as columns, useful when several active queries make the table wide; the ranking recomputes over the queries shown. The All / Shared toggle filters the rows: All lists every entity that ranks for any shown query, while Shared keeps those that rank for at least two, isolating the common ground. Each row offers a single Append to all queries action, which adds the entity to every query in the bar so the comparison stays aligned after the refinement; run Analyze to apply it.