Segment Bar User Guide

Copy

• Copy for LLM

  Copy page as Markdown for LLMs
• View as Markdown

  Open this page as Markdown
• Open in ChatGPT

  Get insights from ChatGPT
• Open in Claude

  Get insights from Claude

Overview

The Segment Bar is a powerful filtering and search tool that helps you build, save, and reuse complex queries to find relevant social media posts and news articles in your datasets. Think of it as a smart search bar combined with advanced filters that remembers your work.

What You Can Do

• Search using boolean queries with AND, OR, NOT operators
• Filter by platform, date range, risk signals, and custom classifiers
• Save your searches as reusable "segments"
• Switch between pages while keeping your filters active

Getting Started

Basic Search

The simplest way to use the Segment Bar is to type a search query:

1. Type your search terms in the query field

   • Example of a keyword: climate
   • Example of a phrase: "climate change"
   • Example with operators: "climate change" AND (policy OR regulation)

2. Click "Apply" or press Enter to run the search

3. View results below the filter bar

Search Operators

The query field supports boolean logic:

NOTE: The NOT operator must be used in conjunction with AND to denote the relationship with the preceding term.

Example:

policy AND NOT conspiracy

Pro Tip: Operators must be in UPPERCASE (AND, OR, NOT) to be recognized.

Adding Filters

Beyond text search, you can narrow results using dropdown filters:

Available Filters

1. Date Range

   • Click the calendar icon
   • Choose from quick presets (Last 7 days, Last 30 days, etc.)
   • Or select custom start/end dates
   • Click Apply to activate

2. Platforms

   • Select which social media platforms to include
   • Options: X (Twitter), Facebook, Instagram, Reddit, etc.
   • Multiple selections allowed
   • Shows post counts when available

3. Risk Signals (Proprietary Classifiers)

   • Bot-like: Content from automated or suspicious accounts
   • Toxic: Content flagged for harmful language
   • High-risk: Content meeting multiple risk criteria

4. Additional Filters (if available in your project)

   • Cohorts
   • Entities
   • Engagement ranges
   • Location spoofing indicators
   • Authenticity scores

How Filters Work Together

• All filters use AND logic: posts must match ALL active filters
• Example: Platform = X AND Date Range = Last 7 days AND Risk Signal = Bot-like
• Active filters appear as removable pills below the search bar

Removing Filters

• Individual: Click the × on any filter pill
• All at once: Click "Clear Segment" (see Segments section below)

Field-Specific Search (Advanced)

You can search within specific data fields using the format fieldName:value

Auto-Suggest Fields

When you type in the query editor, field suggestions appear automatically:

1. Start typing a field name (e.g., auth)
2. Dropdown shows matching fields with descriptions
3. Use Arrow keys to navigate suggestions
4. Press Enter or Tab to insert the field name with :
5. Type your search value after the colon

Example Fields:

• author.username:john_doe - Search for posts by specific authors (read-only)
• platform:x - Filter by platform in the query (aggregate)
• author.followers:>1000 - Authors with many followers (filterable)
• lang:en - Filter by language (filterable)

Field Search Syntax

fieldName:searchValue
fieldName:"exact phrase"
fieldName:>100 (numeric comparison)

Combined Example:

(climate OR environment) AND author.displayName:"EPA" AND author.followers:>500

Available Field Types

The system provides access to fields across different entity types. Here are the fields available for filtering:

Post Content Fields (Publicly Filterable)

• text - Main text content of the post (string, filterable)
• title - Post title for Reddit/articles (string, filterable)
• body - Body text, separate from title (string, filterable)
• lang - Detected language code (e.g., "en", "es") (string, filterable)

Author Fields

Publicly Filterable:

• author.displayName - Author's display name (string, filterable)

• author.followers - Follower count (integer, filterable)

• author.following - Following count (integer, filterable)

• author.isBotLike - Bot-like behavior flag (boolean, filterable)

• author.platform - Author's platform (string, filterable)

• author.postCount - Total posts by author (integer, filterable)

• author.profileBio - Author's bio/description (string, filterable)

• author.cohorts - Author classification cohorts (dict_of_double, filterable)

  • Query as: author.cohorts:LabelName>=0.5

Read-Only:

• author.username - Author's handle/username (string, read-only in queries)

Hidden (Not Available in Public Queries):

• author.verified - Verified status (hidden)

Aggregate/ID Fields

• platform - Platform where post originated (string, aggregate)
• authorId - Unique author identifier (string, aggregate)
• postId - Unique post identifier (string, aggregate)
• createdAt - Post creation timestamp (timestamp, aggregate)
• narrativeId - Unique narrative identifier (string, aggregate)

Narrative/Story Fields (Publicly Filterable)

• narrative.title - Narrative title (string, filterable)
• narrative.postCount - Posts in narrative (long, filterable)
• narrative.compassHeadline - AI-generated headline (string, filterable)
• narrative.compassOneSentence - One-sentence summary (string, filterable)
• narrative.compassSummary - Full summary (string, filterable)
• narrative.executiveBriefing - Executive briefing text (string, filterable)
• narrative.totalEngagement - Total engagement in narrative (long, filterable)
• narrative.totalEngagementWithoutNews - Engagement excluding news (long, filterable)
• narrative.weightedRiskScore - Risk score (double, filterable)

Hidden Fields (Internal Use Only)

These fields exist but are NOT available in public queries:

• engagement.total, engagement.likes, engagement.score - All engagement metrics
• emotion.anger, emotion.fear - Emotion scores
• toxicity.toxic - Toxicity score
• isExtended - Extended content flag
• narrative.botPostPercentage - Bot percentage in narrative
• author.verified - Verification status
• cluster - Cluster identifier

Field Access Levels:

• filter (public) - Can be used in queries with : syntax
• aggregate - Can be used for grouping and filtering
• read - Display only, limited query use
• hidden - Not available in public API queries

Important: If you try to use a hidden field in your query, it will not filter results. Only use fields marked as "filter" (public) or "aggregate" in your queries. NOTE: Fields can be exposed for search. We started with the most commonly used fields. If other fields are needed, please notify Product team.

Working with Multi-Line Queries

For complex queries, you can expand the editor to multiple lines:

Auto-Expansion

• The query field automatically grows as you type
• Scrolling enabled if content exceeds max height

Collapse/Expand Toggle

For long queries, a chevron button appears in the bottom-right corner:

• Chevron Down (↓): Collapses query to 2-line preview
• Chevron Up (↑): Expands to show full query

When to Collapse:

• Save vertical space on your screen
• Keep other filters visible while working
• Quick overview of a complex query

When to Expand:

• Edit the query
• Review full logic
• When clicking into the collapsed query (auto-expands)

Segments: Save Your Work

Segments let you save and reuse filter combinations.

What Gets Saved in a Segment

• Search query text
• All active filters (platforms, dates, risk signals, etc.)

Creating a Segment

1. Build your filters using search and/or dropdown filters
2. Click the Save icon (disk icon) in the top-right
3. Name your segment in the dialog
4. Click Save

Result: Your segment appears in the segments dropdown

Loading a Saved Segment

1. Click the Segments dropdown (shows current segment or "No Segment")
2. Select a segment from the list
3. All filters and query automatically apply

Visual Indicator: Selected segment name appears in the dropdown button

Modifying a Segment

When you change any filter or query text while a segment is active:

Dirty State Indicator:

• Query editor border changes from indigo (clean) to teal (modified)
• Only visible when the query field is focused
• Signals "You have unsaved changes"

Two Options:

1. Update Segment (Save icon)

   • Overwrites the existing segment with new values
   • Click the Save icon (becomes enabled when modified)

2. Revert Changes (Counter-clockwise arrow icon)

   • Discards your changes
   • Restores segment to last saved state
   • Instant action, no confirmation

Clearing a Segment

To remove the active segment and start fresh:

1. Click the Clear icon (X icon) in the top-right

2. If you have unsaved changes, a confirmation dialog appears:

   • "Clear Segment?" with warning about discarding changes
   • Click Clear Segment to confirm
   • Or Cancel to keep working

3. If segment is unchanged, clearing is immediate

Result: All filters reset, query clears, no segment selected

Deleting a Segment

To permanently remove a saved segment:

1. Open the Segments dropdown
2. Hover over the segment you want to delete
3. Click the trash icon (appears on hover)
4. Confirm deletion

Warning: This action cannot be undone. The segment is permanently deleted.

Active Filter Pills

Below the search bar, you'll see pills representing active filters.

What Pills Show

• Date Range: "Jan 1, 2024 - Jan 31, 2024"
• Platform: "X", "Facebook" (one pill per platform)
• Risk Signals: "Bot-like", "Toxic", etc.
• Search Query: "Search: your query text"
• Other Filters: Cohorts, entities, engagement ranges, etc.

Managing Pills

• Click the × on any pill to remove that specific filter
• Pills are color-coded by filter type for easy scanning
• No limit to number of active filters

Navigation & Persistence

Cross-Page Filtering

The Segment Bar stays active as you navigate between pages:

• Move from Posts view → Analytics → Maps
• Your filters remain applied
• Same dataset filtered across all views
• Segment selection persists

URL Sync

• Active segments are reflected in the browser URL
• You can bookmark URLs with segments applied
• Page refresh maintains your active segment

Tips & Best Practices

For Better Search Results

1. Start broad, then narrow

   • Begin with general terms
   • Add filters incrementally
   • Save successful combinations as segments

2. Use quotes for exact phrases

   • "climate change" finds the exact phrase
   • climate change finds posts with both words anywhere

3. Combine operators strategically

   ("climate change" OR "global warming") AND policy AND NOT conspiracy

For Segment Management

1. Name segments descriptively

   • Good: "Q4 Climate Policy X Bot Analysis"
   • Less helpful: "Segment 1"

2. Create templates

   • Save base segments for recurring analyses
   • Clone by loading → modifying → saving as new

3. Use the dirty state indicator

   • Teal border = unsaved changes
   • Update immediately or revert if experimenting

For Complex Queries

1. Use multi-line format

   • Break long queries across lines for readability
   • Collapse when not editing to save space

2. Test incrementally

   • Build complex queries piece by piece
   • Verify results at each step

3. Leverage field search

   • More precise than general text search
   • Use auto-suggest to discover available fields

Troubleshooting

"No results found"

Check:

• Are your filters too restrictive? Try removing some
• Is your date range too narrow? Expand it
• Are boolean operators in UPPERCASE? (AND, OR, NOT)
• Do quotes match? Every " needs a closing "

"Segment appears modified immediately after loading"

Possible causes:

• Default filter values differ from segment
• Browser auto-filled a field
• Try clicking Revert to restore exact segment state

"Can't save segment / Save button disabled"

Requirements:

• At least one filter must be active OR
• Query field must have content
• Must have project access/permissions

"Typeahead suggestions not appearing"

Ensure:

• You're typing in the query field (not a dropdown)
• Field names start with lowercase letter
• Wait a moment after typing for suggestions to load

Keyboard Shortcuts

Feature State Reference

Visual Indicators

Button States

FAQs

Q: Can I have multiple segments active at once? A: No, only one segment can be active at a time. However, you can quickly switch between saved segments.

Q: What happens if I close the browser with a segment active? A: If the segment is saved, it will reload when you return (via URL). Unsaved changes are lost.

Q: Can I export my segments? A: Segments are saved to your project and persist across sessions.

Q: Do segments work across different projects? A: No, segments are project-specific. Create new segments for each project.

Q: How many segments can I save? A: There's no hard limit in the interface, but check with your administrator for any project quotas.

Q: Can I rename a segment after saving? A: Not directly in the current version. Best practice: save as a new segment with the desired name, then delete the old one.

Summary

The Segment Bar is your command center for data exploration:

• Search with powerful boolean queries
• Filter with multi-dimensional criteria
• Save successful searches as segments
• Reuse segments across sessions and pages

Master these features to work more efficiently with large datasets and collaborate effectively with your team.