ThemysThemys/Docs

Extraction View

How to read and navigate the full extraction breakdown of a Terms of Service document.

What the Extraction Shows

The Extraction view takes the raw Terms of Service text and organizes it into structured, readable sections. Instead of scrolling through dense legal paragraphs, you get a clean breakdown with plain-English summaries.

Each section in the extraction represents a logical topic from the original document: data handling, billing terms, content ownership, and so on.

Section Types

The extraction groups clauses into these standard categories:

SectionWhat It Covers ------------------------- Data PrivacyData collection, storage, sharing, cookies, third-party access, GDPR/CCPA mentions ArbitrationMandatory arbitration, class action waivers, dispute resolution, governing law LiabilityLiability caps, disclaimers, indemnification, warranty limitations Content & IPUser content ownership, license grants, DMCA, intellectual property rights BillingPricing, auto-renewal, refunds, cancellation, price changes AccountRegistration, account termination, suspension, data deletion GeneralMiscellaneous clauses that don't fit the above categories

Not every document will have all section types. A simple privacy policy might only contain Data Privacy and General sections.

Key Points

Each section includes a list of key points: concise bullet summaries that capture the important details. These are written in plain English, not legalese.

Example:

code
## Data Privacy

  • The service collects your name, email, IP address, and device information
    • 

  • Data may be shared with third-party advertisers for targeted advertising
    • 

  • You can request data deletion by emailing privacy@example.com
    • 

  • The service uses cookies for analytics and session management
    • 

  • Data is stored on servers in the United States
    •

Key points highlight what matters. If a section is routine (e.g., standard cookie disclosure), it may have only one or two points. Sections with aggressive terms will have more detailed breakdowns.

Severity Ratings

Within each section, individual key points can be tagged with a severity rating:

RatingColorMeaning ------------------------ InfoBlueStandard or expected clause. Good to know. CautionAmberNotable clause that could affect you. Read carefully. WarningRedAggressive or one-sided clause. Significant concern.

Not every key point has a severity tag. Routine items are presented without a badge. The tag appears inline next to the point text.

Quoted Text

When the extraction references a specific clause, it can include the exact text from the original Terms of Service. This appears as a quoted block below the key point.

Use quoted text to:

  • Verify that the extraction accurately captured the original language
  • Understand the exact wording if you need to reference it
  • Make your own judgment about what the clause means

Navigating the Extraction

Section Headers

Each section header is clickable. Click to expand or collapse the section's key points.

Expand/Collapse All

At the top of the extraction view, there's an Expand All / Collapse All toggle:

  • Expand All opens every section so you can scroll through the full breakdown
  • Collapse All collapses every section to just the headers for a quick overview

Search

Use your browser's built-in search (Ctrl+F / Cmd+F) to find specific terms within the extraction. Common searches:

  • "arbitration" for dispute resolution terms
  • "terminate" for account termination clauses
  • "third party" for data sharing terms
  • "refund" for cancellation and refund policies

Extraction vs. Original Text

The extraction is a summary, not a copy. It distills the important points from the original text but may not capture every detail. Use it as:

  • A fast way to understand what the Terms say
  • A map to identify which sections you should read in full
  • A reference for the most common concerns

If you need the exact legal language, always refer back to the quoted text or the original document.

Source Quality

The quality of the extraction depends on the source text:

SourceExpected Quality ------------------------- Clean HTML page (most ToS pages)High: full text extraction, accurate sections JavaScript-rendered pageMedium: may miss dynamically loaded content

If the extraction looks incomplete, try providing the URL directly instead of pasting text. Themys uses Jina Reader API to render JavaScript and extract clean text, with a direct HTML fetch fallback.

Example Extraction

Here's what a typical extraction looks like for a streaming service's Terms of Service:

code
## Data Privacy [Warning]

  • Collects viewing history, search queries, device info, and location data
    • 

  • Shares viewing data with content partners for royalty calculations
    • 

  • Uses cookies and tracking pixels for ad targeting on third-party sites
    • 

  • Data retained for 2 years after account deletion
    • 

  • No option to download your data (GDPR right to portability not honored)
    • 



Content & IP [Caution]


  • You retain ownership of reviews and ratings you post
    • 

  • Service gets a perpetual, worldwide license to display your reviews
    • 

  • You cannot record, copy, or redistribute streamed content
    • 

  • Violation of DRM restrictions may result in account termination without refund
    • 



Billing [Info]


  • Monthly auto-renewal with email reminder 7 days before charge
    • 

  • Price changes take effect next billing cycle
    • 

  • Cancel anytime, access continues until end of paid period
    • 

  • No refunds for partial months
    • 



Account [Caution]


  • Account may be suspended after 6 months of inactivity
    • 

  • Service can terminate accounts that violate terms without warning
    • 

  • Upon termination, all saved preferences and watch history are deleted
    • 

  • You cannot create a new account after termination for the same email
    •

Each section gives you a clear picture of what the Terms say without needing to parse the legal language yourself.

When Extraction Quality Is Low

Some documents produce poor extractions. Common causes and solutions:

ProblemCauseSolution -------------------------- Missing sectionsJavaScript content not renderedUse the URL input so Jina Reader can render the page Very few key pointsDocument is unusually short or vagueThe ToS may genuinely be minimal, check the original Duplicate sectionsDocument has repeated contentThe LLM may see repetition as separate sections, ignore duplicates

If none of these solutions work, the document may not be suitable for automated extraction. Read the original text directly.

API Access

The extraction data is available programmatically through the API:

bash
curl -X POST https://themys.ca/api/analyze \

  -H "Content-Type: application/json" \


  -H "Authorization: Bearer YOUR_TOKEN" \


  -d '{"text": "Terms of Service content here..."}'

The response includes the fullExtraction array with the same section structure shown in the web UI. See POST /api/analyze for full documentation.

Was this page helpful?