Skip to main content
Enterprise

Core asset: PDF template fingerprint library

ChromaParse's core value isn't OCR or Excel export — it's this **continuously evolving chromatography report parsing fingerprint library**. Every deployment builds on prior project experience; the more it's used, the more accurate it gets.

Current page: Core technology

What is a "PDF template fingerprint library"?

It's a curated collection of rules, features, and parsing strategies that recognize different chromatography report layouts.

Even within Waters Empower, customer variations are huge: Empower 2 / 3 / FR versions; field order differences in report templates; peak table column names that may be RT, Retention Time, or 保留时间; area fields that may be Area, Peak Area, or 面积; PDFs that may be text-based or scanned; tables that may span pages, repeat headers, or merge cells.

So a fingerprint library is more than regexes — it's a seven-layer knowledge architecture.

The core deliverable of an on-premise deployment is exactly this: a knowledge base and parsing engine that precisely recognizes your PDFs. Once the library is adapted, parsing aligns with your LIMS fields, validation records and audit trails are in place — the system becomes a stable part of your QC process from day one.
Seven-layer fingerprint architecture
  • 📄
    Document fingerprint L1
    Identify which instrument, report type, and template version
  • 📐
    Layout fingerprint L2
    Position features of trace, peak table, sample info, and SST regions
  • 🏷️
    Header fingerprint L3
    Column names, aliases, units, cross-page headers, merged-header rules
  • 🔗
    Field mapping L4
    Customer PDF fields → standard fields (RT, Area, Height, %Area)
  • ⚙️
    Parsing strategy L5
    Text-layer extraction, OCR, table reconstruction, coordinate location, numerical validation
  • Validation rules L6
    Decimals, thousand separators, scientific notation, required fields, peak number continuity
  • 🔍
    Source-trace info L7
    PDF source coordinates, page, and text-box position for every extracted field

Public library vs private library

The fingerprint library is two-tiered: public is out-of-the-box; private is customer-specific.

Public fingerprint library

Maintained by ChromaParse, covering major instruments and common templates. Available to both SaaS and on-premise customers.

  • Waters Empower common report formats
  • Agilent ChemStation / OpenLab common formats
  • Thermo Chromeleon common formats
  • Shimadzu LabSolutions common formats
  • Continuously updated — new template versions ship with product upgrades
  • Included in all on-premise plans

Customer private library

Built for a specific pharma's internal report templates; deployed only in that customer's environment. Adapts to private formats while protecting data and business rules.

  • Built from real customer PDF samples
  • Deployed only inside the customer environment or tenant
  • Includes template version management and change history
  • Continuous adaptation when customer report templates change
  • 3–5 templates in Pro plan; unlimited in Enterprise
  • Improves with use — gets more precise the more it's used
IP boundary commitment: Customer PDF source files, field configuration, and private template rules are customer data — not shared. Generalized parsing capability may be abstracted into the public engine, but customer-specific fingerprint libraries are deployed only in the customer's environment.

Template state management

Parsing capability binds to template fingerprints. When customers modify report templates, re-validation or re-adaptation is required. The product surfaces real-time validation state for every template.

Certified

Validated against samples; parsing is stable and reliable for daily processing

Pending

New or recently changed template; needs sample-based validation

Needs adapt

Fingerprint doesn't match current rules; accuracy not guaranteed; professional adaptation required

Deprecated

Template no longer in active use; historical reports still parse correctly

Each template includes: template name, applicable instrument/system, applicable report type, fingerprint version, validation sample count, last update time, status, output field mapping, change history. The state and change history themselves serve as audit-ready evidence.