Skip to content

Releases

papAI 7.4 – Changelog

1. New ML use case : NER & RE (Named Entity Recognition & Relation Extraction)

papAI introduces a new module in ML Lab to automatically structure information from unstructured text, supporting key NLP tasks:

  • NER identifies and classifies entities such as people, locations, dates, monetary values, etc.
  • RE detects semantic relations between entities (e.g., “Person X works for Organization Y”).

NER

RE

1.1. Technical Highlights:

  • Based on SpaCy’s modular NLP engine.
  • Uses Tok2Vec for context-aware token embeddings.
  • Dual-layered architecture:
    • NER layer tags named entities.
    • Relation Extractor identifies meaningful links between entity pairs.
  • Easily customizable: load pretrained models or fine-tune on domain-specific data – all from papAI’s no-code interface.

1.2. Business Use Cases:

  • Resume parsing and automated HR document reading.
  • Legal document structuring (contracts, statutes).
  • Insurance claim processing.
  • Extraction of business data from reports and emails.

2. New Viewer Role for Project

  • A read-only “Viewer” role for project has been added to enable secure access for business profiles.
  • Useful for project stakeholders who only need to explore results and dashboards without editing workflows.

viewer

3. PDP (Partial Dependence Plot) Upgrade

  • Enhanced interpretability for ML models:
    • Combined display of feature distribution and partial dependence.
    • Contextual descriptions show how a change in feature values influences predictions.

pdp

pdp2

4. Frontend Framework Upgrade

  • Upgraded to Angular 19 for better performance, security, and UI extensibility.

5. Minor Improvements

  • Added HTML preview for S3, Azure Blob Storage, Google Cloud storage, and MinIO bucket content.
  • Various stability and UI enhancements.

preview