Documentation for PACTE and PSC Services Stack

What is this?

This is the documentation for a collection of services meant to facilitate the usage of automatic annotation tools for the processing of text documents in the context of the PACTE project developed at CRIM.


PACTE is a collaborative annotation web platform for text content that integrates an array of practical tools for research groups. It offers three annotation modes: manual, semi-automatic and automatic. Manual annotation is carried out by means of an interface optimized to allow rapid entry of the data enriching a text. The automatic mode is composed of all the specialized and configurable annotation services (named entities, disambiguated terminology, etc.). Semi-automatic annotation, using active learning algorithms, allows training of a prediction model with minimal annotation, requiring less effort to annotate large text corpora. See below for more information on these tools.

The collaborative nature of the PACTE web platform allows sharing of analyses and annotations with other researchers, thereby facilitating cooperation and opening the door to large-scale multi-partner studies. PACTE generates important gains in productivity by significantly reducing the time spent on analysis, while improving consistency.

PACTE uses a number of shared components or modules (labeled as PSC) for schema, corpus, project and user management, as well as shared components for annotation and file storage. Shared components were designed to be easily integrated to third party annotation platforms such as VESTA.

Software Architecture

This project encompasses a number of elements which interact together. See diagram below for a general idea of the intended software architecture.



For further information please contact:

For technical support please contact:


This section is split into shared component services (PSC), automatic text annotation services (PACTE) and infrastructure support services (IS). For automatic video and audio annotation services please refer to the VESTA documentation.