Glossary
All the key terms of AI, document processing, and automation, simply explained.
47 terms
A
AI Act
European Union legislation regulating the use of artificial intelligence, particularly regarding safety risks and fundamental rights. The AI Act aims to ensure ethical and secure AI deployment across the EU.
AI Agent
Autonomous computer program capable of perceiving information from its environment, analyzing it to make decisions, and executing actions independently to achieve predefined objectives.
API
Application Programming Interface. A set of rules and protocols enabling different software to communicate, facilitating system integration and automated data exchange between platforms.
Apprentissage Non Supervisé
Machine learning method where models are trained without labeled answers to identify natural patterns or groupings within data, such as segmenting customers into similar groups.
Apprentissage Supervisé
Machine learning method that trains a model with pre-labeled data. Each training example is associated with a correct answer, enabling the model to understand the relationship between inputs and outputs.
C
Case Management
Coordinated and personalized management of a file or complex situation, aiming to assess needs, plan and coordinate services, monitor implementation, and adjust support for effective and continuous care.
Classification Automatique des Documents
Process by which IDP identifies and categorizes documents based on their content. Incoming documents are quickly sorted according to predefined rules.
Collecte des Données
Extracting information from unstructured documents and converting them into a metadata flow. Transforms emails and contracts into organized data.
Compréhension Intelligente du Layout
Technology for analyzing the geometric layout of information in a document. Identifies columns, rows, and fields to process semi-structured documents.
CRM
Customer Relationship Management. System centralizing client information (Salesforce, HubSpot, Zoho CRM). IDP automatically enriches client records with data extracted from various documents.
Croisement des Données
Linking information extracted from different documents to create meaningful connections. Ensures consistency between orders, invoices, and delivery notes.
D
Data Warehouse
Centralized data storage system used for analysis and reporting. Consolidates departmental data for strategic decision-making.
Deep Learning
Advanced artificial intelligence technique using neural networks composed of multiple layers. Applied to IDP for image recognition and natural language processing.
Déploiement Cloud et On-Premises
Two deployment options for IDP solutions. Cloud offers accessibility and scalability; on-premises suits organizations requiring strict data security and regulatory compliance.
Données Complexes
Information with difficult structure, including nested tables, graphs, or text with strong contextual dependencies. Financial reports with annotations exemplify this category.
Données Génériques
Information that can be easily standardized and processed without particular complexity. Names, addresses, and phone numbers follow predefined formats.
E
Entraînement des Modèles
Process improving AI and IDP performance in recognizing and extracting document data. Models learn through thousands of examples to identify key elements accurately.
ERP
Enterprise Resource Planning. System centralizing company activities (finances, HR, supply chain, production). IDP integrates with ERPs like SAP, Oracle, and Microsoft Dynamics to automate document processing.
Extraction de Données
Process by which IDP identifies and extracts relevant information from a document, such as invoice amounts, dates, and supplier details.
Extranet
Extension of a company's internal network, allowing external partners or clients to access certain information securely.
F
Fusion de Documents
Capability to combine multiple files into one while integrating relevant metadata, enabling centralized management of related documents.
G
GED
Electronic Document Management. Centralized system for storing, organizing, and managing an organization's electronic documents.
H
Hallucinations
Instances where an artificial intelligence model generates incorrect or fabricated information that appears plausible. Particularly problematic in precision-critical fields like medical or legal.
I
IA Générative
Subcategory of artificial intelligence that uses algorithms to create new content: text generation, image creation, and automated creative tasks.
IA (Intelligence Artificielle)
Technology that enables machines to simulate human capabilities such as learning, reasoning, and problem-solving.
IA Souveraine
Approach to artificial intelligence that ensures data is hosted and processed locally, without relying on foreign technologies. Essential for regulated industries.
IA Supervisée
Type of Machine Learning where the model is trained from labeled data including inputs and expected results.
ICR
Intelligent Character Recognition. Extension of OCR that reads handwriting and converts it into digital format.
IDP
Intelligent Document Processing. Technology that automates the management of unstructured or semi-structured documents using AI, ML, NLP, and OCR.
Indexation
Using metadata extracted from documents to organize them and facilitate their retrieval.
L
LLM
Large Language Models. AI models capable of processing large amounts of text and extracting relevant information, enabling advanced contextual understanding.
M
Machine Learning
Technique of artificial intelligence that enables systems to learn from data without being explicitly programmed. In IDP, it continuously improves extraction accuracy.
Masquage des Données Sensibles
Feature enabling IDP to automatically detect sensitive information and mask it, ensuring GDPR compliance when sharing documents.
Métadonnées
Information describing the content of a document: author, creation date, document type, subject. They improve document organization and searchability.
N
NLP
Natural Language Processing. Technology enabling computer systems to understand and interpret human language, used in IDP for entity extraction and contextual understanding.
Normalisation des Données
Converting extracted data into consistent, standardized formats (e.g., date standardization). Ensures comparability across sources and compatibility with ERP and CRM systems.
O
OCR
Optical Character Recognition. Converts printed or written text into digital format for automated document processing.
P
POC
Proof of Concept. Demonstration validating an idea's feasibility before full implementation, such as testing IDP solutions on specific document types.
Progiciel
Set of standardized commercial programs designed for use by multiple companies (like ERPs), as opposed to custom-built software.
PSSI
Information Systems Security Policy. Organizational rules protecting information systems, controlling access to sensitive data, and preventing cyber threats.
R
RAG
Retrieval-Augmented Generation. AI technique combining information retrieval with text generation to produce contextual, reliable responses.
Renommage Automatique
Feature extracting document data to automatically standardize file naming, improving organization and traceability.
RGPD
General Data Protection Regulation. European regulation protecting the personal data of EU citizens. Organizations must obtain explicit consent and allow deletion requests.
RPA
Robotic Process Automation. Technology automating repetitive manual tasks by mimicking human computer actions, such as data transfer between systems.
S
Segmentation des Documents
IDP functionality dividing complex documents into distinct sections for individual processing. For instance, a multi-invoice PDF can be split into separate files per invoice.
T
Taux d'Intervention Manuelle
Measurement indicating IDP's reliance on human verification to correct or validate extracted data. A low rate demonstrates sufficient accuracy for handling complex documents without operator involvement.
V
Vision-Language Model (VLM)
Multimodal AI model that understands and links visual and textual information to perform tasks including image description and text-based image retrieval.