facebook/duckling
Duckling
Duckling is a deterministic named entity recognizer and natural language entity extractor. It transforms unstructured text into machine-readable data by mapping language-specific input strings to universal structured formats.
The system utilizes a rule-based engine and composable language rules to resolve entities without relying on probabilistic models. It supports multilingual text parsing across various regional locales, employing a logic-driven approach to normalize diverse natural language expressions into standardized numeric values.
The project covers the extraction and normalization of dates, times, durations, distances, monetary values, temperatures, and quantities. It also recognizes contact information such as email addresses, phone numbers, URLs, and credit card numbers.
Users can extend the system by defining custom dimensions and extraction rules to identify domain-specific information.
Features
- Entity and Relation Extraction - Provides a deterministic system for identifying and extracting named entities from unstructured text.
- Natural Language Entity Extraction - Provides a deterministic engine for extracting structured entities like dates, numbers, and currencies from natural language text.
- Custom Entity Recognizers - Allows for the definition of domain-specific entity detection rules via composable language rules and configurations.
- Multilingual Text Processors - Supports extraction of structured information across various regional locales and different languages.
- Named Entity Recognition - Identifies and classifies entities in unstructured text using predefined linguistic patterns and regular expressions.
- Natural Language Date Parsing - Parses complex natural language date and time expressions into structured machine-readable formats.
- Currency Value Extraction - Identifies currency symbols and spelled-out money amounts to extract structured financial values from text.
- Numerical Value Extraction - Converts written numerals, fractions, and quantities of weight or volume into standardized numeric values.
- Natural Language Value Normalization - Normalizes diverse natural language expressions of time, distance, and currency into standardized numeric formats.
- Rule-Based Text Parsers - Employs rule-based text parsers to tokenize and tag text based on predefined linguistic grammars.
- Text-to-Numeric Normalizers - Converts diverse natural language expressions of time and measurement into standardized numeric values.
- Document and Unstructured Extraction - Parses unstructured natural language strings into machine-readable formats.
- Date and Time Libraries - Parses natural language expressions of dates, times, and holidays into structured formats.
- Deterministic Linguistic Resolution - Employs a logic-driven, rule-based engine to resolve complex entities without relying on probabilistic models.
- Deterministic Entity Parsers - Provides a logic-driven parser that transforms natural language expressions into standardized units of measurement, currency, and time.
- Locale Resolution - Resolves the active language and locale to map input strings to universal structured formats.
- Context-Aware State Engines - Maintains internal variables and state to drive context-dependent behavior when parsing textual environments.
- Locale Mapping Rules - Maps language-specific strings to universal structured formats using locale-specific rule sets.
- Rule-Based Pattern Matching - Compares input strings against predefined linguistic and regular expression patterns to identify entities.
- Parsing Contexts - Uses runtime parsing contexts and state tracking to resolve ambiguous natural language expressions.
- Custom Entity Dimensions - Allows users to update rules and classifiers to add support for new entity types and locales.
- Contact Information Extraction - Recognizes and structures common patterns for email addresses, phone numbers, URLs, and credit card numbers.
- User-Defined Entities - Enables the creation of user-defined data types and extraction rules for domain-specific information.
- Customizable Extraction Rules - Provides customizable extraction rules using regex patterns and production functions for new locales.
- Physical Distance and Duration Extraction - Provides the ability to translate natural language descriptions of time spans and physical lengths into standardized units.
- Extraction Dimension Modules - Implements a modular architecture that organizes extraction logic into independent dimensions to resolve complex entities.
- Extraction Logic Compositions - Organizes extraction rules into independent modular dimensions that combine to resolve complex entities.
- Natural Language Expression Decomposition - Breaks down complex phrases into smaller components to resolve nested entities like durations and dates.
- Natural Language Temperature Extraction - Extracts temperature readings and intervals expressed in various scales and natural language formats.
Star history
How this analysis was created: This summary and feature list are AI-generated from collected project material and can contain mistakes. Stars, license and language are imported from GitHub. Inclusion does not mean that we have tested or audited this project. Check the source documentation for any feature you depend on. Learn more on our About page.
