axa-group/nlp.js
Nlp.js
nlp.js is a JavaScript natural language processing library and development framework used to build natural language understanding engines. It provides a toolkit for creating local machine learning models for intent classification and acts as a multilingual text processor that detects languages and normalizes text across various dialects.
The framework distinguishes itself by supporting local execution on both servers and mobile devices, enabling chatbot functionality without an internet connection. It features a specialized system for conversational slot filling to collect mandatory information and manages stateful conversation contexts to personalize dynamic responses.
The project covers a broad range of NLP capabilities, including named entity recognition for extracting temporal, numerical, and contact data, as well as multilingual sentiment analysis. It also includes utilities for text normalization, such as stemming, tokenization, and spell checking, alongside tools for training language models from JSON or Excel data.
The system can be integrated with HTTP servers, various chat interfaces, and external bot frameworks.
Features
- Intent Classification Pipelines - Implements neural-network based intent classification to identify the purpose of user input.
- Neural Intent Classifiers - Trains neural networks to associate tokenized user input with specific intent labels.
- Chatbot Frameworks - Provides a comprehensive framework for managing conversation context, slot filling, and dynamic response generation.
- Conversation State Management - Maintains stateful conversation contexts to track dialogue progress and manage session variables.
- Template-Based Answer Generation - Produces dynamic responses by replacing placeholders with extracted entities or stored context variables.
- Language Detection Tools - Identifies the language of text and returns a ranked list of possibilities based on an allow-list.
- Lightweight Intent Classifiers - Ships lightweight neural networks for intent classification that run locally on servers or mobile devices.
- Language Model Training - Processes a corpus of utterances and answers from JSON files to train intent recognition models.
- Multilingual Natural Language Understanding - Detects languages and extracts entities across various locales to support global user interactions.
- Multilingual Text Processors - Detects languages and normalizes text across numerous supported languages and dialects.
- Named Entity Recognition - Identifies and extracts structured data like dates, currency, and contact information from unstructured text.
- Natural Language Entity Extraction - Extracts specific entities and metrics from natural language text for structured querying.
- Conversational Answer Generation - Assigns multiple possible responses to specific intents to return varied natural language answers.
- Conversation Slot Filling - Tracks and fills required information slots within a conversation to complete user tasks.
- Natural Language Processing - Tokenizes and stems text in numerous supported languages, including basic tokenization for fantasy languages.
- Text Tokenization - Provides utilities for segmenting raw text into individual words and sentences for downstream linguistic processing.
- Text Normalization - Provides comprehensive text normalization utilities including tokenizers and stemmers to prepare text for natural language understanding.
- Word Stemming - Reduces words to their base or root forms to normalize text across different languages and dialects.
- Local NLP Inference - Allows natural language processing and chatbot functionality to run directly on-device without internet connectivity.
- N-Gram Detection - Identifies the language of a phrase by comparing trigram frequencies against a known training corpus.
- Natural Language Understanding - Provides an engine for classifying user intents and extracting named entities to power conversational interfaces.
- Mobile Inference Deployments - Supports local model training and inference on mobile devices for offline operation.
- Bot Behavioral Pipelines - Scripts the sequence of actions and logic a bot executes in response to specific intents.
- Bot Chat Interfaces - Supports connecting bots to multiple communication channels, including web chats and messaging platforms.
- Model Persistence - Exports a trained neural network to a file format and imports it back to restore the model without retraining.
- Contact Information Extraction - Provides utilities to identify and extract email addresses, phone numbers, and URLs across multiple languages.
- Currency Value Extraction - Identifies and extracts monetary values and their corresponding currency units in multiple languages.
- Network Identifier Extraction - Identifies and extracts IPv4 and IPv6 addresses and hashtags from text across multiple languages.
- Numerical Value Extraction - Identifies and extracts integers, floats, ordinal numbers, and percentages, including text-based numbers.
- Physical Dimension Extraction - Identifies and extracts measurements of length, distance, speed, volume, and area using international acronyms.
- Temporal Information Extraction - Identifies and extracts dates, duration intervals, and ages into standardized time formats.
- Sentiment Analysis Tools - Evaluates the emotional tone of text to determine if input is positive, negative, or neutral.
- Lexicon-Based Sentiment Analyzers - Determines emotional tone across multiple languages using a predefined dictionary of weighted words.
- Spelling Correction - Identifies and corrects spelling errors using a combination of dictionaries and edit distance algorithms.
- Stop Word Filters - Removes high-frequency filler words from text using customizable dictionaries to improve the quality of NLP analysis.
- Local Question Answerers - Processes questions and returns answers using a tab-separated file of question-answer pairs.
- Regex Extraction Utilities - Extracts structured data such as dates and emails from text using regular expressions.
- JSON-Based Persistence - Persists trained machine learning weights to JSON files to eliminate the need for retraining on startup.
- Bot API Server Self-Hosting - Allows hosting the bot on an HTTP server to be accessible to remote web clients.
- Chatbot Integration Frameworks - Provides middleware to connect natural language recognizers to various chatbot frameworks and messaging services.
- Intent-to-Action Handlers - Runs registered functions mapped to specific intents to inject context data or modify answers.
- Intent-Driven Plugin Executors - Triggers custom pipelines or API calls based on recognized user intents to generate responses.
- String Post-Processing Pipelines - Implements pipelines to modify generated answers and update conversation context after processing.
- AI and Machine Learning - NLP library for building bots and sentiment analysis.
- Natural Language Processing - NLP framework for building bots and analysis.
- JavaScript NLP Libraries - Library for building conversational bots and intent parsing.
Star history
How this analysis was created: This summary and feature list were written by an AI model that read the project's README and public documentation pages. Each feature links to the documentation it came from; stars, license and language come straight from the GitHub API. The model does not read the source code, and the analysis is refreshed when the project is re-analysed. Learn more on our About page.
