universal-history 0.1.2

A library for creating, maintaining, and utilizing Universal Histories across multiple domains

Universal History

A library for creating, maintaining, and utilizing Universal Histories across multiple domains.

Overview

The Universal History library provides a standardized framework for recording, analyzing, and retrieving historical information about any subject across various domains. Whether you're tracking educational progress, medical history, employment records, or any other domain-specific information, this library offers a unified approach to managing that data.

Key Features

• Multi-domain support: Track histories across education, health, work, finance, and other domains
• Event recording: Capture individual events with rich context and metadata
• Trajectory synthesis: Generate summaries and analyses from sequences of events
• State document: Maintain an up-to-date representation of the subject's current state
• Domain catalogs: Define and standardize domain-specific terminology and metrics
• LLM integration: Optimize historical data for use as context with language models
• Multiple storage backends: Store data in memory, files, or MongoDB

Installation

pip install universal-history

For MongoDB support:

pip install universal-history[mongodb]

For LLM integration:

pip install universal-history[llm]

Quick Start

from universal_history import UniversalHistoryClient

# Initialize a client with file storage
client = UniversalHistoryClient(storage_dir="./data")

# Create an event
event = client.create_event(
    subject_id="student123",
    domain_type="education",
    event_type="exam",
    content="Student scored 92/100 on Algebra final exam.",
    source_type="institution",
    source_id="school001",
    source_name="Springfield High School",
    metrics={"score": 92, "max_score": 100},
    tags=["math", "algebra", "exam"]
)

# Update the state document with information from events
client.update_state_from_events(
    subject_id="student123",
    domain_type="education"
)

# Get LLM-optimized context
context = client.get_llm_context("student123")
print(context)

Architecture

The Universal History system consists of several key components:

• Event Record (RE): The basic unit of information, capturing specific events or observations
• Trajectory Synthesis (ST): Analysis and summary of multiple events over a period of time
• State Document (DE): Real-time representation of the current state of a subject
• Domain Catalog (CDD): Semantic framework for interpreting information in a specific domain
• Universal History (HU): Container for all components related to a specific subject

Documentation

For detailed documentation, visit https://universal-history.readthedocs.io/

License

This project is licensed under the MIT License - see the LICENSE file for details.