Distributed Deductive System Sorts (DDSS)

DDSS is a distributed deductive system with a scalable architecture. It currently supports distributed engines including forward-chaining, E-graph, and more.

Design Philosophy

DDSS adopts a modular architecture that decomposes the deductive system into independent but collaborative sub-systems:

1. Separation of Concerns: Each module focuses on a specific reasoning task
2. Concurrent Execution: All modules collaborate asynchronously through a shared database, fully utilizing system resources
3. Persistent Storage: Uses a database to store facts and ideas, ensuring data consistency

The system uses a database as the central hub, with two tables (facts and ideas) for interaction between sub-systems:

• Eager engines (e.g., forward-chaining): Read facts and eagerly produce new facts. They also add ideas to broadcast "I want this XXX" - indicating what new facts they need to produce more results.

• Lazy engines (e.g., E-graph): Could produce too many facts if eager, so they quietly accept facts without producing many. They only produce facts when they see ideas from other engines that they can (partially) fulfill.

Modules

• Input (ddss/input.py): Interactive input interface with BNF syntax parsing
• Output (ddss/output.py): Real-time display of facts and ideas from the database
• Load (ddss/load.py): Batch import of facts from standard input
• Dump (ddss/dump.py): Export all facts and ideas to output
• DS (ddss/ds.py): Forward-chaining deductive search engine
• Egg (ddss/egg.py): E-graph based equality reasoning engine

Installation

Using uvx (Recommended)

The simplest way is to run with uvx:

uvx ddss

This automatically installs all dependencies and starts the DDSS system.

Using pip

pip install ddss
ddss

Usage

Basic Usage

Run DDSS with a temporary SQLite database:

ddss

Specifying a Database

DDSS supports multiple database backends using the -a or --addr option:

# SQLite (persistent)
ddss --addr sqlite:///path/to/database.db

# MySQL
ddss --addr mysql://user:password@host:port/database

# MariaDB
ddss --addr mariadb://user:password@host:port/database

# PostgreSQL
ddss --addr postgresql://user:password@host:port/database

Selecting Components

By default, DDSS runs with all interactive components (input, output, ds, egg). You can select specific components using the -c or --component option:

# Run only input and output (no inference engines)
ddss --component input output

# Run with only the forward-chaining engine
ddss --component input output ds

# Run with only the E-graph engine
ddss --component input output egg

Available components:

• input: Interactive input interface
• output: Real-time display of facts and ideas
• ds: Forward-chaining deductive search engine
• egg: E-graph based equality reasoning engine
• load: Batch import facts from standard input
• dump: Export all facts and ideas to output

Interactive Usage

After starting, input facts and rules at the input: prompt. The syntax follows the format premise => conclusion:

Example 1: Simple Inference

Input a fact stating a is true:

input: => a

Input a rule stating if a then b:

input: a => b

The system automatically derives and displays => b:

fact: => b

Example 2: Equality Reasoning

Input an equality relation a == b:

input: => a == b

Input an idea for b == a by creating a rule that requires it:

input: b == a => target

The system will derive both the idea and facts:

idea: => b == a
fact: => b == a
fact: => target

License

This project is licensed under the GNU Affero General Public License v3.0 or later. See LICENSE.md for details.

Links

• GitHub: https://github.com/USTC-KnowledgeComputingLab/ddss