Top-Down Modeling Repositories

World Modelers repositories related to top-down modeling/analysis engines and the aspects of the human-machine interface supporting this functional area

Engine Software Components

Delphi

University of Arizona’s top-down modeling engine.

• Repository: See https://ml4ai.github.io/delphi/index.html for more information.
• License: Apache, see https://ml4ai.github.io/delphi/index.html#license-and-funding
• To run Delphi independently as a standalone engine without the HMI, here are the installation/configuration instructions: https://worldmodelers.com/td-modeling/delphi.html#running-locally-without-the-hmi
• To run Delphi with the Causemos HMI, here are the installation/configuration instructions: https://docs.google.com/presentation/d/19PeNAoCIxNCQxXAZNV4Gn4kMVPCAJKYlLXQCJ6V2SZM/edit#slide=id.g8c51928f8f_0_200

DySE

University of Pittsburgh’s top-down modeling engine.

• Repository: See https://github.com/pitt-miskov-zivanov-lab/dyse_wm for more information.
• License: MIT, see https://github.com/pitt-miskov-zivanov-lab/dyse_wm/blob/main/LICENSE.txt
• To run DySE independently as a standalone engine without the HMI, here are the installation/configuration instructions: https://github.com/pitt-miskov-zivanov-lab/dyse_wm/blob/main/README.md
• To run DySE with the Causemos HMI, here are the installation/configuration instructions: PLACEHOLDER, include all repositories needed

Sensei

Jataware’s top-down modeling engine.

• Repository: See https://github.com/dojo-modeling/sensei for more information.
• License: MIT, see https://github.com/dojo-modeling/sensei/blob/main/LICENSE
• Running Sensei independently as a standalone engine and with the Causemos HMI requires the same setup steps, which are available here.

HMI Software Components

Causemos is the main user-interface in the WorldModelers program, used to synthesize knowledge into runnable graphs, and dervie additional insights using these engines.

API

The API and data-exchange format can be found in

• https://github.com/uncharted-causemos/docs/tree/master/td-models

Causemos

Causemos is split into the following modules to support top-down modeling:

• HMI: https://github.com/uncharted-causemos/causemos
• Causemos data service: https://github.com/uncharted-causemos/wm-go
• Causemos recommendation service: https://github.com/uncharted-causemos/wm-curation-recommendation
• Unstructured extraction pipeline: https://github.com/uncharted-causemos/anansi
• Structured data pipeline: https://github.com/uncharted-causemos/slow-tortoise
• License:

Installation

Causemos requires:

• Elasticsearch with compatible dataset ingested (See Corpus Ingestion and Assembly)
• At least one working inference engine, either external or hosted locally