In Computer Science, **Ontology** is a frame of reference for systems development and data modeling in a specific knowledge domain. It describes: - Relevant entities in the domain - Their properties - The relationships between those Entities in such a way that humans *and* machine can share knowledge *across* diverse subdomains and that diverse databases are interoperable. ![[Pasted image 20250116085636.png]] There are many ontological frameworks available, such as [[OWL 2 Web Ontology Language Mapping to RDF Graphs (Second Edition)]] and [[ISO 21838-2 Basic Formal Ontology.pdf]]. These frameworks specify how to design a proper ontology and the machine language to be used when formalizing an ontology. From a technical standpoint, the **Ontology for Decision-Making under Uncertainty** here presented is a draft design for an application ontology, since it represents an individual view in a specialized domain, whereas formal ontologies must result from consensus within a group, as explained in the 2009 paper by *Guarino, Staab and Oberle* ( [[Guarino et al. What is an Ontology.pdf]]: > Computational ontologies are a means to formally model the structure of a system, i.e., the relevant entities and relations that emerge from its observation, and which are useful to our purposes. An example of such a system can be a company with all its employees and their interrelationships. The ontology engineer analyzes relevant entities3 and organizes them into concepts and relations, being represented, respectively, by unary and binary predicates. The backbone of an ontology consists of a generalization/specialization hierarchy of concepts, i.e., a taxonomy. Supposing we are interested in aspects related to human resources, then Person, Manager, and Researcher might be relevant concepts, where the first is a superconcept of the latter two. Cooperates-with can be considered a relevant relation holding between persons. A concrete person working in a company would then be an instance of its corresponding concept. In 1993, Gruber originally defined the notion of an ontology as an “explicit specification of a conceptualization”. In 1997, Borst defined an ontology as a “formal specification of a shared conceptualization”. This definition additionally required that the conceptualization should express a shared view between several parties, a consensus rather than an individual view. Also, such conceptualization should be expressed in a (formal) machine readable format. In 1998, Studer et al. merged these two definitions stating that: “An ontology is a formal, explicit specification of a shared conceptualization.” --- From a pragmatical AI standpoint, an ontology provides a standard canvas for capturing a specific domain knowledge in interoperable graph format, such as: > [!NOTE] Example > An **Ontology** for medical knowledge provides the canvas for > - A Knowledge graph for diabetes > - A Knowledge graph for hepatitis > > A LLM (Large Language Model) can be trained with the KGs to support specialized diagnosis. This shared knowledge also can be used to structure databases that are easier and faster to query. In the example above, the ontology guides the development of the *schemas* (see [[Fonseca et al.-Ontology vs Schema.pdf]]) of the knowledge graphs. There have been a few efforts to develop ontologies and schemas for Risk Management and Performance Management, but the practice never became mainstream. A 2018 paper, *The Common Ontology of Value and Risk* ([[Guarino et al.-Common-Ontology-Risk.pdf]]) studied the need to combine the concepts of value (i. e., performance) and risk, but the proposed models were kept separated. > Risk analysis is traditionally accepted as a complex and critical activity in various contexts, such as strategic planning and software development. Given its complexity, several modeling approaches have been proposed to help analysts in representing and analyzing risks. Naturally, having a clear understanding of the nature of risk is fundamental for such an activity. Yet, risk is still a heavily overloaded and conceptually unclear notion, despite the wide number of efforts to properly character- ize it, including a series of international standards. In this paper, we address this issue by means of an in-depth ontological analysis of the notion of risk. In particular, this analysis shows a surprising and important result, namely, that the notion of risk is irreducibly intertwined with the notion of value and, more specifically, that risk assessment is a particular case of value ascription. As a result, we propose a concrete artifact, namely, the Common Ontology of Value and Risk, which we employ to harmonize different conceptions of risk existing in the literature. ![[Pasted image 20250116093047.png | 600]] [[Open Risk - Risk Model Ontology]] is focused in financial services and deals only with risk analysis in that industry. [[MONITOR_Base_Ontology_Report_1_0.pdf]] is oriented towards disaster management -- an even more specialized domain. [[Freitas Jr et al. Ontology_for_Performance_Measurement_Indicators.pdf]] proposed a model centered on the entity 'indicator'. ![[Indicator Graph Model.png]] In my opinion, the most salient case of practical success in utilizing ontologies for decision-making, and merging them with AI, comes from an enterprise-software development company: [[Palantir - Ontology]]. Palantir has been growing exponentially, indicating a major challenge for the incumbent enterprise SaaS providers in the future. --- This **Ontology for Decision-Making under Uncertainty (ODMU)** is a proposal based on the experience accumulated in my executive and consulting career, hard-learning how to integrate Risk and Performance Management in the real-world. The resulting method has been used in many projects to map risk and performance drivers to other critical entities, allowing methodological-sound auditing and diagnosis of the inner bearings of a client's management system. **ODMU** is a general-purpose document that fits any type of organization, so I would classify it as a potential ontology for the field of **Decision Modeling**. Risk and Performance drivers are present in the ontology, but they are not its center: in **ODMU**, Risk and Assumption drivers are *convolved* to support **decision-making**, the core of the ontology. A *Dependency Knowledge Graph* that derives from this ontology, built for any given organization, has the potential to replace with many advantages the risk register and the KPI system. To facilitate the exploration by customers and site users, I kept the ontology simple, not yet in a “ready for-developers” OWL format. **ODMU** is currently written in Markdown syntax, a simple language that business persons will interact easily with and that is recognizable by any software. Since it is created to be a tool for training, it will be worked out and refined many times over, based on users' input. Cuurently, **ODMU** has 20 Entities ([[Entities and Definitions]]), presented in standard Obsidian format. You will find each of them under the label **Templates** in the left-side menu. Each template represents an Entity that has pre-defined taxonomies and other properties and links to the other 19 entities. Obsidian's Graph View, Canvas, Juggl and 3-D viewer plugins allow for studying the connections of the whole or any part, zooming in and out and altering perspectives. The **Project Enchiridion**'s open-access site has an always-present Graph View on the right side of the screen. This is a knowledge graph of the full **ODMU** view: [[Overview Enchiridion Ontology.png]] An extensive use-case is under development for teaching and templating for similar applications: the **Verdenost Cooperative**. Click on *'Verdenost Applied Ontology/Verdenost Entities/Decision Domains and Hubs/Sustainability/User and Society Safety'* and follow the links. Of course, if you become a member of the Risk Leap Program, we will teach you Obsidian tricks and you can wander freely around your own vault.