Figure 3 - uploaded by Ghazaleh Kazeminejad
Content may be subject to copyright.
REO-SUMO Linking Model: This model imports the abbreviated SUMO model (containing ENDURANT-type classes only) and the REO reference model. Thus, REO specifies the PERDURANTS of the model, while SUMO specifies the ENDURANTS.
Source publication
Commonsense, real-world knowledge about the events that entities or "things in the world" are typically involved in, as well as part-whole relationships, is valuable for allowing computational systems to draw everyday inferences about the world. Here, we focus on automatically extracting information about (1) the events that typically bring about c...
Context in source publication
Context 1
... linking model imports an abbreviated OWL version of SUMO (containing only the ENDURANT-compatible classes mentioned, their documentation that provides a description of what they are, and their associated WordNet sense keys), and the REO OWL reference model. The resulting model has all of the REO event ontology nested under PERDURANT, and the extracted SUMO content nested under ENDURANT (see Figure 3). Admittedly, there are a variety of weaknesses to this model. ...
Citations
... Direct telic seems roughly analogous to Dispositions.11 [22] explored mining SUMO rules to enhance the Rich Event Ontology. We believe that it could be more effective to incorporate the Rich Event Ontology as a component of the SUMO ontology.12 ...
We explore using the Suggested Upper Merged Ontology (SUMO) to develop a semantic simulation. We provide two proof-of-concept demonstrations modeling transitions in a simulated gasoline engine using a general-purpose programming language. Rather than focusing on computationally highly intensive techniques, we explore a less computationally intensive approach related to familiar software engineering testing procedures. In addition, we propose structured representations of terms based on linguistic approaches to lexicography. 1 Introduction We believe knowledge representation should be fully integrated with programming languages. Therefore, we are exploring the implementation of dynamic semantic simulations based on ontologies using a general-purpose programming language (cf., [4]). These simulations allow model-level constructs such as flows, states, transitions, microworlds, generalizations, and causation, and language features such as conditionals, threads, and looping. In this paper, we provide initial demonstrations for how the Suggested Upper Merged Ontology (SUMO) can be applied to Python-based semantic modeling. SUMO has both a rich ontology and a sophisticated inference environment built to use first-order predicate calculus [9, 15, 16, 25, 27, 28]. 1 The SUMO ontology incorporates approaches from several other ontologies ([28] p94). Like the Descriptive Ontology for Linguistic and Cognitive Engineering (DOLCE) [12], SUMO also incorporates insights from linguistics. In fact, one extension of SUMO explores Natural Language Generation ([17]). The SUMO ontology is implemented with SUO-KIF, which is a subset of KIF (Knowledge Interchange Format) [18]. KIF is a notation based on the operators of first-order logic (FOL). 2 As a type of description logic, SUMO includes rules, which are implemented with formulas. These represent constraints about the world. Computationally intensive theorem proving for large ontologies has been the focus of much of the recent research on SUMO. By comparison, we explore state-based modeling for small example applications. This is a companion to [4] semantic modeling using the Basic Formal Ontology (BFO) [10]. As in that work, the interactions studied are object-driven. We do not focus on complex inference in this paper; rather we apply simple test cases analogous to those used in requirements testing and model checking [14, 18] to detect possible conflicts in domains, states, and relationships following Transitions. Truth maintenance [24] considers how to ensure that there are only true statements in a knowledgebase as new statements are added. The original knowledgebase is assumed to be true and any incoming statements that conflict with those are rejected. Research on Truth Maintenance Systems (TMS) explores robust and general abstractions to detect and resolve conflicts. In the interest of practical applications, we support lightweight, tractable approaches to inference and truth maintenance. These approaches are related to those from software engineering used to 1 General information about the SUMO project is available at http://ontologyportal.com/ap/. SUMO's full KIF files, its code and other tools are at https://github.com/ontologyportal/sumo. The bulk of the SUMO ontology and the software are available with only light restrictions. When first exploring SUMO, we found it best to use only a few of the KIF files, together with the Python interface, and a moderately powerful computer. 2 FOL has been criticized as allowing too much flexibility [26, 33] to be a suitable platform for ontologies. However, much of the SUMO ontology avoids the possible pitfalls and any lapses could be remedied. Indeed, we believe that BFO itself could be largely implemented in SUO-KIF.
... [19] explored mining SUMO Rules to enhance the Rich Event Ontology. We believe that it might be more effective to encompass a Rich Event Ontology within the more comprehensive SUMO ontology. ...
While ontologies are typically applied to static descriptions of the world, we propose to apply them as representations for dynamic simulations. In this paper, we explore using the Suggested Upper Merged Ontology (SUMO) to develop a semantic simulation. We provide two proof-of-concept demonstrations modeling transitions in a simulated gasoline engine. In our models, the knowledge base evolves as the simulation executes. Faults can be detected at run-time.
... A comprehensive set of qualia relations have yet to be defined and organized. We are tackling this challenge and aim to make the qualia usable for visual understanding tasks: qualia have been automatically extracted and evaluated for quality via crowdsourcing (Kazeminejad et al., 2018), then encoded as relations between entities and events in the Rich Event Ontology (REO) (Bonial et al., 2016). Assuming, for example, that the objects in Fig. 1 can be recognized accurately, the resulting list of objects (e.g., pot, cereal) can first be queried for their qualia in REO, to discover: pot is USED FOR cooking, and cereal FUNCTIONS AS nourishing and IS A Prepared Food. ...
We describe the task of Visual Understanding and Narration, in which a robot (or agent) generates text for the images that it collects when navigating its environment, by answering open-ended questions, such as 'what happens, or might have happened, here?'
