[Paper Review] Amnestic Forgery: an Ontology of Conceptual Metaphors

Aldo Gangemi, Mehwish Alam+1 more|arXiv (Cornell University)|May 30, 2018

Language, Metaphor, and CognitionPsychology21 references4 citations

TL;DR

This paper introduces Amnestic Forgery, an OWL-based ontology that formalizes conceptual metaphors by integrating MetaNet’s manually curated metaphor data into the Framester knowledge graph, enabling semantic reasoning over metaphorical mappings, blending, and referential structures in natural language. The key contribution is a reusable, semantically grounded framework for metaphor processing in large-scale knowledge graphs.

ABSTRACT

This paper presents Amnestic Forgery, an ontology for metaphor semantics, based on MetaNet, which is inspired by the theory of Conceptual Metaphor. Amnestic Forgery reuses and extends the Framester schema, as an ideal ontology design framework to deal with both semiotic and referential aspects of frames, roles, mappings, and eventually blending. The description of the resource is supplied by a discussion of its applications, with examples taken from metaphor generation, and the referential problems of metaphoric mappings. Both schema and data are available from the Framester SPARQL endpoint.

Motivation & Objective

• To formalize conceptual metaphor theory within a machine-processable ontology for use in large-scale knowledge graphs.
• To address referential ambiguity in metaphorical language by modeling mappings, roles, and blending structures with formal semantics.
• To enable computational processing of metaphors in natural language, including detection, interpretation, and generation.
• To integrate MetaNet’s metaphor data into Framester, a unified linguistic-factual knowledge graph, using a standardized schema.
• To support empirical research on metaphor phenomena by providing a queryable, extensible ontology with real-world examples.

Proposed method

• The authors designed Amnestic Forgery as an extension of the Framester schema, which unifies linguistic and factual data using formal ontologies.
• They extracted and refactored MetaNet’s informal metaphor representations into a structured, OWL-compliant ontology using the D&S ontology pattern framework.
• The ontology models metaphorical mappings as frame compositions, distinguishing between conservative (inheritance-based) and non-conservative (blending-based) constructions.
• It encodes source and target frames, semantic roles, relational patterns, and inheritance links, enabling SPARQL-based querying via the Framester endpoint.
• The framework supports metaphor generation through compositional reasoning over frame blends, using known mappings from MetaNet and FrameNet.
• The system reuses existing resources such as FrameNet, VerbNet, and BabelNet to enrich metaphor semantics with external linguistic and factual knowledge.

Experimental results

Research questions

• RQ1 How can conceptual metaphors be formally represented in a way that supports semantic reasoning and integration into large knowledge graphs?
• RQ2 What is the role of frame composition and blending in modeling non-conservative metaphorical meaning construction?
• RQ3 How can referential ambiguity in metaphorically filtered situations be modeled and resolved within a formal ontology?
• RQ4 To what extent can an ontology-based approach improve the detection, interpretation, and generation of metaphors in natural language?
• RQ5 How can existing linguistic resources like MetaNet and FrameNet be harmonized into a unified, extensible knowledge graph for metaphor research?

Key findings

• Amnestic Forgery successfully integrates MetaNet’s curated metaphor data into the Framester knowledge graph, enabling formal reasoning over metaphorical mappings.
• The ontology supports both literal and metaphorical frame compositions, distinguishing between conservative inheritance and non-conservative blending mechanisms.
• The system enables the generation of novel metaphors through SPARQL-based queries that compose source and target frames using known relational patterns.
• The framework provides a solution to referential problems in metaphorically induced situations by modeling roles and mappings with formal semantics.
• The resource is publicly available via the Framester SPARQL endpoint and GitHub, supporting reuse in metaphor detection, interpretation, and generation pipelines.
• The integration of Amnestic Forgery with existing knowledge graphs opens new pathways for empirical testing of cognitive metaphor theories and computational models.