Source-linked AI summary
Towards a Foundational Ontology for Identifying and Resolving Contradictions in Dialogue-based Human-Robot Interactions
Maitreyee Tewari, Michele Persiani
TL;DR
Dialogue-based HRI research lacks an interoperable formal framework for representing contradictions such as errors, conflicts, and knowledge issues across HRI and HAI. This work applies METHONTOLOGY and Activity Theory to develop ATFOt, reporting natural-language, Set-Theoretic, and First Order Logic formalizations plus three dialogue-guiding principles.
Problem
Existing work addresses contradictions in domain-specific dialogue-based HRI, but lacks a structured, granular, interoperable formal representation usable across HRI and HAI.
Method
The study applies METHONTOLOGY and uses Activity Theory concepts to conceptualize a foundational ontology for dialogue-based collaborative interactions and contradictions.
Results
The preliminary results include natural-language, Set-Theoretic, and First Order Logic definitions of dialogues and contradictions, plus three novel guiding principles.
Takeaways & Limitations
ATFOt is intended to standardize terminology and support formal representation, reasoning, and management of dialogues and contradictions across human-agent interaction domains.
Abstract
from arXiv · showhide
Existing Human-Robot Interaction (HRI) literature has focused on identifying and structuring errors, failures, conflicts, and knowledge issues (called in this work as contradictions) in domain-specific dialogue-based interactions. However, there is still lack of a formal computational framework to represent and define these contradictions, interoperable and usable across HRI and human-agent interaction (HAI) domains. Thus, this research project aims to capture, represent, and evaluate the notion of (1) dialogue-based collaborative interaction and (2) related contradictions in a foundational ontology. METHONTOLOGY, a systematic approach to build domain-independent ontologies was applied. In the conceptualisation stage of the presented ontology, concepts and models from Activity Theory were used. Preliminary results presented in this short article are: (i) Natural language definitions of dialogues and related contradictions in HRI, (ii) Set Theoretic definitions of dialogues and contradictions, and (iii) First Order Logic (FoL) formulation of the contradiction concepts and three novel principles guiding dialogue-based interactions between humans and robots. In summary, we report on ongoing work to develop a foundational ontology based on Activity Theory called Activity Theory-based foundational ontology (ATFOt) to capture and represent the notion of contradictions in HRI.
I. INTRODUCTION
Dialogue-based HRI faces communication failures, conflicts, and knowledge issues, while existing representations remain fragmented across domain-specific work. The paper therefore asks how a foundational ontology can provide a general representation of these contradictions.
- Research gap: LLM integration in dialogue-based HRI exposed persistent gaps in semantic, intentional, contextual, normative, error-management, and uncertainty understanding.The paper also links confabulation and inconsistent uncertainty expressions to communication failures and ethical concerns.
- Research gap: Prior work offers taxonomies and management methods for robot capabilities, social errors, conflicts, knowledge issues, gestures, speech, interaction patterns, and intentions.
- Objective: The research objective is to formally represent dialogue-based HRI and HAI interactions and their communication errors, conflicts, and knowledge issues in a foundational ontology.Following Activity Theory terminology, the paper collectively calls these errors, conflicts, and knowledge issues contradictions.
- Objective: The research question asks how to create a general representation of contradictions in dialogue-based HRI within a foundational ontology.
- Approach and contribution: Activity System Model analysis of HRI and HAI scenarios informed an ontology hierarchy representing contradictions at four levels of abstraction.
- Approach and contribution: The article contributes a novel formal knowledge representation of contradictions and three principles for guiding and transforming dialogue-based interactions and participants.
II. METHODOLOGY
The ontology is developed through METHONTOLOGY, combining specification, conceptualisation, integration, implementation, and evaluation activities. Set Theory and First Order Logic provide the formal representation, while Protégé supports implementation and evaluation.
- METHONTOLOGY: METHONTOLOGY organizes ontology development into specification, knowledge acquisition, conceptualisation, integration, implementation, and evaluation.
- METHONTOLOGY: Specification defines the ontology’s purpose, use case, formality, scope, characteristics, and granularity, producing a natural-language specification document.
- METHONTOLOGY: Conceptualisation structures domain knowledge in a conceptual model describing the problem and solution specified for the ontology.
- METHONTOLOGY: Integration maps developed concepts to existing meta-ontologies and documents the associations between referenced terms and model concepts.
- Implementation and evaluation: Set Theory and First Order Logic formally define contradiction concepts, while Protégé is used to implement and evaluate the ontology.
- Implementation and evaluation: Evaluation verifies correctness and validates that the ontology represents the intended system knowledge.
- Implementation and evaluation: The adopted workflow uses five stages, with integration embedded within specification and conceptualisation.
A. Defining Object Properties and First Order Logic Formalism
The paper models dialogue-based HRI as a collaborative Activity System Model whose object, subjects, artefacts, community, rules, and labour division mediate activity toward an outcome. These properties are defined for formalisation in Set Theory and First Order Logic.
- Object: An object instantiates an entity’s desires and intentions and forms a hierarchy from motives to action goals and physical affordances.
- Object: The object is represented at each abstraction level as a tuple of needs, desires, intentions, goals, and physical affordances.
- Object: Functions connect needs to desires, desires to intentions, intentions to goals, and goals to affordances across successive abstraction levels.
- Subjects and artefacts: Subjects include sentient humans and animals, as well as software, virtual, embodied, and digital agents with agency that adopt or support human goals.
- Subjects and artefacts: Mediating artefacts include technical and psychological tools, represented formally as the union of technical and psychological artefact sets.
- Activity System Model: The Activity System Model represents collective activity through interactions among subject, object, and community mediated by artefacts, rules, and division of labour.
- Collaborative activity: A collaborative activity is a time-indexed function in which subjects and a care-provider community use artefacts and socio-cultural context to transform an object into an outcome.
B. FoL formalism of ASM
The paper formalises dialogue-based HRI as an Activity System Model and defines contradictions as discrepancies within the system or among related activity systems. It distinguishes primary, secondary, tertiary, and quaternary contradictions by their structural locations.
- FoL formalism: ATFOt object properties and relationships are used to formalise the Activity System Model in First Order Logic.
- FoL formalism: The DialHRI definition requires an Activity System Model with distinct community subjects, needs or adopted needs, an object, mediating tools, socio-cultural context, time, and transformation toward an outcome.
- Contradictions: Contradictions are historically emergent, systemic situations arising from discrepancies within a central Activity System Model or among its activity-system family.
- Contradiction types: Primary contradictions occur within one Activity System Model component, whereas secondary contradictions occur between different constitutive components.
- Contradiction types: Tertiary contradictions arise when a culturally advanced activity pattern collides with remnants of an older pattern, such as introducing robots into care-home processes.
- Contradiction types: Quaternary contradictions arise through tensions between a transformed activity system and partner activities, including conflicts over healthcare rules that can affect agency and autonomy.
- Contradiction types: The formal taxonomy defines primary, secondary, tertiary, and quaternary contradictions by relations within the central system, between components, and between central and supporting or advanced systems.
1) FoL Formalism of Contradictions:
The formalism specifies three principles describing how contradictions transform an ontology and alter the focus or relations of collaborative activity systems.
- Principle 1 states that a contradiction in the central ASM transforms the object and shifts focus to an alternative ASM.
- Principle 2 links a contradiction in the central ASM to a conflict with the ASM family.
- Principle 3 extends a central-ASM contradiction to inherited ASMs, including the culturally more-advanced ASM and its related ASM family.
IV. CONCLUSION AND ONGOING WORK
The paper presents preliminary ontological formalisation for dialogue-based HRI and HAI, using Activity Theory to define contradictions and interactions. It contributes formal contradiction concepts and three transformation principles while continuing ATFOt implementation and evaluation.
- The ontology uses Engeström’s Activity System Model and Activity Theory’s contradiction concept as foundations for conceptualising dialogue-based interactions and contradictions.
- The work formalises contradiction concepts and proposes three principles governing ontology transformations caused by inherent contradictions.
- ATFOt is being implemented with set theory, FoL, and Protégé, then evaluated qualitatively and verified through coverage and competency questions.
- ATFOt is intended to help agents formally represent, reason about, and manage dialogues and contradictions while standardising terminology across human-agent dialogue domains.
V. FUNDING
The research received support from European Union-funded projects and programmes.
- Funding came from the HumanE-AI-Net project under Horizon 2020 and the European Union’s Horizon Europe LearnData project.
- HumanE-AI-Net was supported under grant agreement 952026, while LearnData was supported under grant 101086712.
- The funding supported both the research and authorship of the article.