[Paper Review] Inference and Plausible Reasoning in a Natural Language Understanding System Based on Object-Oriented Semantics

Yuriy Ostapov|arXiv (Cornell University)|Feb 1, 2012

Semantic Web and Ontologies3 references3 citations

TL;DR

This paper presents a natural language understanding system based on object-oriented semantics and predicate calculus that combines deductive and plausible reasoning to answer complex logical questions in social domains. By modeling facts, roles, scripts, and psychological norms, the system enables inference for tasks like cause detection, action planning, and state prediction—demonstrating applicability in criminology, medicine, and business systems.

ABSTRACT

Algorithms of inference in a computer system oriented to input and semantic processing of text information are presented. Such inference is necessary for logical questions when the direct comparison of objects from a question and database can not give a result. The following classes of problems are considered: a check of hypotheses for persons and non-typical actions, the determination of persons and circumstances for non-typical actions, planning actions, the determination of event cause and state of persons. To form an answer both deduction and plausible reasoning are used. As a knowledge domain under consideration is social behavior of persons, plausible reasoning is based on laws of social psychology. Proposed algorithms of inference and plausible reasoning can be realized in computer systems closely connected with text processing (criminology, operation of business, medicine, document systems).

Motivation & Objective

• To enable computer systems to answer complex logical questions in natural language when direct database matching fails.
• To integrate both deduction and plausible reasoning—grounded in social psychology—into a semantic processing framework.
• To model social behavior, roles, scripts, and plans using predicate calculus for real-world application in domains like criminology and healthcare.
• To develop a knowledge representation system that supports scalable, extensible, and modular inference using relational databases.
• To demonstrate the feasibility of automating reasoning tasks in enterprise management and criminal investigation using natural language queries.

Proposed method

• Represent natural language sentences as sets of predicates describing persons, objects, actions, and states using first-order logic.
• Store facts and knowledge in relational databases using tables mapped to predicates for efficient storage and retrieval.
• Implement inference through rule-based deduction when direct predicate matching fails to answer a question.
• Apply plausible reasoning based on social psychological norms and role behavior to infer non-obvious conclusions (e.g., motives, causes, states).
• Use scripts (e.g., ship distress, illness) and plans (e.g., city layouts, office robbery) as structured knowledge to guide reasoning about sequences of events.
• Combine multiple reasoning paths: check hypotheses, determine motives, plan actions, and infer causes or states using symptom, script, and event-based triggers.

Experimental results

Research questions

• RQ1 How can a natural language understanding system infer answers to logical questions when direct predicate matching fails?
• RQ2 What role do social psychological norms play in enabling plausible reasoning about human behavior in non-typical or ambiguous situations?
• RQ3 How can structured knowledge—such as scripts, plans, and disease descriptions—be encoded and used to support inference in social domains?
• RQ4 In what ways can plausible reasoning be formalized using logical rules to ensure consistency and validity in reasoning about motives and causes?
• RQ5 How can such a system be integrated into real-world applications like criminology, medicine, or enterprise management to automate complex decision-making?

Key findings

• The system successfully supports inference for four core problem classes: hypothesis validation, non-typical action analysis, action planning, and cause/state determination.
• Plausible reasoning based on social psychology laws enables inference about motives and behaviors outside standard role expectations.
• Scripts and plans (e.g., ship distress, office robbery) are effectively encoded as predicate-based knowledge structures to model sequences of events and actions.
• The system can answer natural language questions like 'Why is a girl dead on November 7th in 9 Street1 Street?' by identifying causal actions and events.
• The integration of relational databases with logical inference allows scalable processing of large volumes of factual and conceptual data.
• The approach enables automation of complex reasoning tasks in domains such as criminology and enterprise management, reducing reliance on manual data processing and expert interpretation.