Aryan Singh Dalal

Ontologies support shared meaning across Semantic Web applications, but selecting an appropriate ontology for reuse remains difficult because existing candidates often differ in documentation quality, semantic clarity, structural connectivity, and logical soundness. Reuse decisions are therefore frequently based on familiarity or intuition rather than reproducible evidence. This paper extends WiseOWL [A. S. Dalal, M. Baloch, A. Y. Lin, A. M. Masci, K. M. Jagodnik and H. K. McGinty, WiseOWL: A methodology for evaluating ontological descriptiveness and semantic correctness for ontology reuse and ontology recommendations, in 2026 Int. Conf. Semantic Computing, 2026, pp. 9–15], a schema-level methodology and web-based tool for evaluating ontology reuse readiness from an ontology’s own RDF/OWL content. WiseOWL defines four core quality metrics: Well-Described, which measures the coverage of human-readable annotations; Well-Defined, which evaluates the semantic alignment and adequacy of labels and definitions using transformer-based embeddings and textual heuristics; Connection, which assesses the breadth, diversity, and richness of object–property relationships; and Hierarchical Breadth, which measures the balance of class-taxonomy depth and branching, including hierarchy inferred from complex OWL axioms. To complement these aggregate scores, this extension introduces three diagnostic metrics: Logical Consistency, Structural Distinctiveness, and Semantic Distinctiveness, which identify unsatisfiable classes, redundant subclass assertions, and candidate near-duplicate concepts, respectively. The methodology is now formalized using RDF graph notation, bounded score definitions, algorithmic pseudocode, and stated metric properties. WiseOWL is implemented as a Streamlit application that accepts OWL files, converts them to RDF Turtle, computes normalized 0–10 scores, and presents interactive visualizations with exportable results. Evaluation is extended to seven publicly available ontologies — Plant Ontology (PO), Gene Ontology (GO), Semanticscience Integrated Ontology (SIO), Food Ontology (FoodON), Friend of a Friend (FOAF), Dublin Core Terms, and GoodRelations. The results show that WiseOWL produces distinguishable quality profiles and identifies concrete areas for improvement.

Constructing expressive axioms in the Web Ontology Language and description logics, such as subclass, disjointness, domain, range, and general concept inclusion axioms, remains a bottleneck in ontology learning, because manual authoring does not scale to large knowledge graphs. Recent surveys have covered parts of this field, but none that we identified places lexical methods, symbolic and neural-symbolic description logic learners, triple-based inductive and logical methods, and large language model methods side by side on shared dimensions. This narrative survey unifies these four families under a single taxonomy of axiom construction over Resource Description Framework data and related inputs, including triples, text, labeled examples, ontology signatures, and prompts. We compare the families by input requirements, axiom coverage, controllability, scalability, and dependence on triples. We identify a recurring architectural pattern in the language model literature, in which a generative model proposes candidate axioms and a verification layer based on retrieval, schema repair, entailment, methodology audit, or description logic reasoning constrains them. Because the studies use different benchmarks and metrics, we do not treat their results as directly comparable, and we discuss a hybrid pipeline argument in which lexical and language model proposers, triplebased scorers, and description logic learners and reasoners each carry the operational load they are best suited for. The principal open problem is evaluation, especially benchmark leakage in the language model era.

Large language models can draft ontologies, but unverified extraction yields hallucinated triples—producing plausible yet incorrect facts. EchoLLM is a text-only, evidence-grounded pipeline for ontology construction. Candidate triples are first extracted with an instruction-following LLM. A hybrid retriever (BM25 + dense) gathers sentence-level evidence for each triple. Natural language inference then tests whether the evidence entails the triple; only entailed, lexically consistent hypotheses are accepted, and all decisions are logged. Accepted entities are embedded and clustered to induce classes and a lightweight hierarchy; rdfs:comment is generated from supporting text. The result is a validated triple set and an initial ontology suitable for bootstrapping domain knowledge graphs. The construction design favors high precision which requires no domain-specific rules, and surfaces failure modes (extraction, retrieval, verification). This enables authors and subject-matter experts to build trustworthy knowledge graphs quickly while keeping model and cost choices flexible.

The traditional approach to ontology development is characterized by its labor-intensive nature, requiring extensive effort and domain expertise to define intricate structures, relationships, and concepts accurately. This study proposes a paradigm shift in ontology development by harnessing the capabilities of large language models (LLMs). This methodology entails the creation of an interactive interface that empowers users to query LLMs using prompts, facilitating the retrieval of pertinent information with ease. Subsequent analysis of this information allows for identifying key relationships, which are then transformed into graph structures using the Web Ontology Language (OWL). The outcome of this process is OLIVE—an ontology development workflow engineered to streamline manual efforts and minimize the risk of errors.