Lexical Candidate Generation for Knowledge Graph and Ontology Matching

This deposit contains all result files, configuration, and analysis summaries reported in the paper. The experimental pipeline evaluates four lexical candidate generation methods — exact label matching, character n-gram retrieval, word-level TF-IDF, and BM25 — across 34 datasets spanning five OAEI benchmark tracks. Experiments were run on an Apple M4 Pro Mac mini (64 GB RAM, macOS Tahoe 26.3.1, Python 3.12.12). Result files are organised into two groups: primary results, which are the raw outputs directly cited in the paper, and analysis summaries, which are derived files produced by the reporting pipeline to support specific claims and figures. All CSV files use UTF-8 encoding with comma delimiters. The hyperparameters column in several files contains JSON-encoded parameter dictionaries. Further details of the result files are available in readme.md file.