COMPUTATIONAL LINGUISTICS
Radixor
Compiling morphological knowledge into a deterministic language primitive.
Learns word-to-stem transformations from dictionaries and compiles reduced patch-command tries, with independent models and reproducible multilingual evaluation.
Research & publications¶
Nine author-released research manuscripts are maintained together with experimental methods, result data and reproducibility references. The publications are project-associated manuscripts; availability on ResearchGate is not a peer-review claim.
Does a Lexicon-Trained Stemmer Generalize? A 20-Language Study of Held-Out Lexical Families
Edit Costs as Structural Hyperparameters in Dictionary-Induced Stemming: A Multilingual Sensitivity Study
Classifying Multilingual Stemmers by Pareto Strata: Error Regimes in Pairwise Lexical Conflation
The Stemming Policy Is Data: Adapting a Lexicon-Trained Stemmer to Alternative Conflation Objectives
Runtime Efficiency of Lexicon-Trained Stemming: Compiled Structure and a Speed–Reference-Agreement Envelope in Radixor
Lexical Resource Consistency and Learned Transformation Vocabulary in Multilingual Stemming
Compiling Learned Lexical Transformations: Semantic Canonicalization and Accepting-Leaf Generalization in Patch-Command Tries
Where Retrieval Quality Goes: Rethinking Stemming in Modern Search
From Lexical Edits to Verified Stemming Policies: Two-Sided Adaptation in Radixor
View the complete research series, papers and evidence ↗
Engineering model¶
Lexical examples are transformed into compact executable patch commands and a reduced trie. Language knowledge is learned and validated offline; production lookups execute a compiled read-only model. Deterministic alternative outputs preserve ambiguity for downstream search or analysis.
Research lineage¶
Radixor revisits the original Egothor/Stempel stemming lineage as a modern multilingual toolkit. Compact compiled tries and deterministic behaviour connect that earlier language-independent stemming work with contemporary quality engineering. The earlier portfolio places the original search and stemming work in its historical context.