Researchers address open questions in language identification in the limit by defining computational traces directly from languages rather than underlying machine models. They demonstrate that positive identification results are achievable using a small alphabet of tokens linear in the size of the language's alphabet.
- Traces are defined directly from the language without requiring an explicit automata-theoretic machine model.
- The token alphabet size is linear relative to the language's alphabet and independent of other language properties.
- This approach overcomes previous limitations where traces relied on large alphabets from specific generating machines.