{ "$schema": "https://json-schema.org/draft/2020-12/schema", "$id": "#/components/schemas/TokenizerUserDictConfig", "title": "TokenizerUserDictConfig", "type": "object", "description": "A list of pairs of strings that should be replaced with another string during tokenization.", "properties": { "tokenizer": { "type": "string", "description": "The tokenizer to which the user dictionary should be applied. Currently, only the `kagame` ja and kr tokenizers supports user dictionaries." }, "replacements": { "type": "array", "items": { "type": "object", "required": [ "source", "target" ], "properties": { "source": { "type": "string", "description": "The string to be replaced." }, "target": { "type": "string", "description": "The string to replace with." } } } } } }