dataset/noising_resources/accents.json

{"ă": [
        ["ắằẳẵặ", "âấầẩẫậ", "aáàảãạ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "â": [
        ["ấầẩẫậ", "aáàảãạ", "ăắằẳẵặ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "á": [
        ["aàảãạ", "ăắằẳẵặ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "à": [
        ["aáảãạ", "ăắằẳẵặ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ả": [
        ["aáàãạ", "ăắằẳẵặ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ã": [
        ["aáàảạ", "ăắằẳẵặ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ạ": [
        ["aáàảã", "ăắằẳẵặ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ắ": [
        ["ăằẳẵặ", "aáàảãạ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ằ": [
        ["ăắẳẵặ", "aáàảãạ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ẳ": [
        ["ăắằẵặ", "aáàảãạ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ặ": [
        ["ăắằẳẵ", "aáàảãạ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ẵ": [
        ["ăắằẳặ", "aáàảãạ", "âấầẩẫậ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ấ": [
        ["âầẩẫậ", "aáàảãạ", "ăắằẳẵặ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ầ": [
        ["âấẩẫậ", "aáàảãạ", "ăắằẳẵặ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ẩ": [
        ["âấầẫậ", "aáàảãạ", "ăắằẳẵặ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ẫ": [
        ["âấầẩậ", "aáàảãạ", "ăắằẳẵặ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ậ": [
        ["âấầẩẫ", "aáàảãạ", "ăắằẳẵặ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "é": [
        ["eèẻẽẹ", "êếềểễệ"
        ],
        [
            0.7,
            0.3
        ]
    ], "è": [
        ["eéẻẽẹ", "êếềểễệ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ẻ": [
        ["eéèẽẹ", "êếềểễệ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ẽ": [
        ["eéèẻẹ", "êếềểễệ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ẹ": [
        ["eéèẻẽ", "êếềểễệ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ê": [
        ["eéèẻẽẹ", "ếềểễệ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ế": [
        ["eéèẻẽẹ", "êềểễệ"
        ],
        [
            0.3,
            0.7
        ]
    ], "ề": [
        ["eéèẻẽẹ", "êếểễệ"
        ],
        [
            0.3,
            0.7
        ]
    ], "ể": [
        ["eéèẻẽẹ", "êếềễệ"
        ],
        [
            0.3,
            0.7
        ]
    ], "ễ": [
        ["eéèẻẽẹ", "êếềểệ"
        ],
        [
            0.3,
            0.7
        ]
    ], "ệ": [
        ["eéèẻẽẹ", "êếềểễ"
        ],
        [
            0.3,
            0.7
        ]
    ], "í": [
        ["iìỉĩị", "ýỳỷỹỵ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ì": [
        ["iíỉĩị", "ýỳỷỹỵ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ỉ": [
        ["iíìĩị", "ýỳỷỹỵ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ĩ": [
        ["iíìỉị", "ýỳỷỹỵ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ị": [
        ["iíìỉĩ", "ýỳỷỹỵ"
        ],
        [
            0.7,
            0.3
        ]
    ], "ó": [
        ["oòỏọõ", "ôốồổỗộ", "ơớờởợỡ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ò": [
        ["oóỏọõ", "ôốồổỗộ", "ơớờởợỡ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ỏ": [
        ["oóòọõ", "ôốồổỗộ", "ơớờởợỡ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "õ": [
        ["oóòỏọ", "ôốồổỗộ", "ơớờởợỡ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ọ": [
        ["oóòỏõ", "ôốồổỗộ", "ơớờởợỡ"
        ],
        [
            0.7,
            0.15,
            0.15
        ]
    ], "ô": [
        ["oóòỏọõ", "ốồổỗộ", "ơớờởợỡ"
        ],
        [
            0.15,
            0.7,
            0.15
        ]
    ], "ố": [
        ["oóòỏọõ", "ôồổỗộ", "ơớờởợỡ"
        ],
        [
            0.15,
            0.7,
            0.15
        ]
    ], "ồ": [
        ["oóòỏọõ", "ôốổỗộ", "ơớờởợỡ"
        ],
        [
            0.15,
            0.7,
            0.15
        ]
    ], "ổ": [
        ["oóòỏọõ", "ôốồỗộ", "ơớờởợỡ"
        ],
        [
            0.15,
            0.7,
            0.15
        ]
    ], "ộ": [
        ["oóòỏọõ", "ôốồổỗ", "ơớờởợỡ"
        ],
        [
            0.15,
            0.7,
            0.15
        ]
    ], "ỗ": [
        ["oóòỏọõ", "ôốồổộ", "ơớờởợỡ"
        ],
        [
            0.15,
            0.7,
            0.15
        ]
    ], "ơ": [
        ["oóòỏọõ", "ôốồổỗộ", "ớờởợỡ"
        ],
        [
            0.15,
            0.15,
            0.7
        ]
    ], "ớ": [
        ["oóòỏọõ", "ôốồổỗộ", "ơờởợỡ"
        ],
        [
            0.15,
            0.15,
            0.7
        ]
    ], "ờ": [
        ["oóòỏọõ", "ôốồổỗộ", "ơớởợỡ"
        ],
        [
            0.15,
            0.15,
            0.7
        ]
    ], "ở": [
        ["oóòỏọõ", "ôốồổỗộ", "ơớờợỡ"
        ],
        [
            0.15,
            0.15,
            0.7
        ]
    ], "ợ": [
        ["oóòỏọõ", "ôốồổỗộ", "ơớờởỡ"
        ],
        [
            0.15,
            0.15,
            0.7
        ]
    ], "ỡ": [
        ["oóòỏọõ", "ôốồổỗộ", "ơớờởợ"
        ],
        [
            0.15,
            0.15,
            0.7
        ]
    ], "ú": [
        ["uùủũụ", "ưứừữửự"
        ],
        [
            0.7,
            0.3
        ]
    ], "ù": [
        ["uúủũụ", "ưứừữửự"
        ],
        [
            0.7,
            0.3
        ]
    ], "ủ": [
        ["uúùũụ", "ưứừữửự"
        ],
        [
            0.7,
            0.3
        ]
    ], "ũ": [
        ["uúùủụ", "ưứừữửự"
        ],
        [
            0.7,
            0.3
        ]
    ], "ụ": [
        ["uúùủũ", "ưứừữửự"
        ],
        [
            0.7,
            0.3
        ]
    ], "ư": [
        ["uúùủũụ", "ứừữửự"
        ],
        [
            0.3,
            0.7
        ]
    ], "ứ": [
        ["uúùủũụ", "ưừữửự"
        ],
        [
            0.3,
            0.7
        ]
    ], "ừ": [
        ["uúùủũụ", "ưứữửự"
        ],
        [
            0.3,
            0.7
        ]
    ], "ử": [
        ["uúùủũụ", "ưứừữự"
        ],
        [
            0.3,
            0.7
        ]
    ], "ữ": [
        ["uúùủũụ", "ưứừửự"
        ],
        [
            0.3,
            0.7
        ]
    ], "ự": [
        ["uúùủũụ", "ưứừữử"
        ],
        [
            0.3,
            0.7
        ]
    ], "ý": [
        ["yỳỷỵỹ", "iíìỉĩị"
        ],
        [
            0.7,
            0.3
        ]
    ], "ỳ": [
        ["yýỷỵỹ", "iíìỉĩị"
        ],
        [
            0.7,
            0.3
        ]
    ], "ỷ": [
        ["yýỳỵỹ", "iíìỉĩị"
        ],
        [
            0.7,
            0.3
        ]
    ], "ỵ": [
        ["yýỳỷỹ", "iíìỉĩị"
        ],
        [
            0.7,
            0.3
        ]
    ], "ỹ": [
        ["yýỳỷỵ", "iíìỉĩị"
        ],
        [
            0.7,
            0.3
        ]
    ]
}