Data quality is enhanced when most entities have human-readble label. Once such a label is available, data quality is higher when such labels do not contain typo's. If a dataset has labels without typo's for all entities, it becomes easier to search for entities based on their name (text search).
The following table shows entities that have multiple labels. As can be observed, alternative labels often are typographic mistakes of a correct label, or they are abbreviations of the full label. In both cases, such labels are redundant (abbreviated labels do not need to be stored separately, but can be based on the full label).
The following table shows a place that has two different names.