Which statement about data linking is true?

• A. It never deals with missing data
• B. It only uses machine learning models
• C. It can use deterministic or probabilistic methods
• D. It is unrelated to data quality

Answer: (C)

Data linking is about figuring out whether records from different sources refer to the same real-world entity. It can be done in a deterministic way, using exact matches on identifiers or strict rules, or in a probabilistic way, where we assess the likelihood that two records match based on similarities across fields (like name, date of birth, address) and a computed probability. This flexibility is why the statement is true: data linking is not restricted to one method. It isn’t limited to machine learning models; ML can be used to learn similarity scores, but many linking approaches rely on rule-based or probabilistic techniques. Missing or noisy data are common in linking scenarios, and probabilistic methods help handle that uncertainty. Moreover, data quality matters for linking accuracy, since errors, duplicates, and inconsistencies can hinder correctly identifying matches.