Which analytics functions require the documents to be in the data source?

The correct answer is that clustering and categorization require the documents to be in the data source because both functions rely on direct access to the content and context of the documents for effective analysis.

Clustering involves grouping similar documents based on patterns and features within the text, which necessitates evaluating the actual content of those documents. This process typically utilizes algorithms that analyze various attributes such as word frequency, term relevance, and semantic relationships among the documents, which can only be done if the documents are readily accessible in the data source.

Categorization, on the other hand, works by labeling documents based on predefined categories or classifications. This function requires evaluating and interpreting the actual text within the documents to determine their relevance and fit within those categories, further emphasizing the need for the documents to be included in the data source.

In contrast, keyword analysis can often be performed on metadata or by using text indices, which may not require the documents to be physically present in the data source. This helps clarify why clustering and categorization are dependent on having access to the documents, affirming that those functions require the documents to be in the data source.