Relativity Analytics (Structured & Conceptual) Flashcards

In email threading results, what does an "inclusive" email mean?

Answer: An email that contains unique content not found in any other email in the thread

Inclusive emails contain content (text or attachments) that appears nowhere else in the thread. Reviewing only the inclusive emails still covers everything in the thread.

Which structured analytics operation groups documents whose extracted text is highly similar based on a minimum similarity percentage?

Answer: Textual near duplicate identification

Textual near duplicate identification compares extracted text and groups documents that meet the minimum similarity percentage, with one principal document per group.

What is the key difference between structured analytics and conceptual analytics in Relativity?

Answer: Structured analytics works from text and metadata patterns; conceptual analytics uses a conceptual index to find meaning-based relationships

Structured analytics (threading, near-dupes, language ID, repeated content) analyzes text structure and metadata. Conceptual analytics uses a conceptual index (LSI) for clustering, categorization and concept searching.

Which conceptual analytics feature groups documents into a hierarchy by concept with no example documents from the reviewer?

Answer: Clustering

Clustering is unsupervised: it groups documents by conceptual similarity without examples. Categorization needs example documents.

Which repeated content should usually be filtered out before you build a conceptual analytics index?

Answer: Boilerplate such as confidentiality footers and disclaimers

Boilerplate such as email disclaimers and footers can distort conceptual relationships. Repeated content identification finds it so it can be applied as a filter on the index.

A reviewer highlights a paragraph and wants to find other documents about the same idea, even if they use different words. Which feature fits best?

Answer: Concept search

Concept search uses the conceptual index to return documents that are similar in meaning to the sample text, even when they don't share the same keywords.