Download Relativity Analytics Specialist Practice Exam and more Exams Technology in PDF only on Docsity!
Question 1. Which of the following best describes the primary purpose of a Relativity Analytics Index? A) To store raw document files for e‑discovery B) To provide a searchable repository for structured metadata only C) To enable conceptual analytics such as clustering and concept searching D) To replace the Relativity database entirely Answer: C Explanation: An Analytics Index is built to support conceptual analytics features like clustering, concept search, and keyword expansion, not just raw file storage. Question 2. Structured analytics differs from conceptual analytics in that it primarily handles: A) Semantic relationships between terms B) Exact text matching and duplicate detection C) Machine‑learning classification of concepts D) Visualization of document clusters Answer: B Explanation: Structured analytics focuses on tasks like email threading and near‑duplicate identification, which rely on exact or pattern‑based matching. Question 3. The Latent Semantic Indexing (LSI) model in Relativity Analytics is used to: A) Compress document images for faster upload
B) Discover underlying concepts by analyzing term co‑occurrence C) Encrypt index data for security compliance D) Generate PDF renditions of native files Answer: B Explanation: LSI creates a mathematical representation of concepts based on term patterns, enabling concept discovery. Question 4. When creating a new Analytics Index, which of the following must be defined first? A) The visual theme for clustering results B) The Data Source saved search that supplies documents to the index C) The user permissions for index access D) The email notifications for index build completion Answer: B Explanation: The Data Source saved search determines which documents are ingested into the index and is required before any further configuration. Question 5. Which file type is generally excluded from an Analytics Index because it contains insufficient textual content? A) PDF with searchable text B) Plain‑text (.txt) C) Scanned image PDF without OCR D) Microsoft Word (.docx)
Question 8. Enabling the “Optimize Training Set” option removes documents that contain: A) More than 10,000 characters of text B) Only numbers, placeholder text, or no meaningful words C) Embedded images with captions D) Custom metadata fields with null values Answer: B Explanation: Optimization removes low‑value documents such as those with only numbers or placeholder text to improve model quality. Question 9. Managing stop words in an Analytics Index helps to: A) Increase the index size for better performance B) Prevent common, non‑informative terms from influencing concept formation C) Encrypt the index data for compliance D) Automatically tag privileged documents Answer: B Explanation: Stop words are excluded to avoid diluting the conceptual model with high‑frequency, low‑value terms. Question 10. The main difference between a Full Build and an Incremental Build of an Analytics Index is: A) Full Build creates visualizations, Incremental Build does not
B) Full Build processes all source documents, Incremental Build processes only new or changed documents C) Full Build is only available for structured analytics, Incremental Build for conceptual analytics D) Full Build runs on a weekly schedule, Incremental Build runs hourly Answer: B Explanation: A Full Build re‑indexes the entire data set, while an Incremental Build updates the index with only new or modified documents. Question 11. Which statistic is most useful for identifying a high error rate during an Analytics Index build? A) Number of documents indexed B) Percentage of documents with “Failed” status C) Average document size in megabytes D) Total time taken to complete the build Answer: B Explanation: The “Failed” status percentage directly indicates errors encountered during indexing. Question 12. If a document fails to index due to “Unsupported file type,” the appropriate remediation is to: A) Increase the index’s memory allocation B) Add the file type to the “Allowed File Types” list in the index configuration
Explanation: By default, only documents with a Rank Score of 0.5 or higher are returned, though this can be adjusted. Question 15. Which type of source document yields the most accurate results when using the “Find Similar Documents” feature? A) A document containing multiple unrelated topics B) A very short email signature only C) A document focused on a single, clear concept or topic D) A scanned image without OCR text Answer: C Explanation: A clear, single‑topic document provides a strong conceptual signature for similarity matching. Question 16. Keyword Expansion in Relativity Analytics primarily helps reviewers to: A) Increase the number of documents in the index automatically B) Discover additional terms that are conceptually related to the original keyword set C) Remove stop words from the original query D) Translate keywords into multiple languages Answer: B Explanation: Keyword Expansion returns conceptually related terms, broadening search coverage.
Question 17. After running a Keyword Expansion, the recommended next step is to: A) Re‑build the entire Analytics Index B) Apply the expanded terms to a new Structured Search C) Use the expanded term list in a Concept Search or Boolean query to retrieve more relevant documents D) Delete the original keyword list to avoid duplication Answer: C Explanation: Expanded terms are intended to improve subsequent searches for related documents. Question 18. When creating a clustering operation, which of the following must be specified? A) The color palette for the visualization B) The number of clusters to generate or let the engine determine automatically C) The exact file path for each cluster’s output file D) The user’s email address for notification Answer: B Explanation: The clustering process needs a target number of clusters (or an automatic setting) to partition the data.
B) Coherence Score C) Recall D) Elusion Rate Answer: B Explanation: The Coherence Score quantifies how tightly related the documents are inside a cluster. Question 22. To review documents inside a specific cluster, a user should navigate to: A) The Index Settings page B) The “Documents” tab within the clustering results view C) The “Analytics Dashboard” home screen D) The “User Management” area Answer: B Explanation: The Documents tab lists all items belonging to the selected cluster for review. Question 23. When creating a Categorization Set, the “Example Data Source” saved search is used to: A) Define the set of documents that will be automatically classified by the model B) Provide labeled example documents that teach the model each category C) Exclude privileged documents from classification D) Generate a visual heat map of category distribution
Answer: B Explanation: Example Data Source supplies training examples for each target category. Question 24. After running a Categorization Set, a high “Confidence Score” for a document indicates: A) The document is definitely irrelevant to the case B) The model is highly certain about the document’s assigned category C) The document contains no searchable text D) The document should be manually reviewed before acceptance Answer: B Explanation: Confidence reflects the model’s certainty in its categorization decision. Question 25. In Continuous Active Learning (CAL), the “Pre‑coded Documents” serve to: A) Randomly generate synthetic documents for testing B) Seed the initial model with known relevant and non‑relevant examples, improving early ranking C) Disable the model until enough user feedback is collected D) Automatically tag privileged material Answer: B
Question 28. The Coverage Review Queue is designed to: A) Present documents that are already known to be irrelevant B) Provide documents that will most improve the model’s accuracy when coded C) Show only privileged documents for attorney review D) List documents that have failed indexing Answer: B Explanation: Coverage Review selects documents that fill gaps in the model’s knowledge, enhancing training. Question 29. When should the Random Sampling Queue be used in an Active Learning project? A) Only after the model reaches 95% recall B) At the start of a project to create an unbiased baseline and for statistical validation later C) When the Prioritized Queue runs out of documents D) Only for documents with missing metadata Answer: B Explanation: Random sampling provides an unbiased sample for validation and initial model seeding. Question 30. Synchronization in a CAL project refers to: A) The process of encrypting all coded decisions for security compliance
B) Updating the machine‑learning model and document rank scores whenever a reviewer codes a document C) Replicating the project to a secondary server for disaster recovery D) Aligning the project’s start date with the case timeline Answer: B Explanation: Synchronization ensures that each new coding decision is incorporated into the model and ranking. Question 31. Which metric indicates the proportion of truly responsive documents that have been identified by the model? A) Richness B) Recall C) Elusion D) Precision Answer: B Explanation: Recall measures the percentage of all relevant documents that the model has retrieved. Question 32. A high “Richness” value in a CAL project suggests: A) The model is overfitting to a small set of documents B) There is a high proportion of responsive documents in the overall data set C) The index contains many duplicate files D) The project has a large number of privileged documents
Question 35. To conduct an Elusion Test, a reviewer must: A) Randomly select a sample of documents from the entire collection and manually code them to estimate missed responsive items B) Run a full rebuild of the Analytics Index C) Export all coded documents to an external spreadsheet D) Disable the Prioritized Review Queue temporarily Answer: A Explanation: Elusion testing involves sampling to estimate the proportion of responsive documents not retrieved by the model. Question 36. During a Recall Test, the primary goal is to: A) Verify that the model’s precision is above 90% B) Measure the percentage of all truly responsive documents that have been identified by the model C) Determine the average time reviewers spend per document D) Identify privileged documents within the collection Answer: B Explanation: Recall testing assesses how comprehensively the model has captured relevant material. Question 37. Which permission is required for a user to add new documents to an existing Active Learning project? A) View Project Only
B) Edit Project Settings C) Delete Project D) Export Project Data Answer: B Explanation: Editing project settings includes the ability to modify the document set. Question 38. In Review Center, a document that appears in the “Skipped” category typically means: A) The document failed to load due to a corrupted file B) The document did not receive a rank score because it lacked sufficient text for the model C) The document was marked as privileged and removed from review D) The document was already coded in a previous project Answer: B Explanation: Skipped documents are those the model could not evaluate, often because they contain insufficient textual content. Question 39. A common cause of “Insufficient Positive Documents” in a CAL project is: A) Too many documents with large file sizes B) An overly restrictive keyword set that fails to retrieve relevant material early on C) The presence of duplicate documents in the data set
Explanation: Population Statistics provide a breakdown of document processing outcomes during indexing. Question 42. Which of the following is NOT a recommended best practice for documents included in an Analytics Index training set? A) Ensure each document contains at least 100 characters of searchable text B) Include a mix of document types (emails, PDFs, Word files) for broader coverage C) Add system log files that contain only timestamps and IDs D) Remove documents that are pure placeholders or contain only numbers Answer: C Explanation: System logs lack meaningful text and can degrade the conceptual model. Question 43. A “Coherence Score” of 0.85 for a cluster indicates: A) The cluster contains 85% duplicate documents B) The documents in the cluster share a high degree of conceptual similarity C) 85% of the cluster’s documents are unranked D) The cluster was built using 85% of the total index vocabulary Answer: B Explanation: High coherence reflects strong internal conceptual alignment among the cluster’s documents.
Question 44. In a Concept Search, increasing the “Minimum Rank Score” threshold will: A) Return more documents, including less relevant ones B) Reduce the number of returned documents, focusing on higher similarity matches C) Change the underlying LSI model parameters D) Disable the use of stop words in the query Answer: B Explanation: Raising the threshold filters out lower‑scoring results, yielding a more focused set. Question 45. Which visualization best helps an analyst identify clusters that are conceptually close to a selected cluster? A) Timeline view B) Nearby Clusters view C) Document list view D) Heat map of file extensions Answer: B Explanation: The Nearby Clusters visualization highlights clusters with minimal conceptual distance to the selected one. Question 46. When setting up a new Active Learning project, the “Project Validation” saved search is used to:
