Download Relativity Analytics Repeated Content Filters Practice Test and more Exams Nursing in PDF only on Docsity!
Relativity Analytics Repeated Content Filters Practice
Test
Section 1: Conceptual Analytics & Repeated Content Filters
Question 1: What is the primary purpose of Repeated Content Identification (RCI) in a conceptual analytics index? A) To flag privileged documents for review B) To identify and filter out boilerplate text and other repetitive low-value content that could distort conceptual relationships C) To automatically categorize documents based on their metadata D) To encrypt sensitive data before indexing
Answer: B
Rationale: The main goal of RCI is to locate boilerplate or repetitive text (e.g., disclaimers, headers, footers) and remove it from the index. This prevents that recurring text from skewing the conceptual analysis, so the index more accurately reflects the unique, substantive content of documents.
Question 2: Which of the following tasks lists the correct order for filtering repeated content from a conceptual index? A) Run repeated content identification → Evaluate and tag desired filters → Apply filters to the Analytics index → Click Run: Full B) Apply filters to the Analytics index → Run repeated content identification → Evaluate and tag desired filters → Click Run: Full C) Create conceptual index → Apply filters → Run identification → Evaluate filters D) Evaluate and tag filters → Apply filters → Run identification → Click Run: Full
Answer: A
Rationale: The process must start by identifying the repeated content. After analysis, the administrator evaluates and tags the filters they want to exclude. Only then are the filters applied to the index, which requires a full population ("Click Run: Full") to rebuild the index without the filtered text.
Question 3: By default, how does Relativity select which Repeated Content filters to link to a new conceptual index? A) All filters identified are linked B) The first 200 filters found in the workspace, sorted alphabetically C) The top 200 filters found, sorted in descending order by number of occurrences D) Only filters the user manually selects at creation
Answer: C
Rationale: When a conceptual index is built, the system automatically links the 200 most frequently occurring repeated content patterns. This prioritization ensures the most impactful repetitive content is excluded first.
Question 4: You have a conceptual index that currently has 200 Repeated Content filters linked. You change the "ready to index" field to "Yes" for 5 lower-ranked filters. After the next full population, approximately how many filters will be linked to the index? A) 5 B) 195 C) 200 D) 205
Answer: C
Rationale: The index automatically links the top 200 filters based on occurrence frequency. Manual selections outside the top 200 are not included unless they rank within that threshold. The index will always link the 200 highest-occurrence filters.
Question 5: Can Repeated Content filters be manually added or edited? A) No, all filters are generated automatically by the system B) Yes, administrators can manually create or edit new repeated content filters C) No, only Relativity support can modify them D) Yes, but they can only be added, not edited
Answer: B
Rationale: Repeated content filters can be created in structured analytics and can also be added by hand. Administrators can manually create or edit new repeated content filters, which is helpful to include repetitive phrases or patterns not caught automatically.
Question 6: After applying a new Repeated Content filter to a conceptual index, what action must be taken for the filter to take effect? A) No action is needed; the filter applies immediately B) Run a full population on the conceptual index C) Restart the Relativity application D) Recreate the index from scratch
Answer: B
Rationale: Applying a new filter changes which text is indexed. A full population (Full run) is required to rebuild the index and remove the filtered content. Incremental updates will not incorporate the change.
Question 11: Which statement best describes the difference between conceptual and structured analytics? A) Structured analytics works only on images; conceptual works on text B) Structured analytics uses exact matches; conceptual analytics groups by meaning C) Structured analytics requires manual tagging; conceptual does not D) Structured analytics is only available in the cloud
Answer: B
Rationale: Structured analytics (e.g., email threading, near-duplicate detection) relies on deterministic rules and exact criteria. Conceptual analytics uses semantic similarity to group conceptually related documents even if they lack identical keywords.
Question 12: What is the proper order for filtering repeated content from a conceptual index? A) Run repeated content identification → Evaluate and tag desired filters → Apply filters to the Analytics index → Click Populate Index: Full B) Apply filters to the Analytics index → Run repeated content identification → Evaluate and tag desired filters → Click Populate Index: Full C) Create conceptual index → Apply filters → Run identification → Evaluate filters D) Evaluate and tag filters → Apply filters → Run identification → Click Populate Index: Full
Answer: A
Rationale: The correct sequence is to first run identification to find repeated content patterns, then evaluate and tag which filters to use, apply them to the index, and finally run a full population to rebuild the index and remove the filtered text.
Question 13: What does the "optimize training set" option do when creating a conceptual index? A) It removes duplicate documents from the training set B) It selects the most relevant documents from the training data source to improve concept learning C) It applies repeated content filters automatically D) It limits the training set to 1,000 documents
Answer: B
Rationale: The optimize training set feature evaluates the training data source and selects a representative subset for model building. This can improve index quality and reduce build time by focusing on the most diverse and relevant documents.
Question 14: What must you do after applying a new Repeated Content filter to a conceptual index? A) Immediately run a full population to rebuild the index B) Restart the Relativity application
C) Delete and recreate the index D) No action; filters apply automatically
Answer: A
Rationale: Applying a new filter changes which text is indexed. A full population (Full run) is required to rebuild the index and remove the filtered content. Incremental updates will not incorporate the change.
Question 15: By default, when a conceptual index runs, how many repeated content filters will it link to the index? A) All filters identified B) The first 200 filters found in the workspace, sorted alphabetically C) The top 200 filters found, sorted in descending order by number of occurrences D) Only filters the user manually selects at creation
Answer: C
Rationale: When a conceptual index is built, the system automatically links the 200 most frequently occurring repeated content patterns. This prioritization ensures the most impactful repetitive content is excluded first.
Question 16: What is the role of the "Searchable Set" in an Analytics index? A) It contains documents that will be excluded from all analytics functions B) It defines the documents that will be returned by analytics operations like clustering and concept search C) It is only used for Active Learning model training D) It stores a backup of the original documents
Answer: B
Rationale: The searchable set contains all documents on which analytics functions (clustering, categorization, concept search, etc.) will be performed. Only documents in this set are returned when running these features.
Question 17: What is the difference between conceptual and structured analytics? A) Structured analytics works only on images; conceptual works on text B) Structured analytics uses exact matches; conceptual analytics groups by meaning C) Structured analytics requires manual tagging; conceptual does not D) Structured analytics is only available in the cloud version
Answer: B
Rationale: Structured analytics uses deterministic rules and exact matches (e.g., email threading, near- duplicate detection). Conceptual analytics uses semantic similarity to group conceptually related documents even without identical keywords.
Answer: B
Rationale: Active Learning requires a minimum training foundation. The model builds for the first time once the project has at least five documents coded with the positive choice and five with the negative choice.
Question 22: What does the Prioritized Review queue display before the first model is built? A) Only documents coded as Responsive B) Only documents coded as Not Responsive C) Random documents from the review set D) Documents selected by the Coverage Queue
Answer: C
Rationale: Until the first model is built and rank scores are generated, the system has no basis for prioritization. It therefore serves random documents from the review pool.
Question 23: Which review queue is designed to serve documents that are optimal for training the model by selecting those with high uncertainty? A) Prioritized Review queue B) Coverage queue C) Random queue D) Elusion Test queue
Answer: B
Rationale: The Coverage queue uses an algorithm that picks documents the model is most uncertain about. Coding these efficiently trains the model to define the boundary between Responsive and Not Responsive documents.
Question 24: True or False: Family-based review can be turned off after Prioritized Review has started. A) True B) False
Answer: B
Rationale: The decision to use family-based review must be made at project setup. Once the project has started, you cannot change this selection because it would disrupt the integrity of the review workflow.
Question 25: Can a reviewer change their coding decision on a document after it has been reviewed in an Active Learning project? A) Yes, but the updated coding will not affect the model B) Yes, and the next model build will include the most recent coding update
C) No, coding decisions are final D) Yes, but only project owners can change coding decisions
Answer: B
Rationale: Reviewers can change their coding decisions at any time. The next time the model builds, it will incorporate the most up-to-date coding information for that document.
Question 26: What is the primary purpose of the Elusion Test in Active Learning? A) To identify documents that should be reviewed first B) To validate that the model is not missing a significant number of responsive documents C) To train the initial model D) To automatically code documents as Not Responsive
Answer: B
Rationale: The Elusion Test samples documents coded as Not Responsive to estimate the number of potentially responsive documents the model may have missed. It is a validation tool for project quality control.
Question 27: What is the ideal minimum document count for obtaining meaningful clustering results? A) 100 B) 1, C) 5, D) 10,
Answer: B
Rationale: Clustering algorithms perform best with larger datasets. A minimum of roughly 1, documents is recommended to produce meaningful, stable clusters that reveal true conceptual themes.
Question 28: Which security permissions are required to view and edit an Active Learning project? (Select all that apply) A) View Project B) Edit Project C) Delete Project D) Run Project
Answer: A, B, and D
Rationale: To view and edit an Active Learning project, users need View Project, Edit Project, and Run Project permissions. Delete Project is not required for viewing or editing.
Answer: B
Rationale: The training data source contains the documents from which the LSI model learns term relationships and concepts. These documents must be representative of the overall dataset for accurate concept modeling.
Question 33: Which operation will trigger the need to run a Full population on a conceptual index instead of an Incremental update? A) Adding new documents to the data source saved search B) Editing the extracted text of a document or adding a Regular Expression filter C) Changing the concept stop word list D) Running a concept search
Answer: B
Rationale: Changes that fundamentally alter the underlying text of documents (editing extracted text) or how text is filtered (adding regex filters) require a full rebuild. Simply adding documents can often be handled by an incremental update.
Question 34: Which warning message might appear when creating a new conceptual index? A) "Training data source is empty" B) "No documents appear in the saved search" or "Search contains fields that would cause the index to error" C) "Insufficient system memory" D) "Invalid language detected"
Answer: B
Rationale: Common warnings when creating an index include an empty saved search or the presence of fields that conflict with analytics processing (e.g., containing unsupported data types). These issues can prevent successful index creation.
Question 35: Where are analytics indexes stored when being actively processed? A) On disk B) In memory (RAM) C) In the database D) In a temporary file
Answer: B
Rationale: Analytics indexes are stored in memory (RAM) while being worked with, which allows for very fast response times. This design is critical for interactive analytics features like concept search and clustering.
Question 36: When creating multiple analytics indexes for a single workspace, what is a valid reason to do so? A) To limit search results to certain document groups or to handle multiple languages B) To improve overall system performance C) To reduce storage requirements D) To avoid using concept stop words
Answer: A
Rationale: Multiple indexes are recommended when you need to limit analytics to specific document groups or when the document set contains multiple languages. Each index can be optimized for its specific subset.
Question 37: What must you do after changing the extracted text of a document? A) Run an incremental population B) Run a full population of the conceptual index C) Restart the Relativity application D) No action needed
Answer: B
Rationale: Changes that fundamentally alter the underlying text of documents require a full rebuild of the conceptual index. Simply adding documents can often be handled by an incremental update, but altering extracted text necessitates a full population.
Question 38: What is the concept space in a conceptual index? A) The physical storage location of the index B) The mathematical model created from the training set that maps relationships between terms and documents C) The list of concept stop words applied to the index D) The set of documents excluded from the searchable set
Answer: B
Rationale: The concept space is the mathematical model created from the training set. It maps relationships between terms and documents, allowing the system to perform conceptual comparisons like concept searches and clustering.
Question 39: What is the primary difference between a conceptual index and a classification index? A) Conceptual indexes are faster to build B) Classification indexes are used solely by Active Learning to predict document relevance based on coded examples C) Conceptual indexes require more storage space D) Classification indexes cannot be used with repeated content filters
C) Only the email date D) Only the email attachments
Answer: A
Rationale: Email threading uses multiple header fields (From, To, Date, Subject) and body segments to accurately identify and group emails into complete conversation threads.
Question 44: What email relationships can be determined by email threading? A) Email threads, participants, and attachments B) Only email threads C) Only file system relationships D) Only email attachments
Answer: A
Rationale: Email threading reconstructs complete conversation chains, identifies all participants involved, and (when parent ID is provided) includes related attachments.
Question 45: Which structured analytics feature groups documents with highly similar but not identical content? A) Email threading B) Near-duplicate identification C) Language identification D) Repeated content identification
Answer: B
Rationale: Near-duplicate identification detects documents that are almost identical, allowing reviewers to efficiently handle minor variations (e.g., changed signature blocks or phone numbers).
Question 46: Which of the following best describes email threading as a structured analytics function? A) It groups conceptually similar emails regardless of header information B) It uses exact header and body matching to reconstruct email conversations C) It only groups emails with identical subject lines D) It is a conceptual analytics feature
Answer: B
Rationale: Email threading is a structured analytics function that uses deterministic matching of email headers (From, To, Date, Subject) and body segments to reconstruct complete email conversation threads.
Question 47: What is the primary distinction between Structured Analytics and Conceptual Analytics? A) Structured analytics works only on images; conceptual analytics works on text B) Structured analytics relies on exact matches, while conceptual analytics groups documents by meaning C) Structured analytics requires manual tagging; conceptual analytics does not D) Structured analytics is only available in the cloud version
Answer: B
Rationale: Structured analytics uses deterministic rules and exact matches (e.g., email threading, near- duplicate detection). Conceptual analytics uses semantic similarity to group conceptually related documents even without identical keywords.
Question 48: True or False: Near-duplicate identification is a type of structured analytics that groups documents with very similar but not identical content. A) True B) False
Answer: A
Rationale: Near-duplicate identification is a structured analytics feature that detects and groups documents with highly similar content (e.g., minor differences like a changed phone number or signature block).
Question 49: What are the two components required for setting up repeated content identification? A) Saved search and Structured analytics set B) Conceptual index and Active Learning project C) Data source and classification index D) Email threading set and near-duplicate set
Answer: A
Rationale: The setup for running repeated content identification is comprised of two components: a saved search and a structured analytics set.
Question 50: In structured analytics, how many filters can be applied at once? A) Only one filter B) Multiple filters C) Unlimited filters D) Up to 10 filters
Answer: A
Rationale: Note that in structured analytics, only one filter can be applied.
Question 55: Can clusters be moved inside the cluster pack panel? A) Yes, click and drag B) No, they are fixed C) Only by an administrator D) Only during initial setup
Answer: A
Rationale: Clusters can be moved inside the cluster pack panel by clicking and dragging. Changes can be saved with your dashboard.
Question 56: What can you submit for clustering? (Select all that apply) A) Documents returned in a saved search B) Documents located in a specific folder C) All documents in the workspace D) Only documents coded as Responsive
Answer: A, B, and C
Rationale: You can submit documents returned in a saved search, documents located in a specific folder, or all documents in the workspace for clustering.
Question 57: For categorization, the concept rank indicates: A) The distance between the example document and the resulting document B) The number of documents in the category C) The accuracy of the categorization D) The time taken to categorize
Answer: A
Rationale: For categorization, the concept rank indicates the distance between the example document and the resulting document. It helps measure how closely a document matches the example.
Question 58: What is an ideal example document for categorization? A) Focused on a single concept, at least one full paragraph, and free of distracting text/errors B) Long and detailed with multiple concepts C) Short and concise with minimal text D) Containing images and metadata
Answer: A
Rationale: An ideal example document should be focused on a single concept, contain at least one full paragraph, and be free of distracting text, errors, or metadata.
Question 59: How many example documents are recommended for creating a categorization set? A) 1 to 4 B) 5 to 20 C) 50 to 100 D) Several thousand
Answer: B
Rationale: It is recommended to tag at least 5 to 20 documents as examples for a category. This provides a sufficient basis for the categorization algorithm to identify other conceptually similar documents.
Question 60: Are numbers considered in categorization? A) Yes B) No
Answer: B
Rationale: Numbers are not considered in categorization. Only text should be used as examples, as the algorithm focuses on textual content for conceptual matching.
Section 6: Administration & Troubleshooting
Question 61: After what action must you run a full population on your conceptual index? (Select all that apply) A) Editing the extracted text of a document B) Adding a regular expression filter to the index C) Adding new documents to the data source D) Changing the concept stop word list
Answer: A and B
Rationale: Changes that fundamentally alter the underlying text of documents (editing extracted text) or how text is filtered (adding regex filters) require a full rebuild. Simply adding documents can often be handled by an incremental update.
Question 62: What is the proper order for filtering repeated content from a conceptual index? A) Run repeated content identification → Evaluate and tag desired filters → Apply filters to the Analytics Index → Click Populate Index: Full B) Apply filters to the Analytics index → Run repeated content identification → Evaluate and tag desired filters → Click Populate Index: Full C) Create conceptual index → Apply filters → Run identification → Evaluate filters D) Evaluate and tag filters → Apply filters → Run identification → Click Populate Index: Full
Answer: A
Rationale: The index remains available for queries until you explicitly disable queries and rebuild the index. This allows continuous analytics functionality during index updates.
Question 67: Are email notifications sent for manual index creation? A) Yes, if enabled, notifications will be sent for both manual and automatic creation B) No, only for automatic creation C) Yes, but only for errors D) No, never for manual creation
Answer: A
Rationale: If email notifications are enabled, they will be sent for both manual and automatic index creation, keeping administrators informed of index status.
Question 68: Does Active Learning work when you analyze the content of families? A) Yes, but it is not recommended as it can introduce bias B) No, it works only on individual document content C) Yes, families are recommended for better results D) No, families are automatically excluded
Answer: A
Rationale: Active Learning can work with family content, but it is not recommended as it can introduce bias and affect model training. The algorithm works better analyzing individual document content.
Question 69: What is the purpose of the "Concept Stop Word" list? A) It determines the words you want the conceptual index to suppress B) It defines the keywords for concept search C) It lists words to include in the index D) It controls email threading parameters
Answer: A
Rationale: The Concept Stop Words list determines the words you want the conceptual index to suppress. You can add or remove stop words from the list.
Question 70: What is the "minimum similarity percentage field" used for in clustering? A) It determines when a document is placed in a TND Group B) It sets the minimum number of documents per cluster C) It defines the maximum number of clusters D) It controls the display of cluster titles
Answer: A
Rationale: A document is placed in a TND Group when it's close enough to the Principal, as defined by the Minimum Similarity Percentage Field. This threshold determines how closely a document must match to be included in a cluster.
Section 7: Advanced Concepts
Question 71: What is the role of synchronization in Active Learning? A) The act of telling the analytics server about updates to documents coded on the Review Field B) The process of building the initial model C) The method of prioritizing documents for review D) The validation step for model accuracy
Answer: A
Rationale: Synchronization is the act of telling the analytics server about updates to documents coded on the Review Field. This ensures the model incorporates the latest coding decisions.
Question 72: What Relativity Analytics feature involves classifying documents into conceptually similar groups based on user examples? A) Clustering B) Categorization C) Active Learning D) Concept search
Answer: B
Rationale: Categorization involves classifying documents into conceptually similar groups based on user examples. It uses manually selected example documents as a basis for identifying and grouping other conceptually similar documents.
Question 73: For categorization, the concept rank indicates: A) The distance between the example document and the resulting document B) The number of documents in the category C) The accuracy of the categorization D) The time taken to categorize
Answer: A
Rationale: For categorization, the concept rank indicates the distance between the example document and the resulting document. It helps measure how closely a document matches the example.
Question 74: Only documents included in the data source are returned when you run clustering, categorization, etc. True/False.
