Analytics - Name Normalization
- Operation parses header data (From, To, Cc, Bcc) from every segment within an email segment using same logic as email threading
- Name Norm identifies aliases within each section, looking for semi-colon delimiters
- entities with same first and last name values are automatically merged (as well as merged with those created by Legal Hold / Processing / Case Dynamics)
- uses segment matching to infer relationships between different aliases that appear in the email headers
Name Normalization - Special Considerations
- Recommend run in its own structured analytics set
- must have at least a “From” field and one other email header field (To / CC / BCC / Subject / Date Sent); if fields don’t exist, name norm will attempt to analyze extracted text to find a From field
- if Processing / Legal Hold are installed, strongly recommend using a “Classification” value to existing entities to differentiate between them and entities created by name normalization
Name Normalization Results
Aliases - Name - Domain (full domain after @) - Primary Domain - Type [Proper Name / Email Address / Extended Email Address / Exchange / Phone Number / Undefined] - Entity - entity the alias belongs to Multiple object fields link aliases to documents: - Alias From - Alias To - Alias CC - Alias BCC - Alias Recipient - Alias Participant
Name Normalization Results - Entities
Analytics merges multiple aliases into a single entity (same first and last) - Full Name - First Name - Last Name - Classification - Aliases Multiple-Object fields link to docs: - Entity From - Entity To - Entity CC - Entity BCC - Entity Recipient - Entity Participant
Name Normalization - Documents
Analytics creates and populates the following fields on the document object:
- Alias From
- Alias To
- Alias CC
- Alias BCC
- Alias Recipient
- Alias Participant
- Entity From
- Entity To
- Entity CC
- Entity BCC
- Entity Recipient
- Entity Participant
Name Normalization - Adjusting Results
Assign to Entity - Mass Operation, lets you select and re-assign an alias to a different entity (can’t create entity on the fly)
- can select up to 50 aliases at a time
Merge - Mass operation on the entities tab that lets you select and merge multiple entities into a single entity (up to 50)
- Merge logic - all entities sorted by Artifact ID, with lowest at the top -> if entity is associated with processing data source or legal hold, moved to the top
- assigns all values to first entity on the list for multiple object / multiple choice fields
- assigns first value for each field to first entity on the list (fixed-length, long text, etc)
Name Normalization - Communication Analysis
Security -> Entity - View; Documents tab visibility; Admin Operations - Communication Analysis Widget; Item-level security - Entity From/Entity Recipient field
Nodes - each entity represented by a blue circle called a node, based on number of times entity appears in Entity From / Entity Recipient fields
Links - Gray lines that represent communication between 2 entities; width is based on amount of bidirectional comms
Actions
- Hover over node [Name/Sent/Received/Total]
- Left click [add search condition to search panel]
- Filter on doc list
- Move the visualization
- Zoom in and out
- Maximize or remove widget
Active Learning - Basic Active Learning Workflow
- Run Structured Analytics
- Remove Large/Non-Text Docs
- Remove Out of Scope Docs
- Create Saved Search
- Classification Index
- Reviewer Group and Review Field
- Pre-code documents/richness sample
- New Active Learning Project
- Turn on Prioritized Review
- Monitor Review
- Run Project Validation
Active Learning - Choosing an Active Learning review queue
Prioritized Review:
- need to quickly locate and review most relevant docs in set
- want to review relevant and family together
- have a doc set with low richness
- Unsure which queue to select
Coverage:
- Need to classify documents quickly into rel/not rel sets
- You have large case and don’t need to review and code every relevant document
Project Validation and Elusion Testing - Key Definitions
- Discard pile
- Discard-pile Elusion Rate
- Sample Elusion Rate
- Pre-coded documents in sample - docs in sample not served up; with Elusion with Recall is selected, samples across entire Active Learning Project
- Rank cutoff
Project Validation - Elusion with Recall
- Samples all documents, regardless of rank and coding status, and calculates elusion rate, richness, recall and precision
- Richness estimate helps you interpret the elusion rate
- Recall estimate often requested by receiving parties
- If run near end of the prioritized review, validation type takes roughly the same effort as Elusion Only, but gives a clearer picture of project state
