Beyond Predictive Coding – The True Power of Analytics
Panelists: Hal Marcus, eDiscovery Attorney working with Product Marketing team, Recommind Susan Stone, Senior Solutions Consultant, Advanced Discovery Moderator: Stephen Dooley, Senior Manager - Electronic Discovery & Litigation Support, Sullivan & Cromwell LLP Introductions
Researching your Data Clustering and Organizing Analysis QC and Redactions Categorization Overview
Researching Your Data
Search with text from anywhere: Excerpt from Wikipedia News Media Articles Documents from client Concept Searching
Concept Browsing
Phrase Analysis
Use for internal investigations, especially with fraud components Apply when keyword searches are not yielding much information Target documents of interest and then categorize to find additional documents Researching your Data - Recap
Clustering/Organizing
Clustering Review and filter by like concepts
Organizing with Smart Filters Filter and analyze. Then save into “Review Universes”
Group documents intelligently based on a range of criteria Overlay search, metadata, analytics Organize for priority treatment and/or discrete review workflows Clustering and Organizing - Recap
Analysis
Threading
“End of Branch”
Hypergraph Analysis
Reduce the volume of review substantially Ensure consistency in coding (view and/or bulk code as desired) Identify possibly missing components Analysis – Recap
QC and Redactions
Identify Textual Near Duplicates
Textual near Duplicate Compare
Consistent Coding
Smart Redactions
Global Smart Redactions
Textual Near Duplicates Identify near duplicates Review changes Provide consistent coding Smart Redactions Identify PII, PCI, PHI Redacts regular expressions Redact full phrases/sentences Use colors and reasons for QC QC and Redactions – Recap
Categorization
Coding just 5 documents able to categorize hundreds
Docs with priv search terms Docs suggested by machine learning Relevant and privileged 2nd: Use relevant priv docs to train for privilege - and find anything missed by keyword searches Total document corpus Help identify potentially privileged docs 1st: Prioritize your review of docs hitting on priv search terms
Use Clustering to identify materials provided by opposition Use categorization on documents from your review set to identify hot materials from opposition documents. Where machine learning overlaps with search and/or other analytics, you have key documents to review Categorization – Recap