Topic-Model Privilege Ranking for E-Discovery Documents
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional e-discovery methods for identifying privileged documents are inaccurate, inconsistent, time-consuming, and inefficient due to reliance on keyword and name searches, which fail to identify privileged communications without explicit mentions and require recursive reviews.
Innovation Solution
A computer-implemented method using a machine learning topic model to analyze documents by extracting spans, generating topics and scores, and ranking documents based on their potential privilege, adjusting scores based on legal entities and communication networks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If keyword and name searches are used to identify privileged documents, then the review process can be performed using simple search methods, but the accuracy of privilege review deteriorates because searches may fail to identify items that include legal discussion without naming an attorney
Solution Approach 1:
The patent replaces mechanical keyword/name search methods with a machine learning topic model that automatically identifies legal discussions and privileged communications without requiring explicit attorney names or predefined keywords. The system uses probabilistic topic modeling to detect legal concepts and privilege indicators throughout the document corpus.
Solution Approach 2:
The system changes the search parameters from fixed keywords and names to dynamic topic probabilities generated by the machine learning model. By calculating privilege scores based on topic distributions rather than simple keyword matching, the system achieves both ease of operation and improved accuracy.
2Ease of operation
If different reviewers perform keyword searches with varying keywords or names, then individual reviewer flexibility is maintained, but consistency of review deteriorates as different reviewers may have different opinions on whether a document is privileged
Solution Approach 1:
The machine learning topic model serves as a universal review system that applies consistent privilege detection criteria across all documents and reviewers. The model's probabilistic framework provides a standardized method for identifying privileged communications, eliminating reviewer-specific variations while maintaining operational flexibility through automated scoring and ranking.
3Measurement precision
If quality control processes are applied to identify new keywords or names, then review quality may be improved, but the process becomes recursive and time-consuming requiring repeated reviews of the entirety of electronic data
Solution Approach 1:
The system performs preliminary topic modeling and privilege score calculation on the entire document corpus in advance, before any quality control review. This pre-processing creates a ranked list of potentially privileged documents, allowing reviewers to focus only on high-probability candidates rather than performing repeated full-corpus searches when new keywords are identified.
Solution Approach 2:
The machine learning model automatically adapts to new keywords and concepts through its probabilistic framework, eliminating the need for manual quality control iterations. The system self-adjusts by recalculating topic probabilities based on newly identified legal terminology, maintaining high review quality without requiring recursive human review of entire datasets.
4Reliability
If all documents identified as potentially privileged are reviewed without additional guidance, then comprehensive review is performed, but efficiency deteriorates due to lack of prioritization
Solution Approach 1:
The system segments the set of potentially privileged documents into ranked categories based on privilege scores derived from topic model probabilities. By dividing the review workload into priority tiers (high, medium, low confidence), the system maintains comprehensive review coverage while enabling reviewers to efficiently process high-priority documents first, improving overall productivity without sacrificing completeness.
Data Source
AI summary
A method for identifying and ranking potentially privileged documents using a machine learning topic model may include receiving a set of documents. The method may also include, for each of two or more documents in the set of documents, extracting a set of spans from the document, generating, using a machine learning topic model, a set of topics and a subset of legal topics for the set of spans, generating a vector of probabilities for each span with a probability being assigned to each topic in the set of topics for the span, assigning a score to one or more spans in the set of spans by summing the probabilities in the vector that are assigned to a topic in the subset of legal topics, and assigning a score to the document. The method may further include ranking the two or more documents by their assigned scores.