Privileged Document Identification via Automated Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The current manual review of documents for identifying privileged information in litigation is time-consuming, costly, and prone to inconsistencies and errors, leading to potential legal and financial risks.
Innovation Solution
A system and method for identifying potentially privileged documents by comparing document attributes against a datastore containing predefined and user-specific indicators, allowing for automated flagging and segregation of documents, reducing the need for full human review and minimizing errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual review of documents is performed to identify privileged information, then accuracy and reliability can be maintained, but time consumption and cost increase significantly
Solution Approach 1:
The document review process is segmented into multiple stages: automated preliminary screening using machine learning models to identify potentially privileged documents, followed by selective human review only of documents flagged by the automated system. This segmentation allows the bulk of documents to be processed automatically while maintaining human oversight for critical decisions.
Solution Approach 2:
An automated machine learning-based intermediary system is introduced between the document corpus and human reviewers. This intermediary performs preliminary classification and prioritization, acting as a filter that selects only the most relevant documents for human review, thereby reducing the overall time and cost while maintaining reliability.
2Reliability
If manual review of documents is performed to identify privileged information, then accuracy can be maintained, but cost increases significantly
Solution Approach 1:
The review process divides documents into categories: those automatically identified as non-privileged (which can be discarded or processed differently) and those flagged as potentially privileged (which require human review). This segmentation reduces the quantity of documents requiring expensive human review while maintaining accuracy through selective oversight.
Solution Approach 2:
The machine learning system performs self-service by autonomously analyzing document content, metadata, and patterns to automatically identify and classify privileged documents without requiring continuous human intervention. This self-service capability significantly reduces the cost of document review while maintaining reliable identification through the system's trained models.
3Productivity
If multiple human reviewers work on documents, then comprehensive coverage can be achieved, but consistency and accuracy decrease due to subjective judgments and fatigue
Solution Approach 1:
The system changes the fundamental parameter of review from human-based to machine-based. Machine learning models provide consistent, objective classification decisions that do not vary based on reviewer fatigue, subjectivity, or experience levels. The model applies the same criteria uniformly across all documents, ensuring measurement precision and consistency while maintaining high productivity through automated processing.
4Loss of time
If automated systems are used to identify privileged documents, then time and cost are reduced, but reliability may be compromised
Solution Approach 1:
The system incorporates feedback mechanisms where human reviewers verify and correct automated classifications. The outcomes of these human reviews are fed back into the machine learning model to continuously improve its accuracy. This feedback loop ensures that automated processing maintains high reliability while achieving the time and cost reductions associated with automation.
Data Source
AI summary
A litigation discovery document system is provided to help identify documents that might be privileged. In particular, a system is described in which documents within a document set are compared against one or more data sources which store data that indicate whether a particular document is privileged or potentially privileged. Human reviewers can confirm whether such determinations are accurate. Analytic reports can be provided which characterize the underlying methodology used to make such determinations. Related apparatus, systems, techniques and articles are also described.


