Email Classification via Entity Data Objects and Token Extraction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing e-discovery methods are inefficient in identifying potentially privileged communications, as they rely on incomplete attorney lists and fail to capture communications involving legal advice discussed among employees without the attorney's metadata presence.

Innovation Solution

A system that processes electronic communications to identify potentially privileged documents by extracting entities, generating entity data objects, and using a machine learning model to categorize these objects, along with token extraction and joint distance calculation to accurately classify emails.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional e-discovery methods use incomplete attorney lists for identification, then the search process is simpler and faster, but the accuracy of identifying privileged communications deteriorates

Engineering Contradiction:

Improveaccuracy of identifying privileged communicationsVSAvoidcomplexity of identification system

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary machine learning model that acts as a mediator between the email database and the identification process. This model is trained on curated datasets containing entity relationships and communication patterns, enabling it to indirectly identify privileged communications through learned patterns rather than direct attorney list matching, thereby improving accuracy without requiring complete attorney lists

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system performs preliminary actions by pre-training the machine learning model on curated datasets that contain entity relationships, communication patterns, and privileged communication examples before actual e-discovery operations. This preliminary training enables the model to automatically identify privileged communications during runtime without requiring manual attorney list updates or complex configuration

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If manual review methods are used to ensure complete identification of privileged communications, then accuracy improves, but the time and resources required increase significantly

Engineering Contradiction:

Improvecompleteness of privilege identificationVSAvoidtime for recursive searches

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The machine learning model performs self-service by automatically identifying privileged communications through patterns learned during training. The model independently processes emails, extracts entities, calculates joint distances, and categorizes communications without requiring manual review or recursive searching, thereby achieving complete identification while eliminating time losses associated with manual methods

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual review process with an automated machine learning system. The model substitutes human reviewers by performing entity extraction, relationship analysis, and classification tasks algorithmically, achieving both complete identification and time efficiency simultaneously

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If traditional methods rely on attorney metadata presence, then the classification process is simpler, but it fails to capture communications involving legal advice discussed among employees

Engineering Contradiction:

Improveability to capture employee discussions about legal adviceVSAvoidcomplexity of entity extraction and categorization

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The machine learning model achieves universality by being capable of identifying multiple types of privileged communications beyond those involving direct attorney metadata. The model can detect privileged communications between employees discussing legal advice, between attorneys and clients, and other privileged scenarios through learned patterns in entity relationships and communication contexts, making the system adaptable to various privilege types without requiring separate classification mechanisms for each

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250029068A1Automatic document classification

Publication Date: 2025.01.23 RELATIVITY ODA LLC

AI summary

A method to automatically classify emails may include generating multiple entity data objects using entities identified in fields of emails and categorizing the multiple entity data objects into a first set of data objects and a second set of data objects. The method may also include extracting all tokens from each email and searching the extracted tokens for tokens associated with the data objects of the first set of data objects. The method may further include identifying the emails that include the extracted tokens that are associated with the data objects of the first set of data objects, identifying a particular data object of the first set of data objects to which an identified email corresponds, and automatically classifying the identified email in the first category in response to identifying the particular data object of the first set of data objects to which an identified email corresponds.