Electronic Chat Splitting with Gaussian Mixture Models for Precise E-Discovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing e-discovery systems inefficiently process electronic chat data by treating entire chats as single documents, leading to inefficient manual review and inaccurate machine learning classification due to chats covering multiple topics, resulting in unnecessary processing and review of irrelevant content.
Innovation Solution
Adaptive splitting of electronic chats into conversations based on time gaps and Gaussian mixture models, allowing for separate processing and indexing of chat messages, reducing irrelevant content processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If entire electronic chats are treated as single documents for e-discovery, then the document paradigm can be maintained for simplicity, but the processing efficiency and relevance accuracy deteriorate due to chats covering multiple topics
Solution Approach 1:
The patent segments electronic chats into multiple conversations by detecting time gaps between messages. When a time gap exceeds a threshold, the system splits the chat into separate conversations, each treated as an independent document unit. This segmentation allows the document paradigm to be maintained while improving processing efficiency, as each conversation can be reviewed and processed independently rather than requiring review of entire multi-topic chats.
2Device complexity
If entire electronic chats are indexed as single documents, then indexing simplicity is maintained, but search precision deteriorates because chats may contain only a few relevant messages among many irrelevant ones
Solution Approach 1:
The system segments chats into conversations based on time gaps, creating smaller indexed units. Each conversation is indexed separately, allowing search results to be more precise. When users search for keywords, only relevant conversations need to be reviewed rather than entire chats, improving search precision while maintaining reasonable indexing complexity.
Solution Approach 2:
The system extracts and separates relevant conversations from irrelevant content within chats. By identifying time gaps that indicate topic changes or pauses, the system extracts only the relevant portions (conversations) for indexing and search, rather than indexing entire chats, thus improving search precision.
3Extent of automation
If machine learning classifiers process entire electronic chat documents, then classification can be performed at the document level, but classification accuracy deteriorates because chats covering multiple topics cannot be accurately classified
Solution Approach 1:
The patent segments chats into conversations that can be processed as separate document units for machine learning classification. Each conversation is classified independently, improving accuracy by ensuring that single-topic or single-conversation units are classified rather than multi-topic chats. The segmented conversations maintain document-level processing capability while achieving more accurate classification results.
4Reliability
If manual review of electronic chats is required, then complete content can be examined, but review time increases significantly due to the need to review entire chats even when only a few messages are relevant
Solution Approach 1:
The system segments chats into conversations based on time gaps, creating manageable units for manual review. Reviewers can focus on specific conversations rather than reading entire multi-topic chats, significantly reducing review time while maintaining the ability to examine complete content within each conversation unit. The segmentation allows selective review of relevant portions only.
Data Source
AI summary
Systems, methods, and computer program products for adaptively splitting electronic chats are provided. One embodiment includes receiving, by an electronic discovery system, an electronic chat comprising a set of electronic chat messages, each of the electronic chat messages in the set of electronic chat messages having a timestamp; determining a set of time gaps between the electronic chat messages from the set of electronic chat messages, based on selecting a Gaussian mixture model as a model of the time gaps, splitting the set of electronic chat message into a set of conversations based on the Gaussian mixture model; performing a text analysis on the set of conversations based on a chat subject matter identified in the set of electronic chat messages; and splitting the set of conversations based on the chat subject matter.