Self-Training Parser Using Document Metadata for E-Discovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for training communication document parsers require significant manual review and annotation, which is time-consuming and inefficient, especially when dealing with large volumes of electronic communication documents during processes like eDiscovery.
Innovation Solution
A self-training method and system that utilize metadata from electronic communication documents to annotate and re-train parsers, reducing the need for manual annotations by using metadata as truth to identify entities and improve parsing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual annotation is used to train the parser, then parsing accuracy is improved, but the time and resources required for training increase significantly
Solution Approach 1:
The system performs preliminary actions by automatically generating annotations from metadata before manual review. The metadata extraction and initial annotation processes are executed in advance, creating a pre-annotated dataset that reduces the time required for manual annotation while maintaining sufficient training quality.
Solution Approach 2:
The parser system performs self-service by automatically training itself using metadata from the documents it processes. The system extracts its own training data from document metadata, generates initial annotations, and iteratively improves its own performance without requiring extensive external manual annotation resources.
2Measurement precision
If manual annotation is used to train the parser, then parsing accuracy is improved, but the cost and complexity of the training process increase
Solution Approach 1:
The system automates the training process by using its own processed document metadata as training data. The parser independently extracts features, generates annotations, and retrains itself, eliminating the need for complex manual annotation workflows and reducing overall system complexity.
Solution Approach 2:
The system implements multi-functionality by using the same metadata extraction and processing pipelines for both document analysis and parser training. The metadata that serves the primary function of document characterization also serves the secondary function of training data generation, reducing the need for separate training infrastructure.
3Reliability
If extensive manual review is conducted to generate training annotations, then parser reliability is improved, but productivity decreases
Solution Approach 1:
The parser achieves self-service by automatically generating training annotations from metadata and iteratively improving its own reliability. This self-training mechanism maintains parser reliability while eliminating the bottleneck of manual review, thereby preserving high document processing throughput.
Solution Approach 2:
The system implements feedback loops where the parser's performance on processed documents feeds back into automated retraining using metadata from those same documents. This continuous feedback mechanism improves parser reliability over time without interrupting the document processing workflow, maintaining high productivity.
4Measurement precision
If the parser is re-trained frequently with new data, then parsing accuracy is improved, but the computational resources and time required increase
Solution Approach 1:
The system applies partial retraining by using only the metadata from recently processed documents as training data, rather than retraining on the entire corpus. This selective approach maintains parsing accuracy while significantly reducing computational resources and time required compared to full retraining cycles.
Solution Approach 2:
The system performs preliminary feature extraction and annotation generation from metadata before the actual retraining process. By preparing training data in advance during document processing, the system reduces the computational burden during retraining and enables more frequent, efficient updates to parsing accuracy.
Data Source
AI summary
Systems, methods, and computer readable media for self-training a parser of electronic communication documents, such as emails, are provided. These techniques may include applying a parser to a batch of electronic communication documents to identify entities included in unstructured text. The outputs of the parser are used to identify entries in a metadata file associated with the electronic communication documents to generate training data for the parser. The parser is then re-trained using the training data and applied to an additional batch of documents. Through this process, the systems, methods, and computer readable media are able to re-train the parser without obtaining manual annotations of electronic communication documents.