Automated Document Graphing for Legal E-Discovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Lawyers and professionals face challenges in evaluating the similarity of numerous documents manually, making it impractical during tasks like litigation e-discovery and legal obligation mapping due to the sheer volume of documents.
Innovation Solution
Automated document graphing systems and methods that involve text extraction, cleaning, lemma conversion, stop word removal, and vector representation, allowing for the calculation of cosine similarity and visualization of document relationships, facilitated by a knowledge graph and distributed analytics engine.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual document review is used to evaluate document similarity, then evaluation accuracy can be maintained, but the time required and productivity deteriorate due to the sheer number of documents
Solution Approach 1:
The patent replaces manual mechanical review with automated computational systems. Text extraction, cleaning, lemma conversion, and vector representation algorithms automatically process documents, while cosine similarity calculations and machine learning models evaluate document relationships without human intervention, maintaining accuracy while dramatically increasing productivity
Solution Approach 2:
The patent introduces intermediate processing steps including text extraction, cleaning, lemma conversion, and vector representation as mediators between raw documents and similarity evaluation. These intermediary transformations convert unstructured text into standardized vector formats that enable efficient automated comparison while preserving semantic meaning
2Measurement precision
If text preprocessing steps (cleaning, lemma conversion, stop word removal) are applied, then vector representation accuracy improves, but processing time and computational complexity increase
Solution Approach 1:
The patent applies preliminary text preprocessing actions including cleaning, lemma conversion, and stop word removal before vector representation. These preparatory steps organize and standardize the text data in advance, improving the quality and consistency of vector representations while enabling more efficient subsequent processing through structured input
3Loss of information
If cosine similarity calculation is performed on all document pairs, then comprehensive document relationship analysis is achieved, but computational complexity and processing time increase exponentially
Solution Approach 1:
The patent segments the computational task by first calculating cosine similarities for individual document pairs, then using these results to build a document graph structure. This segmentation allows the system to process documents in manageable units and reconstruct comprehensive relationships through graph algorithms rather than requiring exhaustive pairwise comparison of all documents simultaneously
Data Source
AI summary
Systems and methods for automated document graphing are disclosed. In one embodiment, in an information processing apparatus comprising at least one computer processor, a method for automated document graphing may include: (1) receiving, from a document repository, a plurality of source documents; (2) extracting text from each of the plurality of source documents; (3) cleaning the extracted text by removing unnecessary characters from the extracted text; (4) rewriting multi-word proper nouns in the extracted text; (5) converting a plurality of words in the extracted text to lemma form; (6) rewriting compound terms in the extracted text; (7) removing stop words from the extracted text; (8) representing each of the remaining words in the extracted text as a vector; and (9) storing the vectors in a library.