Enterprise Knowledge Graph Mining with Neural Entity Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in creating accurate and up-to-date knowledge graphs for enterprises due to unique vocabularies, private project names, and non-standard use of words, leading to inaccurate search results and inefficient information retrieval, which hampers productivity and requires significant manual effort.
Innovation Solution
The implementation of a multi-toolkit enterprise mining system that uses neural entity recognition and natural language processing to mine and annotate enterprise knowledge graphs, incorporating multiple AI toolkits to specialize in different entity types and sources, such as emails, OneDrive, and external data, to generate a comprehensive and accurate knowledge graph.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional knowledge graph systems are used, then information retrieval can be performed, but accuracy and completeness deteriorate due to unique enterprise vocabularies, private project names, and non-standard word usage
Solution Approach 1:
The system implements user-based mining where each user's content (emails, OneDrive files, etc.) is processed individually to extract entity mentions and relationships specific to that user's context. This local processing allows the knowledge graph to adapt to unique enterprise vocabularies and non-standard word usage while maintaining high search accuracy for each user's specific needs
Solution Approach 2:
The knowledge graph is segmented into user-specific subgraphs, each containing entities and relationships relevant to that particular user. This segmentation allows the system to handle diverse enterprise contexts independently, improving both accuracy for individual users and adaptability across the entire enterprise
2Loss of information
If manual knowledge graph creation is performed, then completeness can be improved, but productivity deteriorates due to significant manual effort required
Solution Approach 1:
The system automatically mines entity mentions and relationships from user content without requiring manual input. Users simply provide their existing content (emails, documents, etc.), and the system self-services by extracting structured knowledge graph data, achieving both completeness and high productivity
Solution Approach 2:
Manual knowledge graph creation processes are replaced with automated AI-based entity recognition and relationship extraction systems. The mechanical process of manually curating knowledge graphs is substituted with intelligent automated mining from unstructured user content
3Adaptability or versatility
If multiple AI toolkits are implemented, then coverage and flexibility improve, but system complexity increases
Solution Approach 1:
The system implements a universal toolkit architecture where a single framework can process multiple entity types (people, projects, products, etc.) and multiple content sources (emails, OneDrive, external data). This multi-functional design provides broad coverage and flexibility while managing complexity through unified processing logic
Data Source
AI summary
Examples described herein generally relate to a computer system including a knowledge graph storing a plurality of entities. A mining of a set of enterprise source documents within an enterprise intranet is performed, by a user-based mining system, to determine a plurality of entity names. An entity record is generated within a knowledge graph for a mined entity name from the linked entity names based on an entity schema and ones of the set of enterprise source documents associated with the mined entity name. The entity record includes attributes aggregated from the ones of the set of enterprise source documents associated with the mined entity name.