Automated User Profile Building via Encyclopedia Summaries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Organizations face difficulties in tracking and utilizing employee knowledge and expertise due to challenges in defining and maintaining ontologies, and the error-prone nature of natural language processing tools, which burden users with manual reporting and require human review.
Innovation Solution
A system that automatically builds user profiles by crawling internal web pages, segmenting text around employee names, and summarizing it using an electronic encyclopedia to form name and subject pairs, stored in a database for efficient searching and discovery of expertise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If natural language processing tools are used to discover expertise, then expertise discovery can be automated, but the tools are error prone and produce a large number of terms requiring human review
Solution Approach 1:
The patent introduces an intermediary validation mechanism where extracted expertise terms are cross-checked against a predefined ontology and taxonomic hierarchy. This intermediary layer filters and validates the output of NLP tools, reducing errors and false positives while maintaining automation. The system uses the ontology as a mediator to verify whether extracted terms are valid and meaningful expertise indicators.
Solution Approach 2:
The system implements feedback loops where the results of expertise discovery are validated against the ontology structure, and incorrect or ambiguous terms are fed back for refinement. The taxonomic hierarchy provides a feedback mechanism to ensure extracted terms align with established knowledge categories, allowing the system to learn and improve from validation results.
2Loss of information
If users manually report their skills according to an ontology, then expertise information can be collected, but users are burdened with manual reporting and updating
Solution Approach 1:
The system enables self-service by automatically extracting expertise information from various sources such as employee profiles, project documents, and performance reviews. The NLP-based extraction system allows employees to have their expertise automatically discovered and recorded without manual intervention, while the ontology ensures the information is structured and complete.
Solution Approach 2:
The ontology and taxonomic hierarchy are pre-established and prepared in advance, providing a ready-made framework for organizing expertise information. This preliminary structure allows the system to automatically categorize and store extracted expertise data without requiring users to manually define categories or structures during the reporting process.
3Adaptability or versatility
If an ontology of skills is defined and users report according to it, then expertise can be tracked, but the ontology is difficult to define and keep current
Solution Approach 1:
The system implements a dynamic ontology that can automatically adapt to new expertise domains and emerging skills. The taxonomic hierarchy is designed to be extensible, allowing new categories and subcategories to be added based on discovered expertise patterns. The system can learn from new data sources and automatically update the ontology structure to reflect current organizational needs.
Solution Approach 2:
The ontology is designed with universal categories and hierarchical structures that can accommodate multiple domains and types of expertise. By creating a flexible taxonomic framework with broad top-level categories that can be specialized, the system can track diverse expertise areas without requiring separate ontologies for different domains, reducing overall complexity.
Data Source
AI summary
Described are various embodiments which enable organizations to track and use knowledge and expertise of their associated individuals. An organization can use exemplary embodiments to automatically summarize the expertise of each individual from documents available from internal or external web sites. For example, a web crawler crawls a computer network to identify documents that name an individual. Summaries of the documents are generated based on articles in an encyclopedia, and a profile is built of the individual using the summaries. These summaries are used for automatically searching and automatically discovering individuals having particular knowledge or expertise on certain topics and subjects.