An intelligent graph-based expert searching system

Simple SummaryContent extracted from patent full text and abstract with AI.

This patent describes an intelligent, ontology-driven system for automatically connecting users seeking knowledge with suitable experts in real time. The core method models areas of expertise and user queries as vectors in a graph-based hierarchical structure (ontology tree), and uses similarity algorithms to match questions to relevant experts efficiently. The system supports dynamic, scalable, and accurate expert retrieval, making it particularly suitable for web-based communities or corporate intranets.

Use CasesContent extracted from patent full text and abstract with AI.

• Enterprise knowledge management systems that allow employees to find internal experts on specific topics quickly.
• Online Q&A platforms or forums that want to automate expert assignment for user-submitted questions.
• Customer support systems that intelligently route queries to the most qualified support agent or specialist.
• Academic or research networks connecting students and researchers with subject matter experts for mentoring or collaboration.
• Healthcare or telemedicine services directing patient inquiries to the most relevant doctors or medical specialists.
• Community-based help desks or IT support systems in large organizations.

BenefitsContent extracted from patent full text and abstract with AI.

• Improves speed and accuracy in finding the right expert for a question, reducing search time and effort.
• Scalable to large organizations or online communities with many users and fields of expertise.
• Dynamic: continuously updates expert profiles and knowledge fields as user activities and expertise evolve.
• Reduces dependence on static keyword lists or manual routing, utilizing semantic relationships within the ontology structure.
• Allows for real-time, direct communication between question posers and selected experts, enhancing user experience.
• Efficient in storage and computation by representing expertise and queries with compact vectors rather than large unstructured texts.
• Robust against sparse or incomplete queries due to hierarchical expert mapping.

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

A query-answering system (100) and an ontology-based method for automatically peering knowledge seekers with knowledgeable experts (103) has been proposed. It comprises an intelligent, real-time expert searching engine (116) for web clients (101, 102, 103) that can effortlessly be deployed in enterprise Intranets or general interest online communities. A graph-based representation scheme has been proposed that comprises an ontology tree (10) wherein each node (20 to 70n) corresponds to a topic or subtcpic and is associated with a bag of words. This allows to represent the fields of expertise of the participating experts as well as incoming queries as subtrees of the ontology tree (10). Subsequently, we have addressed the resulting graph matching problem of assigning queries to relevant experts on a vector space, which follows from a serialization of the ontology tree, using simple dot products of respective vectors.