18122691. Scalable and Resource-Efficient Knowledge-Graph Completion simplified abstract (Microsoft Technology Licensing, LLC)
Contents
Scalable and Resource-Efficient Knowledge-Graph Completion
Organization Name
Microsoft Technology Licensing, LLC
Inventor(s)
Xiaodong Liu of Woodinville WA (US)
Sanxing Chen of Durham NC (US)
Jianfeng Gao of Woodinville WA (US)
Scalable and Resource-Efficient Knowledge-Graph Completion - A simplified explanation of the abstract
This abstract first appeared for US patent application 18122691 titled 'Scalable and Resource-Efficient Knowledge-Graph Completion
The technique described in the patent application performs knowledge-graph completion in a scalable and resource-efficient manner.
- The technique identifies a source entity with a source-target relation connecting it to a target entity yet to be determined.
- It also identifies a source-entity data item containing text related to the source entity.
- A machine-trained encoder model is used to map the source-entity data item to encoded information.
- The technique predicts the identity of the target entity based on the encoded information of the source entity and the encoded information of the source-target relation.
- Prediction of the target entity may also consider neighboring entities connected to the source entity and their respective relations.
- Knowledge transfer across knowledge-graph training stages is enabled.
Potential Applications: - Enhancing knowledge graph databases - Improving search engine results - Enhancing recommendation systems
Problems Solved: - Scalability issues in knowledge-graph completion - Resource inefficiency in completing knowledge graphs - Improving accuracy of predicting target entities in knowledge graphs
Benefits: - Increased efficiency in completing knowledge graphs - Enhanced accuracy in predicting target entities - Improved performance of recommendation systems
Commercial Applications: Title: "Enhancing Knowledge Graphs for Improved Search Results" This technology can be utilized in search engines, recommendation systems, and data analytics platforms to enhance the accuracy and efficiency of information retrieval and analysis.
Questions about the technology: 1. How does the technique ensure scalability and resource efficiency in knowledge-graph completion? 2. What are the potential implications of using machine-trained encoder models in predicting target entities in knowledge graphs?
Original Abstract Submitted
A technique performs the task of knowledge-graph completion in a manner that is both scalable and resource efficient. In some implementations, the technique identifies a source entity having a source-target relation that connects the source entity to a yet-to-be-determined target entity. The technique also identifies a source-entity data item that provides a passage of source-entity text pertaining to the source entity. The technique uses a machine-trained encoder model to map the source-entity data item to source-entity encoded information. The technique then predicts an identity of the target entity based on the source-entity encoded information, and based on predicate encoded information that encodes the source-target relation. In some implementations, the technique also predicts the target entity based on a consideration of one or more neighboring entities that are connected to the source entity and their respective source-to-neighbor relations. The technique further allows transfer of knowledge across knowledge-graph training stages.
