Knowledge graph representation learning-oriented distributed framework construction method

Abstract

The invention discloses a knowledge graph representation learning-oriented distributed framework construction method, which comprises the following steps of: performing data mapping of RDF triples, establishing hash mapping from original character string type data to integer type IDs, and enabling the mapped RDF triples to participate in subsequent processing; scrambling and segmenting the RDF triples according to the computing capacity of each computing group and distributing the scrambled and segmented RDF triples to each computing group; secondly, randomly generating a group of vectors as an initialized representation model, logically segmenting the representation model, sending a logical segmentation result to each calculation group, and then constructing a local sub-model by each calculation node of the calculation group according to the logical segmentation result; adopting a self-adaptive method, and enabling each computing node to automatically complete collection of required representation model vectors on related computing nodes according to a random sampling result of each round; and carrying out distributed model training and distributed model aggregation.

Description

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine

Login to View More

AI Technical Summary

Benefits of technology

The technical feature described in this patented describes methods called Distributed Computational Librations (DCL) or Structured Query Language (SQL). These techniques allow multiple computers connected together without overloading their storage capacity. They enable efficient execution across different computer systems while keeping them up-to-date about what they were doing previously. Additionally, these processes involve storing complex representations into smaller spaces instead of relying heavily upon mainframe servers. Overall, DCLI helps cram thousands of people working collaboratively towards big problem addressed in this research field.

Problems solved by technology

Technological Problem addressed in this patents relates to improving the efficiency and scalable performance of collaborative systems with big world dimensions. Current methods require expensive hardware resources and slow down computation times due to their size limitations. However, these techniques have drawbacks like lack of continuum growth capabilities and instabilities caused by factors like storage capacity limits.

Images