Neural Machine Translation for Yang Model XPath Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manual mapping process between different Yang models in network devices is time-consuming and prone to inaccuracies due to the need for domain knowledge and the complexity of network modules with thousands of attributes, such as the Openconfig BGP model, which has over 1,000 unique attributes, making it difficult for vendors to implement CRUD operations via North-bound interfaces.
Innovation Solution
A neural machine translation (NMT) model is employed to automatically generate mappings between Yang models by treating the path translation as a machine translation task, using an encoder-decoder architecture with attention mechanisms and a Sequence-to-Sequence (Seq2Seq) model to convert external Yang XPath to internal Yang XPath, reducing the need for manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual mapping is used to transform XPath between Yang models, then domain knowledge and expertise can be applied to ensure accuracy, but the process becomes extremely time-consuming and labor-intensive
Solution Approach 1:
The patent replaces the manual mechanical mapping process with an automated neural machine translation system. The NMT model learns mapping patterns from training data and automatically transforms XPath expressions between different Yang models, eliminating the need for manual domain expert intervention while maintaining high accuracy through learned relationships from extensive training examples
Solution Approach 2:
The patent performs preliminary training of the NMT model using labeled mapping examples between source and target Yang models. This pre-training phase allows the system to learn transformation patterns in advance, so that during actual operation, the mapping can be performed automatically without requiring manual domain knowledge, thus resolving the time-consuming issue while preserving accuracy through pre-learned expertise
2Measurement precision
If manual mapping is performed by domain experts, then accuracy can be maintained, but the complexity and cost of the process increases due to the need for specialized knowledge
Solution Approach 1:
The patent substitutes the complex manual process requiring domain experts with a trained NMT system. The complexity of acquiring and maintaining domain knowledge is transferred to the training phase, where the model learns from labeled examples. Once trained, the system performs mappings automatically without requiring ongoing domain expert involvement, thus reducing operational complexity while maintaining accuracy
Solution Approach 2:
The patent creates a learned representation (copy) of the mapping knowledge from training data. Instead of relying on human experts to repeatedly apply domain knowledge, the NMT model creates internal representations of mapping relationships that can be automatically applied to new XPath transformation tasks, reducing the complexity of the operational process
3Productivity
If automated mapping is implemented without domain knowledge, then the process becomes faster and more scalable, but accuracy and reliability of the mappings deteriorate
Solution Approach 1:
The patent performs preliminary training with labeled mapping examples to teach the NMT system the correct transformations. This pre-learning phase enables the automated system to achieve high accuracy by incorporating domain knowledge into the training data, so that subsequent automated mappings maintain both speed and precision without requiring ongoing human intervention
Solution Approach 2:
The patent uses labeled mapping examples as training feedback to teach the NMT system correct transformations. The model learns from these feedback examples during training, adjusting its parameters to minimize errors. This feedback mechanism enables the automated system to achieve high accuracy comparable to manual expert mapping while maintaining automated throughput
Data Source
AI summary
Described herein are systems and methods for neural machine translation (NMT) of languages between network operation systems. The languages may be a query language, such as a XML path, to navigate through elements and attributes in an XML document. An NMT model comprises an encoder and a decoder to implement the machine translation. The encoder encodes a source sentence as a sequence of encoder hidden states. The decoder may incorporate attention mechanism to generate a target sentence, conditioned on the encoder hidden states. The NMT model may also use a modified beam search with variable beam width and search scope for each search step to speed up search process with a balance of accuracy and processing cost. Evaluation results demonstrate that embodiments of the present disclosure may be used in a recommender system for XPath auto-generation between different network operation systems.