US 2022/0012433 A1 · AUTO TRANSFORMATION OF NETWORK DATA MODELS USING NEURAL MACHINE TRANSLATION

Abstract

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.

Claims20

Dependent claims

A computer-implemented method for neural machine translation comprising:

encoding, using an encoder, a source expression into a sequence of encoder hidden states, the source expression comprising one or more words;
generating, using a decoder, decoder hidden states conditioned at least on the sequence of encoder hidden states, the decoder uses beam search comprising multiple search steps for target words, the beam search has a variable beam width and a variable search scope within a target word vocabulary during the multiple search steps;
producing, from an output prediction layer, a probability distribution among target words searched for each of the multiple search steps;
and
generating, using the probability distribution for each of the multiple search steps, one or more target expressions corresponding to the source expression after the multiple search steps, each of one or more the target expressions comprising one or more target words.

2 more elements

A non-transitory computer-readable medium or media comprising one or more sequences of instructions which, when executed by at least one processor, causes steps for neural machine translation be performed comprising:

encoding, using an encoder, a source expression into a sequence of encoder hidden states, the source expression comprising one or more words;
generating, using a decoder, decoder hidden states conditioned at least on the sequence of encoder hidden states, the decoder uses beam search comprising multiple search steps for target words, the beam search has a variable beam width and a variable search scope within a target word vocabulary during the multiple search steps;
producing, from an output prediction layer, a probability distribution among target words searched for each of the multiple search steps;
and
generating, using the probability distribution for each of the multiple search steps, one or more target expressions corresponding to the source expression after the multiple search steps, each of one or more the target expressions comprising one or more target words.

2 more elements

Dependencies are inferred from the claim text.

Auto transformation of network data models using neural machine translation

Life of this patent

Auto transformation of network data models using neural machine translation

Life of this patent

Abstract

Claims20