Relational reasoning method and system based on dependency graph

Abstract

The invention provides a relation reasoning method and system based on a dependency graph. The relation reasoning method comprises the following steps: performing word division and word feature construction on a given sentence pair by using word meaning features; obtaining a dependency relationship tree between words extracted from the text after word division through a dependency extractor; taking the dependency relationship as a basis for word feature updating, and learning and updating word features in a given sentence pair in combination with a deep learning network; taking the plurality of updated word features obtained by the given sentence pair as local features, and performing feature fusion to obtain global features; performing interaction between two sentences by taking the global features as sentence meaning features, inputting the sentence meaning features into an output layer to obtain output, comparing the output with a real label, and calculating a loss function of a learning model; and correcting the learning model according to a loss function calculation result of the learning model, and determining a target parameter corresponding to the learning model. And the expression of the syntactic dependency tree on natural language reasoning is effectively improved.

Description

technical field [0001] The present invention relates to the field of computers, in particular to a method and system for relational reasoning based on a dependency graph. Background technique [0002] With the continuous development of deep learning models, the main trend of natural language reasoning tasks is to use more complex network models to obtain the semantic information of sentences and to determine the relationship between them. However, in the traditional network, the location information comes from the direct capture of the structure of the long-term and short-term network, which can only obtain the location information in space, and cannot capture the relative location information in the deep semantics. The deep learning network that operates directly on the syntactic dependency tree is usually based on a complex tree neural network, whose training process is slow, and usually cannot combine information other than child nodes. [0003] Therefore, the Chinese pa...

AI Technical Summary

Problems solved by technology

However, in the traditional network, the location information comes from the direct capture of the structure by the long-term and short-term network, which can only obtain the spatial location information, and cannot capture the relative location information in the deep sentence sense.

The deep learning network that operates directly on the syntactic dependency tree is usually based on a complex tree neural network, whose training process is slow, and usually cannot combine information other than child nodes.

[0003] Therefore, the Chinese patent of the existing patent document CN109902301A discloses a relational reasoning method based on a deep neural network. The feature based on the grammatical dependency tree used in this method only utilizes the dependency feature on the path, and there is no effective combination Dependency type in syntax dependency tree, extraction features are not sufficient, not specific

Although LSTM and CNN are used to capture the features on the path, some of the off-path information cannot be captured, which is not accurate enough

Images