Task type dialogue state tracking method fusing slot association and semantic association

Abstract

The invention discloses a task type dialogue state tracking method fusing slot association and semantic association, which is used for fusing slot association information and mining the relationship between slots by using a graph neural network. Firstly, a multi-relation mode graph is constructed for slots, and then attention scores of a relation layer and a node layer are respectively calculated by utilizing a hierarchical graph attention network, so that an explicit or implicit relation between the slots is captured. In order to fuse the semantic relationship, the method introduces a word-level semantic similarity vector in a slot gate mechanism to obtain local semantic matching information. According to the mode, the local semantic features of the slot position and the current dialogue can be obtained, and the similarity vector is used as the enhanced feature of the slot gate mechanism, so that whether the slot position is involved in the dialogue or not is better judged, and the prediction mode of the slot value is determined. And finally, decoding different prediction modes through two sub-modules in the slot value decoder, so that the method can effectively improve the prediction precision and speed of the dialogue state.

Description

technical field [0001] The invention relates to the technical field of deep learning, in particular to a task-based dialogue state tracking method integrating slot association and semantic association. Background technique [0002] In recent years, artificial intelligence has developed rapidly, and more and more applications are inseparable from intelligent interaction. Among them, the most significant development is human-computer dialogue interaction. In layman's terms, it is a robot dialogue. Human-computer dialogue interaction can be divided into chat-type dialogue system and task-type dialogue system according to the content of dialogue. The chat-type dialogue system belongs to the open domain question and answer, and there is no specific chat field, such as Xiao Ai speakers, Microsoft Xiaobing and other voice assistants. Task-based dialogue systems are designed to help users complete one or more specific tasks, such as a ticket reservation system, an intelligent cust...

AI Technical Summary

Problems solved by technology

[0006] (1) High coding complexity

During each round of state tracking, existing methods usually encode all the dialogue history. As the number of dialogue rounds increases, the length of the dialogue context also increases, which increases the difficulty of encoding, computational complexity and memory consumption. also increased

[0007] (2) The relationship between slots is not considered

[0008] (3) Insufficient semantic correlation information

This method will lead to inefficient and time-consuming dialogue state tracking process. Not only that, the input information of the slot door mechanism is too simple, and the global semantic association information cannot be fully extracted, causing the state tracking to miss the slots mentioned in the current dialogue. bit

Images