Corpus collecting method and device and computer equipment

Abstract

The invention provides a corpus collecting method and device and computer equipment. The corpus collecting method comprises the following steps: receiving a query request input by a user; showing templated response for the user, wherein the templated response comprises a feedback portion and a guiding portion, the feedback portion comprises result feedback to a queried text input by the user, andthe guiding portion comprises guidance of the user to a next input; and receiving corpuses input by the user. A real dialogue scenario can be shown for a corpus collecting person, and collection difficulty of task-guided multiple corpuses is reduced by balance between efficiency and validity of conversation.

Description

technical field [0001] The present application relates to the technical field of man-machine dialogue, in particular to a method, device and computer equipment for collecting corpus. Background technique [0002] Now, there are more and more scenarios for task-oriented dialogue understanding and interactive applications, but most of the scenarios only support a single round of dialogue. The technical implementation of multi-round dialogue understanding and interaction is much more difficult than single-round dialogue. The primary reason is that the acquisition of corpus for multi-round dialogue is much more difficult than single-round dialogue. The corpus of a single round of dialogue can be obtained through simple and direct enrichment by relevant personnel who are familiar with the business, but due to the interactive process of multiple rounds of dialogue, it cannot be enriched out of thin air, so it is more difficult to obtain, which directly leads to the loss of Techno...

AI Technical Summary

Problems solved by technology

[0004] However, the above-mentioned corpus collection method is far away from the real scene, and it is easy to fall into personal thinking and language stereotypes, resulting in poor authenticity of the corpus, and the need to imagine the scene by yourself, which is inefficient; invest a lot of manpower to develop a complete dialogue understanding and interaction system, and the labor cost is high. , the development cycle is long, and it cannot meet the rapidly developing technology and product requirements. Each module needs to be developed and tuned separately, but the tuning of individual modules also needs corpus support; in the absence of corpus, it is very difficult to optimize the effect of each module , the effect is difficult to achieve the ideal

Images