Inspection method, device and terminal of knowledge base

Abstract

The invention relates to a knowledge base check method and device, and a terminal. A knowledge base comprises multiple problems. The method comprises the following steps of: receiving multiple test problems and corresponding expected problems, wherein the expected problems are problems in the knowledge base; for the test problems, calculating the highest semantic similarity between each test problem and the multiple problems in the knowledge base; and, determining a check result of each test problem based on the highest semantic similarity, wherein the check result comprises whether the test problem is matched with the problem in the knowledge base or not, and whether the matched problem is the same as the expected problem corresponding to the test problem or not when the test problem is matched with the problem in the knowledge base. By means of the technical scheme in the invention, the convenience and the check efficiency for knowledge base check are improved.

Description

technical field [0001] The invention relates to the technical field of natural language processing, in particular to a knowledge base checking method, device and terminal. Background technique [0002] With the development of intelligent question answering technology, more and more platforms (for example, QQ, Skype, e-commerce customer service systems, MSN platforms, WeChat platforms, short message service platforms, etc.) are using intelligent question answering systems. Multiple knowledge points are stored in the knowledge base applied to the intelligent question answering system, and each knowledge point includes one or more preset questions and corresponding answer information. When the user asks a question by inputting request information, the semantic similarity between the request information and the preset question is obtained, and if there is a preset question whose semantic similarity is greater than the preset threshold, the answer information corresponding to the...

AI Technical Summary

Problems solved by technology

[0004] However, the existing inspection method requires repeated input by multiple testers, and only focuses on the accuracy of the output answer of the intelligent question answering system; it takes a lot of manpower and energy to analyze the reasons for the wrongly answered user questions.

Especially when the knowledge base in the intelligent question answering system changes, the above actions need to be repeated, which is very time-consuming and labor-intensive

Images