Command recognition method based on keywords and Word2Vec

Abstract

The invention relates to a command recognition method based on keywords and Word2Vec. The method comprises the following steps: 1) acquiring a command text; 2) performing syntactic analysis on the command text, and extracting keywords in the command text based on a syntactic analysis result; 3) based on action keywords and object keywords in the keywords, acquiring a synonym set of the action keywords and a synonym set of the object keywords from a pre-established word vector dictionary, and constructing a command alternative phrase set, wherein the pre-established word vector dictionary is established based on Word2Vec; 4) respectively matching each command alternative phrase in the command alternative phrase set with each standard command in a pre-established command phrase library to obtain a command matching result; and 5) combining the command matching result and the keywords extracted in the step 2) to form a command recognition result. Compared with the prior art, the method eliminates the interference of redundant information, and has the advantages of high generalization performance of command matching, low manpower and time cost and the like.

Description

technical field [0001] The invention relates to the field of command recognition, in particular to a command recognition method based on keywords and Word2Vec. Background technique [0002] Human-computer interaction is a study of the interactive relationship between the system and the user. The system can be a computerized system and software or a robot. Service robot is a young member of the robot family. It integrates many disciplines such as machinery, electronics, materials, computers, sensors, and control. It is an important symbol of the country's high-tech strength and development level. [0003] Command recognition is an important field in human-computer interaction. It needs to define all supported command phrase libraries in advance, and match the commands issued by the user with the commands in the command phrase library during use, so as to execute the corresponding commands. Its main advantage is that the user does not need to use mouse, keyboard, remote contr...

AI Technical Summary

Problems solved by technology

The similarity at the string level is greatly affected by the length of the sentence, while the similarity at the shallow semantic level will lead to a decrease in accuracy with the increase of sentence components and redundant information

[0011] 2) Due to the existence of a large number of polysemy and synonyms in Chinese, the voice commands issued by the user to the robot are often quite different from the subscript forms in the command phrase library, resulting in a greatly reduced matching accuracy

If users are required to issue commands completely in the format of the command phrase library, problems such as high barriers to entry and poor scalability will arise.

[0012] 3) Although the deep semantic level method can often extract the semantic information of sentences, it requires a large amount of labeled data. In such a relatively new field, the acquisition of labeled data is extremely large, even in a short period of time. Impossible directly

Images