Search

CN-121997929-A - Equipment control method, device, medium and electronic equipment

CN121997929ACN 121997929 ACN121997929 ACN 121997929ACN-121997929-A

Abstract

The disclosure provides a device control method, a device, a medium and electronic equipment, and relates to the field of device control, wherein the device control method comprises the steps of obtaining natural language data of a target user; the natural language data are respectively input into a first semantic recognition model and a second semantic recognition model, a first intention list output by the first semantic recognition model and a second intention list output by the second semantic recognition model are obtained, the first semantic recognition model is different from the second semantic recognition model, the target intention is determined according to the first intention list and the second intention list, the target intention comprises target equipment to be controlled and target control intention of the target equipment, and the target equipment is controlled according to the target control intention. According to the method and the device, based on two different semantic recognition models, natural language data of the user can be processed in parallel, and the obtained target intention is more in line with the actual intention of the user to control the device, so that the device is controlled more accurately.

Inventors

  • ZHANG PENGFEI
  • FAN GE
  • YAN YUKUN

Assignees

  • 北京小米移动软件有限公司
  • 北京小米松果电子有限公司

Dates

Publication Date
20260508
Application Date
20241106

Claims (16)

  1. 1.A device control method, characterized by comprising: Acquiring natural language data of a target user; The natural language data are respectively input into a first semantic recognition model and a second semantic recognition model, a first intention list output by the first semantic recognition model and a second intention list output by the second semantic recognition model are obtained, and the first semantic recognition model is different from the second semantic recognition model; Determining a target intention according to the first intention list and the second intention list, wherein the target intention comprises target equipment to be controlled and target control intention of the target equipment; And controlling the target equipment according to the target control intention.
  2. 2. The method of claim 1, wherein determining a target intent from the first and second intent lists comprises: Scoring each intention in the first intention list and the second intention list to obtain the confidence of each intention; And determining the target intention according to the confidence degree of each intention.
  3. 3. The method of claim 2, wherein scoring each intent in the first and second lists of intents to obtain a confidence level for each intent comprises: Determining slot information of the intention for each intention; Determining a target correction value corresponding to the intention according to the intended slot position information; and correcting the preset confidence coefficient according to the target correction value to obtain the confidence coefficient of the intention.
  4. 4. A method according to claim 3, wherein determining a target correction value corresponding to the intention from the intended slot information comprises: determining at least one of equipment information, control intention information and slot number corresponding to the intention according to the intended slot information; And determining the target correction value according to at least one of equipment information, control intention information and the number of slots corresponding to the intention.
  5. 5. The method of claim 4, wherein determining the target correction value based on the device information corresponding to the intent comprises: determining a matching result between equipment corresponding to the equipment information and equipment associated with the target user, and/or acquiring a first historical use frequency of the equipment corresponding to the equipment information; and determining the target correction value according to the matching result and/or the first historical use frequency.
  6. 6. The method of claim 4, wherein determining a target correction value for correcting the preset confidence level based on the number of slots to which the intent corresponds comprises: determining the duty ratio of the number of the slots corresponding to the intention in the preset number of the slots; And determining the target correction value according to the duty ratio.
  7. 7. The method according to claim 4, wherein determining the target correction value based on control intention information corresponding to the intention includes: determining whether the control intention corresponding to the control intention information is a preset confusing intention or not, and/or acquiring a second historical use frequency of the control intention corresponding to the control intention information; Determining the target correction value according to the confusing intention and/or the second historical use frequency.
  8. 8. The method of any of claims 2-7, wherein determining the target intent based on the confidence level of each intent comprises: And determining the intention with the highest confidence in the first intention list and the second intention list as the target intention according to the confidence of each intention.
  9. 9. The method of any of claims 2-7, wherein determining the target intent based on the confidence level of each intent comprises: determining whether an intention of a target type exists according to the confidence coefficient of each intention, wherein the confidence coefficient corresponding to the intention of the target type is larger than or equal to a confidence coefficient threshold value; When the intention of the target type exists, the intention with the highest confidence degree in the intention of the target type is determined to be the target intention.
  10. 10. The method according to claim 9, wherein the method further comprises: and outputting prompt information for prompting the failure of semantic recognition when the intention of the target type is determined not to exist.
  11. 11. The method according to any of claims 1-7, wherein a first semantic recognition model is used to obtain the first list of intents according to: Carrying out semantic recognition on the natural language data to obtain a key information unit; According to a preset mapping table, matching the key information units to obtain standard intention information corresponding to each key information unit, wherein the standard intention information is information in standard intention for equipment control; And obtaining the first intention list according to the standard intention information corresponding to each key information unit.
  12. 12. The method of any of claims 1-7, wherein obtaining natural language data of the target user comprises: acquiring voice data of the target user; And preprocessing the voice data to obtain natural language data of the target user, wherein the preprocessing at least comprises format conversion operation for converting the voice data into text data.
  13. 13. The method according to any one of claims 1-7, wherein after controlling the target device according to the target control intention, the method further comprises: And training the first semantic recognition model and the second semantic recognition model according to the feedback data of the target user on the target intention and the execution result of controlling the target device.
  14. 14. An apparatus control device, comprising: The acquisition module is configured to acquire natural language data of a target user; The obtaining module is configured to input the natural language data into a first semantic recognition model and a second semantic recognition model respectively, and obtain a first intention list output by the first semantic recognition model and a second intention list output by the second semantic recognition model, wherein the first semantic recognition model is different from the second semantic recognition model; A determining module configured to determine a target intention including a target device to be controlled and a target control intention for the target device according to the first intention list and the second intention list; and the control module is configured to control the target equipment according to the target control intention.
  15. 15. A computer readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the device control method of any of claims 1-13.
  16. 16. An electronic device, comprising: storage means for storing a computer program; Execution means for executing the computer program to implement the device control method of any one of claims 1 to 13.

Description

Equipment control method, device, medium and electronic equipment Technical Field The disclosure relates to the technical field of equipment control, and in particular relates to an equipment control method, an equipment control device, a medium and electronic equipment. Background With the development of internet of things (Internet of Things, ioT) technology, more and more devices are connected to the internet, creating a huge IoT ecology. In the related art, equipment in the internet of things can be controlled based on voice of a user or text input by the user at a control end. Disclosure of Invention The disclosure provides a device control method, a device, a medium and electronic equipment, so as to provide a device control scheme which is more accurate and more suitable for actual control intention of a user. According to a first aspect of an embodiment of the present disclosure, there is provided an apparatus control method, including: Acquiring natural language data of a target user; The natural language data are respectively input into a first semantic recognition model and a second semantic recognition model, a first intention list output by the first semantic recognition model and a second intention list output by the second semantic recognition model are obtained, and the first semantic recognition model is different from the second semantic recognition model; Determining a target intention according to the first intention list and the second intention list, wherein the target intention comprises target equipment to be controlled and target control intention of the target equipment; And controlling the target equipment according to the target control intention. Optionally, determining the target intention according to the first intention list and the second intention list includes: Scoring each intention in the first intention list and the second intention list to obtain the confidence of each intention; And determining the target intention according to the confidence degree of each intention. Optionally, scoring each intention in the first intention list and the second intention list to obtain a confidence of each intention includes: Determining slot information of the intention for each intention; Determining a target correction value corresponding to the intention according to the intended slot position information; and correcting the preset confidence coefficient according to the target correction value to obtain the confidence coefficient of the intention. Optionally, determining, according to the intended slot information, a target correction value corresponding to the intent includes: determining at least one of equipment information, control intention information and slot number corresponding to the intention according to the intended slot information; And determining the target correction value according to at least one of equipment information, control intention information and the number of slots corresponding to the intention. Optionally, determining the target correction value according to the device information corresponding to the intention includes: determining a matching result between equipment corresponding to the equipment information and equipment associated with the target user, and/or acquiring a first historical use frequency of the equipment corresponding to the equipment information; and determining the target correction value according to the matching result and/or the first historical use frequency. Optionally, determining, according to the number of slots corresponding to the intent, a target correction value for correcting the preset confidence level includes: determining the duty ratio of the number of the slots corresponding to the intention in the preset number of the slots; And determining the target correction value according to the duty ratio. Optionally, determining the target correction value according to control intention information corresponding to the intention includes: determining whether the control intention corresponding to the control intention information is a preset confusing intention or not, and/or acquiring a second historical use frequency of the control intention corresponding to the control intention information; Determining the target correction value according to the confusing intention and/or the second historical use frequency. Optionally, determining the target intention according to the confidence of each intention includes: And determining the intention with the highest confidence in the first intention list and the second intention list as the target intention according to the confidence of each intention. Optionally, determining the target intention according to the confidence of each intention includes: determining whether an intention of a target type exists according to the confidence coefficient of each intention, wherein the confidence coefficient corresponding to the intention of the target type is larger than or equal to a