CN110675876A - Intelligent control method, intelligent controller, system and storage medium for semantic recognition - Google Patents
Intelligent control method, intelligent controller, system and storage medium for semantic recognition Download PDFInfo
- Publication number
- CN110675876A CN110675876A CN201910990126.0A CN201910990126A CN110675876A CN 110675876 A CN110675876 A CN 110675876A CN 201910990126 A CN201910990126 A CN 201910990126A CN 110675876 A CN110675876 A CN 110675876A
- Authority
- CN
- China
- Prior art keywords
- intention
- voice
- knowledge
- operator
- mode
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000011156 evaluation Methods 0.000 claims abstract description 27
- 238000012545 processing Methods 0.000 claims description 26
- 238000000605 extraction Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 8
- 230000004927 fusion Effects 0.000 claims description 6
- 239000000284 extract Substances 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 6
- 238000012549 training Methods 0.000 abstract description 3
- 230000000875 corresponding effect Effects 0.000 description 24
- 241001672694 Citrus reticulata Species 0.000 description 17
- 230000008859 change Effects 0.000 description 14
- 238000001816 cooling Methods 0.000 description 10
- 238000010438 heat treatment Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000005057 refrigeration Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 229910001316 Ag alloy Inorganic materials 0.000 description 1
- BQCADISMDOOEFD-UHFFFAOYSA-N Silver Chemical compound [Ag] BQCADISMDOOEFD-UHFFFAOYSA-N 0.000 description 1
- 238000004378 air conditioning Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/28—Constructional details of speech recognition systems
- G10L15/30—Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/223—Execution procedure of a spoken command
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a semantic recognition intelligent control method and an intelligent control system, wherein a voice control instruction is obtained, a voice control signal is analyzed, recognized and processed, when the intention of an operator can be recognized, an operation control command associated with the intention is directly executed, corresponding control is realized, when the intention cannot be recognized, a voice manual association control mode is entered, when the intention of the operator can be recognized, corresponding control is realized, when the intention of the operator cannot be recognized, a voice system model analysis mode is entered, when the intention of the operator can be recognized, corresponding control is realized, and when the intention of the operator cannot be recognized, iterative update is carried out on a voice system model until the intention of the operator can be recognized. Repeated semantic training is avoided by establishing the voice system model, the system running speed is increased, the evaluation condition is set, the probability of obtaining the intention of an instructor from an unknown voice control instruction is increased, the recognition effect is good, and a user has better use experience.
Description
Technical Field
The invention relates to the technical field of intelligent home voice control, in particular to a semantic recognition intelligent control method, an intelligent controller, a system and a storage medium.
Background
Along with the rapid development of the internet of things, due to the advantages of simplicity, rapidness, good interactivity and the like, more and more intelligent devices adopt voice interaction at present and gradually become a preferred interaction mode for people. However, there is a problem that when a user uses dialects, non-standard mandarin or english in various places to control a voice device, the intelligent device has a poor recognition effect, cannot really understand the semantics of the dialects or the non-standard mandarin or english uttered by the user, cannot accurately recognize and understand the intention of the user, and is difficult to realize free voice control of the intelligent device.
Disclosure of Invention
The technical problem to be solved by the invention is that when a user controls a voice device by dialect, nonstandard Mandarin or English, the control device has the conditions of poor recognition effect and even incapability of recognizing, can not really understand the problems proposed by the user, can not accurately recognize and understand the user intention, and has poor user experience.
Mainly comprises the following aspects:
the invention provides a semantic recognition intelligent control method, which comprises the following steps:
acquiring a voice control instruction;
analyzing the voice control signal to identify the intention of the commander;
when the intention of the operator can be identified, executing an operation control command associated with the intention to realize corresponding control, and when the intention of the operator cannot be identified, entering a voice manual association control mode;
under a voice manual association control mode, associating the voice control instruction with the manual control instruction, and identifying the intention of an operator according to the associated voice control instruction;
when the intention of an operator can be recognized, executing an operation control command associated with the intention to realize corresponding control, and when the intention of the operator cannot be recognized, entering a phonetic system model analysis mode;
in the phonetic system model analysis mode, recognizing the intention of the commander by using a phonetic system model based on a knowledge graph;
when the intention of the operator can be identified, executing the operation control command associated with the intention to realize the corresponding control, when the intention of the operator cannot be identified, iteratively updating the phonetic system model, identifying the intention of the operator again by using the updated phonetic system model until the intention of the operator can be identified, and executing the operation control command associated with the intention to realize the corresponding control.
According to an embodiment of the invention, associating the voice control instruction with the manual control instruction comprises:
acquiring a voice control instruction while acquiring a manual control instruction, and analyzing a voice signal;
the instructor's intention, from which the speech signal is parsed, is associated with a manual control instruction.
According to an embodiment of the present invention, a phonetic system model is constructed, specifically,
collecting raw data, the raw data comprising structured, semi-structured, and unstructured data;
carrying out knowledge extraction on the original data and the data of the third-party basic knowledge;
performing knowledge representation on the information extracted by the knowledge, and storing the information into a mode layer of a speech system model;
storing the knowledge representation, the updated knowledge after the knowledge representation and the intention of the commander obtained by combining knowledge reasoning into a mode layer of the voice system model;
comparing the intention of the commander with the data of third-party professional knowledge and the data of an expert system to obtain an intention conclusion of the commander;
and performing quality check evaluation on the intention conclusion of the operator, storing the intention conclusion meeting the evaluation threshold condition into a mode layer of the speech system model, executing an operation control command associated with the intention, realizing corresponding control, and transferring the intention conclusion not meeting the evaluation threshold condition into the step of collecting original data for iteration until the intention conclusion meets the evaluation threshold condition.
According to an embodiment of the present invention, knowledge extraction, in particular,
extracting knowledge facts related to voice control instructions of an operator from original data and data of third-party basic knowledge in an automatic or semi-automatic mode, carrying out knowledge fusion on the knowledge facts, and storing fused information into a data layer of a voice system model;
and extracting knowledge facts related to the voice control instruction of the instructor from the semi-structured and unstructured data in an automatic or semi-automatic mode, and storing the knowledge facts into a data layer of the voice system model.
According to the embodiment of the invention, the automatic mode extraction is that the program automatically extracts knowledge facts, and the semi-automatic extraction is that the knowledge facts are manually extracted. In accordance with an embodiment of the present invention,
the automatic extraction method is characterized in that the knowledge fact is automatically extracted by the program, and the semi-automatic extraction method is characterized in that the knowledge fact is manually extracted.
In a second aspect, the present invention further provides an intelligent controller, which includes a storage module, a processing module and a voice receiving module, wherein the storage module stores thereon a computer program, and when the computer program is executed by the processing module, the steps of any one of the above semantic recognition intelligent control methods are implemented.
In a third aspect, the present invention further provides an intelligent controller, a semantic recognition intelligent control system, including:
the intelligent controller and the data server are described above; the storage module of the data server stores a knowledge graph for constructing a voice system model.
According to an embodiment of the present invention, the server is a cloud server.
According to the embodiment of the invention, the intelligent controller is a mobile terminal.
In a fourth aspect, the present invention also provides a storage medium storing a computer program for implementing the steps of any one of the above semantic recognition intelligent control methods.
One or more embodiments of the present invention may have the following advantages over the prior art:
the intelligent control method for semantic recognition provided by the invention can acquire a voice control instruction, analyze a voice signal and recognize the intention of an operator, if the intention of the operator cannot be recognized, the intelligent control method enters a manual voice association control mode, and if the intention of the operator cannot be recognized, the intelligent control method for semantic recognition enters a voice system model analysis mode, wherein the voice system model is constructed by adopting a semantic recognition mode based on a knowledge graph. Therefore, various voice situations of the operator of Mandarin, nonstandard Mandarin, dialect and other languages can be finally identified through multiple modes, and intelligent control is realized.
The knowledge graph selects original data in various formats such as structured data, semi-structured data, unstructured data and the like and third-party basic knowledge data to construct basic data, and also adopts knowledge extraction, knowledge representation, knowledge fusion and knowledge reasoning processes to construct the knowledge graph.
The voice system model adopts a data layer and a mode layer to store corresponding data, compares the intention of the commander obtained by inference of the voice system model with a third-party professional knowledge database and an expert system database to obtain the intention of the commander, and performs quality check and evaluation, so that the intention meeting the evaluation threshold condition can execute a control instruction associated with the intention, and corresponding control is realized.
Repeated semantic training is avoided by establishing the voice system model, the method is different from the multi-layer knowledge graph in the prior art, the number of layers is reduced, the running speed is improved, meanwhile, third-party comparison processing and evaluation threshold evaluation conditions are set, and the probability of obtaining the intention of an instructor from an unknown voice control instruction is improved, so that the voice recognition probability of dialects, nonstandard mandarins and other languages (such as foreign languages including English) is improved, the recognition effect is good, and a user has better use experience.
The intelligent control system for semantic recognition comprises an intelligent controller and a data server, wherein the intelligent controller can adopt a mobile terminal mode, the data server can adopt a cloud server mode, and a knowledge graph for constructing a voice system type is stored in a storage module of the data server, so that the mobile terminal is convenient to use, or the knowledge graph of the cloud server is called through a network, the intention of an operator can be recognized quickly from the terminal to the terminal and from the terminal to the cloud server, voice intelligent control is realized in time, and the use experience of a user is effectively improved.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a general flow diagram of a semantic recognition intelligent control method based on knowledge-graph according to an example of the present invention;
FIG. 2 is a flow diagram illustrating an exemplary process for constructing a phonetic hierarchy model according to the present invention;
FIG. 3 is a block diagram of an example intellectual property graph based semantic recognition intelligent control architecture of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the following detailed description of the present invention with reference to the accompanying drawings is provided to fully understand and implement the technical effects of the present invention by solving the technical problems through technical means. It should be noted that, as long as there is no conflict, the embodiments and the features of the embodiments of the present invention may be combined with each other, and the technical solutions formed are within the scope of the present invention.
First embodiment
The embodiment provides an intelligent control method for semantic recognition, as shown in fig. 1, including the following steps:
acquiring a voice control instruction;
analyzing the voice control signal to identify the intention of the commander;
when the intention of the operator can be identified, executing an operation control command associated with the intention to realize corresponding control, and when the intention of the operator cannot be identified, entering a voice manual association control mode;
under a voice manual association control mode, associating the voice control instruction with the manual control instruction, and identifying the intention of an operator according to the associated voice control instruction;
when the intention of an operator can be recognized, executing an operation control command associated with the intention to realize corresponding control, and when the intention of the operator cannot be recognized, entering a phonetic system model analysis mode;
in the phonetic system model analysis mode, recognizing the intention of the commander by using a phonetic system model based on a knowledge graph;
when the intention of the operator can be identified, executing the operation control command associated with the intention to realize the corresponding control, when the intention of the operator cannot be identified, iteratively updating the phonetic system model, identifying the intention of the operator again by using the updated phonetic system model until the intention of the operator can be identified, and executing the operation control command associated with the intention to realize the corresponding control.
In this embodiment, the instructor issues a voice control command of mandarin, nonstandard mandarin, dialect, or other languages (e.g., foreign languages such as english, etc.), for example, issues a voice control command of mandarin "change mode"; acquiring a voice control instruction of the mandarin changing mode, and analyzing, identifying and processing the collected mandarin changing mode voice signals; the intention of the operator can be judged by analysis, the intention of the operator can be judged to be a mode change by general analysis and identification, and at the moment, if the intelligent equipment is an intelligent air conditioner, a mode change operation control command associated with the intention can be directly executed, so that voice control is realized.
For another example, the instructor sends out a voice control command of a dialect "ti a o (meaning changed for Mandarin, the same applies hereinafter) mode"; collecting a speech signal of a dialect' ti { hacek over (o) }; analyzing and recognizing the collected dialect' ti { hacek over (o) } mode voice signals; the intention of the operator is analyzed and judged, generally, the meaning of relevant pronunciation such as picking or jumping is identified as ti { hacek over (a) } o', the intention of the operator cannot be accurately judged through analysis and identification, and at the moment, the manual voice association control mode is entered.
Second embodiment
The embodiment provides a semantic recognition intelligent control method, wherein a voice manual association control mode is adopted, as shown in figure 1,
associating the voice control instruction with the manual control instruction, comprising:
acquiring a voice control instruction while acquiring a manual control instruction, and analyzing the voice signal;
the instructor's intention, as analyzed from the speech signal, is associated with a manual control instruction.
In particular to a method for preparing a high-performance nano-silver alloy,
when the operator performs voice control, the intelligent equipment does not respond;
recording a voice control instruction of an operator;
the operator sends out the voice control instruction again;
continuously recording the voice control instruction sent by the commander;
the voice control instruction does not respond for many times, and the operator adopts manual control and performs manual operation by using a control button arranged on a remote controller or intelligent equipment;
recording a manual operation control instruction of an operator, and associating the manual operation control instruction with an instruction controlled by a user voice;
storing a control instruction and an associated voice control instruction for an operator to manually operate the intelligent equipment;
the commander sends out the same voice control instruction again, collects the voice signal, and analyzes and identifies the voice signal of the commander;
analyzing and judging the intention of the operator, executing the operation control command associated with the intention when the intention of the operator can be judged, realizing corresponding control, and entering a phonetic system model analysis process when the intention of the operator cannot be judged.
In this embodiment, the dialect "ti a o mode" voice control is still taken as an example by the instructor, and at this time, the intelligent air conditioner as the intelligent device does not respond; recording a voice control instruction of a dialect' ti { hacek over (o) } mode adopted by an instructor; the instructor sends out the dialect 'ti { hacek over (a) } o mode' voice control instruction again; continuing to record the dialect' ti { hacek over (o) } mode voice control instruction sent by the commander; the dialect 'ti { hacek over (a) } o mode' voice control instruction is not responded for many times, the commander can only adopt manual control at this time, and the commander uses a remote controller or a control button arranged on an intelligent air conditioner to carry out manual operation and change the mode, such as changing from heating to air changing; recording a manual operation control instruction replacement mode of an operator, replacing a heating mode with a wind exchange mode, and associating the manual operation control instruction replacement mode with a voice control instruction of a user dialect 'ti a o mode'; storing a voice control instruction of a control instruction replacement mode of manually operating the intelligent equipment by an operator, wherein the control instruction replacement mode is replaced from a heating mode to a wind exchange mode and is associated with a dialect' ti a o mode; the commander sends out the same dialect 'ti { hacek over (a) } o mode' voice control instruction again, collects the dialect 'ti { hacek over (a) } o mode' voice signal, and carries out analysis and recognition processing on the dialect 'ti { hacek over (a) } o mode' voice signal of the commander; the intention of the operator is analyzed and judged, and the voice control command of the dialect' ti { hacek over (o) } mode is stored before, and the manual control command mode change mode associated with the voice control command is also stored for changing from the heating mode to the air change mode, so that the intention of the operator can be judged, and the mode change operation control command associated with the intention can be directly executed, thereby realizing the voice control.
If the dialect voice control command of the operator is associated with a plurality of manual operation commands or other conditions, so that the intention of the operator cannot be judged yet, the phonetic system model analysis mode is entered.
Third embodiment
The embodiment provides an intelligent control method for semantic recognition, wherein a speech system model analysis mode, specifically,
constructing a phonetic system model, as shown in FIG. 3;
in the phonetic system model analysis mode, recognizing the intention of the commander by using a phonetic system model based on a knowledge graph; when the intention of the operator can be identified, executing the operation control command associated with the intention to realize the corresponding control, when the intention of the operator cannot be identified, iteratively updating the phonetic system model, identifying the intention of the operator again by using the updated phonetic system model until the intention of the operator can be identified, and executing the operation control command associated with the intention to realize the corresponding control.
According to an embodiment of the present invention, the constructing of the phonetic system model, in particular,
collecting raw data, the raw data comprising structured, semi-structured, and unstructured data;
storing the original data and the data of the third-party basic knowledge into a data layer of a voice system model, and extracting the knowledge of the data;
performing knowledge representation on the information extracted by the knowledge, and storing the information into a mode layer of a speech system model;
storing the knowledge representation, the updated knowledge after the knowledge representation and the intention of the commander obtained by combining knowledge reasoning into a mode layer of a voice command system model;
comparing the intention of the commander with the data of third-party professional knowledge and the data of an expert system to obtain an intention conclusion of the commander;
and performing quality check evaluation on the intention conclusion of the operator, storing the intention conclusion meeting the evaluation threshold condition into a mode layer of the voice instruction system model, calling a related operation control command to realize voice control, and transferring the intention conclusion not meeting the evaluation threshold condition into the step of collecting original data to perform iterative updating until the intention conclusion meets the evaluation threshold condition.
A model of the phonetic system is constructed, specifically,
various voice control instructions of an administrator are collected by adopting the prior art, for example, for the same 'conversion mode' control instruction, voice data of different voice colors of foreign languages such as mandarin, dialects of various regions, English and the like are collected, then, various control instructions are classified according to the voiceprint characteristics of the voice data, are converted into knowledge data in a preset format, and are stored in an original database as structured data.
Meanwhile, some semi-structured control instructions OR semi-structured data of the same control instruction, such as fuzzy voice control instructions like 'cooling' OR 'heating', can correspond to a plurality of temperature options and cannot correspond to one another, so that a plurality of corresponding semi-structured control instruction data exist.
There are some unstructured data, such as voice control using online video, at this time, it is impossible to immediately structure the voice control of the video, and only the unstructured data is saved first.
All the most original structured, semi-structured and unstructured data form an original database which is stored in the data layer of the system architecture model, so that calling and accessing are convenient.
Or the operator adopts the dialect 'ti { hacao } mode' to send a voice control instruction, but a plurality of corresponding manual control instruction switching modes are available, such as switching from a heating mode to a cooling mode, switching from the cooling mode to the heating mode, switching from the cooling mode to an air switching mode, switching from the heating mode to a humidification mode and the like, and the original associated voice control instruction and manual control instruction process cannot judge the intention of the operator.
Carrying out knowledge extraction on all related structured, semi-structured and unstructured data of a dialect 'ti { hacek over (a) } o mode';
the method comprises the steps of expressing knowledge of information extracted by knowledge, such as 'mode change' voiceprint information structured data of Mandarin, control instructions of a plurality of manual operation mode change associated with 'mode change' of semi-structured, and unstructured video or recorded data, defining 'mode change' as switching among a plurality of modes of the intelligent air conditioner, and storing the 'mode change' in a mode layer of a system architecture model;
the intention conclusion of an instructor obtained by combining the just-existing knowledge representation, knowledge updated after the knowledge representation (such as adding mode switching) and knowledge reasoning (logic reasoning or common selection), for example, in summer, the mode switching is agreed to change from a heating mode to a cooling mode, and the mode switching is stored in a mode layer of the system architecture model;
the intention conclusion of the commander is changed into a refrigeration mode if the mode change is appointed to be the heating mode in summer, the intention conclusion is compared with data of third-party professional knowledge (such as a peer-to-peer reference database or self-defined control command data) and data of an expert system (such as manual customer service or an expert online system) to obtain the intention conclusion of the commander, an air-conditioning temperature sensor senses that the room is 35 ℃, and the commander sends a voice control instruction of a dialect 'ti/o mode' at this time and judges that the mode is switched into the refrigeration mode from any mode;
then, the model performs quality check evaluation on the intention conclusion "switching from any mode to the cooling mode" of the instructor, for example, when the air conditioner temperature sensor senses that the room is 35 degrees, the instructor sends a voice control instruction of the dialect "ti { hacho } mode" at this time, and among 10 voice control instructions sent, 9 times of voice control instructions are switched from any mode to the cooling mode, the instructor is considered to meet the evaluation threshold condition, the intention conclusion "switching from any mode to the cooling mode" of the instructor meeting the quality check evaluation threshold condition is stored in the mode layer of the architecture model, and then the operation control instruction associated with the intention is executed to switch from any mode to the cooling mode, so that the voice control is realized.
For example, if the air conditioner temperature sensor senses that the room is 35 degrees, the commander sends a voice control instruction of a dialect' ti a { hacek over (o) } mode at this time, and 2 times of the 10 sent voice control instructions are switched from any mode to a ventilation mode, the command is considered to be not satisfied with the evaluation threshold condition, the original intention conclusion of switching from any mode to a cooling mode is not considered as the intention conclusion of the commander, and then the step of collecting original data is carried out for iteration until the intention conclusion of the commander satisfies the evaluation threshold condition.
The specific evaluation threshold condition needs to be set according to the specific voice control instruction and the related actual control instruction.
In an embodiment, the knowledge extraction can also be,
extracting knowledge facts related to voice control instructions of an operator from the structured data and the data of third-party basic knowledge in an automatic or semi-automatic mode, carrying out knowledge fusion on the knowledge facts, and storing the fused information into a data layer of a system architecture model;
and extracting knowledge facts related to the voice control instruction of the instructor from the semi-structured and unstructured data in an automatic or semi-automatic mode, and storing the knowledge facts into a data layer of the architecture model.
The automatic extraction method is characterized in that the knowledge fact is automatically extracted by the program, and the semi-automatic extraction method is characterized in that the knowledge fact is manually extracted.
In summary, the semantic recognition intelligent control method provided by the invention can acquire a voice control instruction, analyze a voice signal, recognize the intention of an operator, enter a voice manual association control mode if the intention of the operator cannot be recognized, and enter a voice system model analysis mode if the intention of the operator cannot be recognized, wherein the voice system model is constructed by adopting a semantic recognition mode based on a knowledge graph. Therefore, various voice situations of the operator of Mandarin, nonstandard Mandarin, dialect and other languages can be finally identified through multiple modes, and intelligent control is realized.
The knowledge graph selects original data in various formats such as structured data, semi-structured data, unstructured data and the like and third-party basic knowledge data to construct basic data, and also adopts knowledge extraction, knowledge representation, knowledge fusion and knowledge reasoning processes to construct the knowledge graph.
The voice system model adopts a data layer and a mode layer to store corresponding data, compares the intention of the commander obtained by inference of the voice system model with a third-party professional knowledge database and an expert system database to obtain the intention of the commander, and performs quality check and evaluation, so that the intention meeting the evaluation threshold condition can execute a control instruction associated with the intention, and corresponding control is realized.
Repeated semantic training is avoided by establishing the voice system model, the method is different from the multi-layer knowledge graph in the prior art, the number of layers is reduced, the running speed is improved, meanwhile, third-party comparison processing and evaluation threshold evaluation conditions are set, and the probability of obtaining the intention of an instructor from an unknown voice control instruction is improved, so that the voice recognition probability of dialects, nonstandard mandarins and other languages (such as foreign languages including English) is improved, the recognition effect is good, and a user has better use experience.
Fourth embodiment
The present embodiment provides an intelligent controller, which includes a storage module, a processing module, and a voice receiving module, where the storage module stores a computer program, and the computer program, when executed by the processing module, implements the steps of any one of the above semantic recognition intelligent control methods.
This embodiment still provides an intelligent control ware, a semantic recognition intelligent control system, includes:
the intelligent controller and the data server are described above; the storage module of the data server stores a knowledge graph for constructing a voice system model.
According to this embodiment, the server is a cloud server.
According to this embodiment, the intelligent controller is a mobile terminal.
The present embodiment also provides a storage medium storing a computer program for implementing the steps of any one of the above semantic identification intelligent control methods.
In summary, the semantic recognition intelligent control system based on the knowledge graph of the embodiment includes an intelligent controller and a data server, wherein the intelligent controller can adopt a mobile terminal mode, the data server can adopt a cloud server mode, and the knowledge graph for constructing a speech system type is stored in a storage module of the data server, so that the mobile terminal is convenient to use, or the knowledge graph of the cloud server is called through a network, so that the intention of an operator can be recognized from the terminal to the terminal and from the terminal to the cloud server, the voice intelligent control is realized in time, and the use experience of a user is effectively improved.
Fifth embodiment
In addition, a semantic recognition intelligent control system based on knowledge graph, as shown in fig. 3, may further include:
the voice receiving module is used for collecting a voice control instruction of an administrator and sending the voice control instruction to the cloud service platform server for recognition processing;
the manual control module is a control button module arranged on a remote controller or intelligent equipment, and an operation control instruction of the manual control module is uploaded to the cloud service platform server and stored after being implemented;
the cloud service platform server stores the voice control instruction and the manual control instruction, associates the voice control instruction with the corresponding manual control instruction, sends the voice control instruction capable of being identified to the intelligent equipment, transfers the voice control instruction incapable of being identified to the knowledge graph server for identification, and sends the voice control instruction corresponding to the intention of the commander sent by the knowledge graph server to the intelligent equipment; the voice control instruction which cannot be identified and processed by the cloud service platform server is transferred to the knowledge graph server for identification and processing;
the cloud service platform server comprises a voice recognition processing system and a server database;
the voice recognition processing system is used for recognizing the voice signal of the commander, analyzing and processing the voice signal, analyzing and judging the intention of the commander, calling a recognizable voice control instruction and sending the recognizable voice control instruction to the intelligent equipment realizing corresponding control; the voice recognition processing system can recognize the voice signal of the operator, analyze and process the voice signal, analyze and judge the intention of the operator, call a control instruction, correspondingly control the intelligent equipment, transfer the voice control instruction which can not be recognized into the knowledge graph server for recognition processing, and then send the voice control instruction corresponding to the intention of the operator transmitted by the knowledge graph server to the intelligent equipment;
and the server database receives and stores the one-time or multiple-time voice control instruction of the operator, stores the one-time or multiple-time manual control instruction of the operator, and associates the voice control instruction with the corresponding manual control instruction.
The voice control instruction which cannot be identified and processed by the cloud service platform server is transferred to the knowledge graph server for identification and processing;
a knowledge graph server comprising a data layer module and a schema layer module, wherein,
the data layer module stores information of the original database, the third-party database and knowledge fusion;
the original knowledge database comprises a structured database, a semi-structured database and an unstructured database;
and the third-party database comprises a third-party basic knowledge database, a third-party professional knowledge database and an expert system database.
The schema layer module stores refined data, the refined data comprising data represented by knowledge, data of an intention conclusion of an instructor, and data of an intention conclusion evaluated by a quality check.
The method comprises the steps that a knowledge graph server receives a voice control instruction which cannot be recognized and processed by a cloud service platform server, loop iteration is carried out inside the knowledge graph server, analysis, recognition and processing are carried out until the intention of an instructor is output, and the intention of the instructor is transmitted to the cloud service platform server;
the intelligent device receives the voice control instruction from the voice recognition processing system of the cloud service platform and starts a corresponding function according to the voice control instruction.
If the voice control instruction can be directly recognized, the control instruction associated with the intention is executed, and the intelligent air conditioner is correspondingly controlled.
If the voice recognition processing system of the cloud service platform server cannot recognize the voice control instruction, the dialect 'ti { hacho } mode' voice control instruction is sent to a cloud service platform server database, after an operator uses the manual control module to operate, the dialect 'ti { hacho } mode' voice control instruction and the manual operation control instruction are correlated and then transmitted back to the voice recognition processing system of the cloud service platform server, and the voice recognition processing system of the cloud service platform server executes the control instruction correlated with the intention, so that the voice control intelligent air conditioner is realized.
If the voice recognition processing system of the cloud service platform server still cannot recognize the control instruction of the intention of the commander from the associated voice control instruction and the manual control instruction, the dialect 'ti { hacek over (a) } o mode' voice control instruction is sent to the knowledge graph server to be subjected to iterative analysis until the control instruction of the intention of the commander is obtained, and then the control instruction is transmitted back to the voice recognition processing system of the cloud service platform server, and the voice recognition processing system of the cloud service platform server executes the control instruction associated with the intention, so that the voice control intelligent air conditioner is realized.
Although the embodiments of the present invention have been described above, the above description is only for the convenience of understanding the present invention, and is not intended to limit the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as disclosed, and that the scope of the invention is not to be limited to the particular embodiments disclosed herein but is to be accorded the full scope of the claims.
Claims (10)
1. An intelligent control method for semantic recognition is characterized by comprising the following steps:
acquiring a voice control instruction;
analyzing the voice control signal to identify the intention of the commander;
when the intention of an operator can be recognized, executing an operation control command associated with the intention to realize corresponding control, and when the intention of the operator cannot be recognized, entering a voice manual association control mode;
under a voice manual association control mode, associating the voice control instruction with the manual control instruction, and identifying the intention of an operator according to the associated voice control instruction;
when the intention of an instructor can be recognized, executing an operation control command associated with the intention to realize corresponding control, and when the intention of the instructor cannot be recognized, entering a phonetic system model analysis mode;
in the phonetic system model analysis mode, recognizing the intention of the commander by using a phonetic system model based on a knowledge graph;
when the intention of the operator can be identified, executing the operation control command associated with the intention to realize the corresponding control, when the intention of the operator cannot be identified, iteratively updating the phonetic system model, identifying the intention of the operator again by using the updated phonetic system model until the intention of the operator can be identified, and executing the operation control command associated with the intention to realize the corresponding control.
2. The intelligent control method for semantic recognition according to claim 1,
the associating the voice control instruction with the manual control instruction comprises:
the voice control instruction is acquired while the manual control instruction is acquired, and the voice signal is analyzed;
associating the analyzed voice signal with a manual control instruction.
3. The intelligent control method for semantic recognition according to claim 2,
the construction of the phonetic system model is, specifically,
collecting raw data, the raw data comprising structured, semi-structured, and unstructured data;
performing knowledge extraction on the original data and the data of the third-party basic knowledge;
performing knowledge representation on the information extracted by the knowledge, and storing the information into a mode layer of a speech system model;
storing the knowledge representation, the knowledge updated after the knowledge representation and the intention of an instructor obtained by combining knowledge reasoning into a mode layer of the speech system model;
comparing the intention of the commander with data of third-party professional knowledge and data of an expert system to obtain an intention conclusion of the commander;
and performing quality check evaluation on the intention conclusion of the instructor, storing the intention conclusion meeting the evaluation threshold condition to a mode layer of the speech system model, executing an operation control command associated with the intention, realizing corresponding control, switching the intention conclusion not meeting the evaluation threshold condition into the step of collecting original data, and performing iteration until the intention conclusion meets the evaluation threshold condition.
4. The intelligent control method for semantic recognition according to claim 3,
the knowledge extraction is carried out by, specifically,
extracting knowledge facts related to voice control instructions of an operator from the original data and the data of the third-party basic knowledge in an automatic or semi-automatic mode, carrying out knowledge fusion on the knowledge facts, and storing the fused information into a data layer of the voice system model;
and extracting knowledge facts related to the voice control instruction of the instructor from the semi-structured data and the unstructured data in an automatic or semi-automatic mode, and storing the knowledge facts into a data layer of the voice system model.
5. The intelligent control method for semantic recognition according to claim 4,
the automatic extraction is to automatically extract knowledge facts for the program itself, and the semi-automatic extraction is to manually extract knowledge facts.
6. An intelligent controller comprising a memory module, a processing module, and a voice receiving module, the memory module having stored thereon a computer program that, when executed by the processing module, performs the steps of the method of any one of claims 1 to 5.
7. A semantic recognition intelligent control system, comprising:
the intelligent controller and data server of claim 6; the storage module of the data server stores a knowledge graph for constructing a voice system model.
8. The intelligent semantic recognition control system of claim 7, wherein the server is a cloud server.
9. The intelligent control system for semantic recognition according to claim 7 or 8, wherein the intelligent controller is a mobile terminal.
10. A storage medium, characterized in that a computer program for implementing the steps of the method of any one of claims 1 to 5 is stored.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990126.0A CN110675876A (en) | 2019-10-17 | 2019-10-17 | Intelligent control method, intelligent controller, system and storage medium for semantic recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910990126.0A CN110675876A (en) | 2019-10-17 | 2019-10-17 | Intelligent control method, intelligent controller, system and storage medium for semantic recognition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110675876A true CN110675876A (en) | 2020-01-10 |
Family
ID=69082846
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910990126.0A Pending CN110675876A (en) | 2019-10-17 | 2019-10-17 | Intelligent control method, intelligent controller, system and storage medium for semantic recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110675876A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111951795A (en) * | 2020-08-10 | 2020-11-17 | 中移（杭州）信息技术有限公司 | Voice interaction method, server, electronic device and storage medium |
| CN113808589A (en) * | 2020-06-15 | 2021-12-17 | 中兴通讯股份有限公司 | Agent call method, computer device, and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202795020U (en) * | 2012-09-24 | 2013-03-13 | 杨炙龙 | Intelligent home control system |
| CN106448671A (en) * | 2016-10-26 | 2017-02-22 | 安徽省云逸智能科技有限公司 | Speech interaction system for controlling smart home device |
| US20180342236A1 (en) * | 2016-10-11 | 2018-11-29 | Mediazen, Inc. | Automatic multi-performance evaluation system for hybrid speech recognition |
| CN108919669A (en) * | 2018-09-11 | 2018-11-30 | 深圳和而泰数据资源与云技术有限公司 | A kind of smart home dynamic decision method, apparatus and service terminal |
| CN109062094A (en) * | 2018-07-24 | 2018-12-21 | 卢坤雄 | Control method, device and the storage medium of model |
| US20190080687A1 (en) * | 2017-09-13 | 2019-03-14 | Hitachi, Ltd. | Learning-type interactive device |
| CN109471678A (en) * | 2018-11-07 | 2019-03-15 | 苏州思必驰信息科技有限公司 | Voice midpoint controlling method and device based on image recognition |
| CN110275959A (en) * | 2019-05-22 | 2019-09-24 | 广东工业大学 | A kind of Fast Learning method towards large-scale knowledge base |
-
2019
- 2019-10-17 CN CN201910990126.0A patent/CN110675876A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202795020U (en) * | 2012-09-24 | 2013-03-13 | 杨炙龙 | Intelligent home control system |
| US20180342236A1 (en) * | 2016-10-11 | 2018-11-29 | Mediazen, Inc. | Automatic multi-performance evaluation system for hybrid speech recognition |
| CN106448671A (en) * | 2016-10-26 | 2017-02-22 | 安徽省云逸智能科技有限公司 | Speech interaction system for controlling smart home device |
| US20190080687A1 (en) * | 2017-09-13 | 2019-03-14 | Hitachi, Ltd. | Learning-type interactive device |
| CN109062094A (en) * | 2018-07-24 | 2018-12-21 | 卢坤雄 | Control method, device and the storage medium of model |
| CN108919669A (en) * | 2018-09-11 | 2018-11-30 | 深圳和而泰数据资源与云技术有限公司 | A kind of smart home dynamic decision method, apparatus and service terminal |
| CN109471678A (en) * | 2018-11-07 | 2019-03-15 | 苏州思必驰信息科技有限公司 | Voice midpoint controlling method and device based on image recognition |
| CN110275959A (en) * | 2019-05-22 | 2019-09-24 | 广东工业大学 | A kind of Fast Learning method towards large-scale knowledge base |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113808589A (en) * | 2020-06-15 | 2021-12-17 | 中兴通讯股份有限公司 | Agent call method, computer device, and storage medium |
| CN111951795A (en) * | 2020-08-10 | 2020-11-17 | 中移（杭州）信息技术有限公司 | Voice interaction method, server, electronic device and storage medium |
| CN111951795B (en) * | 2020-08-10 | 2024-04-09 | 中移（杭州）信息技术有限公司 | Voice interaction method, server, electronic device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210190360A1 (en) | Artificial intelligence device | |
| WO2019101083A1 (en) | Voice data processing method, voice-based interactive device, and storage medium | |
| CN110334347B (en) | Information processing method based on natural language recognition, related equipment and storage medium | |
| CN108932945B (en) | Voice instruction processing method and device | |
| CN109754804A (en) | Voice control method and device, storage medium and intelligent home system | |
| WO2019218820A1 (en) | Method and apparatus for determining controlled object, and storage medium and electronic device | |
| CN109243432A (en) | Speech processing method and the electronic equipment for supporting the speech processing method | |
| CN108153158A (en) | Switching method, device, storage medium and the server of household scene | |
| CN112799747A (en) | Intelligent assistant evaluation and recommendation method, system, terminal and readable storage medium | |
| WO2020253064A1 (en) | Speech recognition method and apparatus, and computer device and storage medium | |
| CN110675876A (en) | Intelligent control method, intelligent controller, system and storage medium for semantic recognition | |
| CN110992937B (en) | Language off-line identification method, terminal and readable storage medium | |
| US11468886B2 (en) | Artificial intelligence apparatus for performing voice control using voice extraction filter and method for the same | |
| US11678017B2 (en) | Electronic device and method for controlling the same, and storage medium | |
| US20230385377A1 (en) | Device, method, and computer program for performing actions on iot devices | |
| CN110597082A (en) | Intelligent household equipment control method and device, computer equipment and storage medium | |
| CN108758989A (en) | A kind of air-conditioning and its application method | |
| CN114925163B (en) | Intelligent device and model training method for intention recognition | |
| WO2020078076A1 (en) | Method and system for controlling air conditioner, air conditioner, and household appliance | |
| CN114676689A (en) | Sentence text recognition method and device, storage medium and electronic device | |
| CN108959889A (en) | A kind of Accreditation System and method of intelligent appliance | |
| CN111710339A (en) | Voice recognition interaction system and method based on data visualization display technology | |
| CN114302168B (en) | Learning video generation method, device, equipment and storage medium based on big data | |
| CN116566760B (en) | Smart home equipment control method and device, storage medium and electronic equipment | |
| CN110719512A (en) | Intelligent remote controller control method and device, intelligent remote controller and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |