CN107678309A - Manipulate clause generation, using control method and device, storage medium - Google Patents
Manipulate clause generation, using control method and device, storage medium Download PDFInfo
- Publication number
- CN107678309A CN107678309A CN201710781250.7A CN201710781250A CN107678309A CN 107678309 A CN107678309 A CN 107678309A CN 201710781250 A CN201710781250 A CN 201710781250A CN 107678309 A CN107678309 A CN 107678309A
- Authority
- CN
- China
- Prior art keywords
- entry
- newly
- increased
- clause
- manipulation
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
Abstract
The disclosure provides a kind of manipulation clause generation method and device, using control method and device, storage medium, electronic equipment.Wherein, manipulation clause generation method includes：Obtain and treat original entry corresponding to the function of manipulation application support, original entry includes original entity and/or original intention；By extension entry corresponding to original entry and original entry, it is defined as newly-increased entry；Can judgement match similar entry corresponding to the newly-increased entry in default clause storehouse；If matching similar entry corresponding to the newly-increased entry, the newly-increased entry is merged into the entry class belonging to the similar entry；The mapping relations between other entry classes in entry class and the default clause storehouse according to belonging to the similar entry, manipulation clause of the generation with the newly-increased entry.Such scheme, manipulation clause can be automatically generated, be favorably improved flexibility, the scalability of manipulation clause generation.
Description
Technical field
This disclosure relates to data processing field, in particular it relates to which a kind of manipulate clause generation method and device, using manipulation
Method and device, storage medium, electronic equipment.
Background technology
With the continuous progress of artificial intelligence technology, man-machine interaction also achieves significant progress, and various man-machine interactions are set
It is standby also to rise without restraint therewith.For example, human-computer interaction device can be mobile phone, smart home, robot, mobile unit etc..
Generally, when carrying out man-machine interaction, it is necessary to be application supported function in advance structure corresponding to manipulation clause, this
Sample, after getting human-machine interaction data, it can first judge whether the human-machine interaction data can match the manipulation sentence built in advance
Formula, if can match, then corresponding function is realized by the manipulation clause in matching, control application.For example, should for music
For, if the manipulation clause in matching is " being switched to next song ", music application can be controlled by the manipulation clause
Carry out song switching.
Such scheme is only applicable to have been built up the application function for manipulating clause, for not yet building answering for manipulation clause
With function, for example, emerging function or functional is new in the supported function of emerging application or application
Saying etc., because in the absence of the manipulation clause that can be matched, semantic understanding can not be carried out, it is difficult to realize application manipulation.
The content of the invention
It is a general object of the present disclosure to provide one kind manipulation clause generation method and device, using control method and device,
Storage medium, electronic equipment, manipulation clause can be automatically generated, be favorably improved the flexibility, expansible of manipulation clause generation
Property.
To achieve these goals, in a first aspect, the disclosure provides a kind of manipulation clause generation method, methods described bag
Include：
Obtain treat manipulation application support function corresponding to original entry, original entry include original entity and/or
Original intention；
By extension entry corresponding to original entry and original entry, it is defined as newly-increased entry；
Can judgement match similar entry corresponding to the newly-increased entry in default clause storehouse；
If matching similar entry corresponding to the newly-increased entry, the newly-increased entry is merged into the similar word
In entry class belonging to bar；
The mapping between other entry classes in entry class and the default clause storehouse according to belonging to the similar entry
Relation, manipulation clause of the generation with the newly-increased entry.
Alternatively, the mode of extension entry is corresponding to acquisition original entry：
Input using original entry as entry extended model, after entry extended model processing, export institute
State extension entry corresponding to original entry.
Alternatively, the mode for building the entry extended model is：
Training human-machine interaction data is obtained, and is every training growth data corresponding to human-machine interaction data marks,
The training includes training with manipulation clause and/or training entry with human-machine interaction data；
Determine the topological structure of the entry extended model；
Using the training human-machine interaction data and the topological structure, training obtains the entry extended model,
The growth data of the entry extended model output is set to comprise at least the growth data of mark.
Alternatively, it is described entry to be extended corresponding to original entry and original entry, it is defined as new epexegesis
Bar, including：
Calculate it is described extension entry and corresponding original entry between the first similarity, and it is described extension entry and its
The second similarity between his original entry；
Judge whether first similarity is more than second similarity；
If first similarity is more than second similarity, and first similarity and second similarity
Difference be not less than preset difference value, then retain the extension entry；
By original entry and the extension entry retained, it is defined as newly-increased entry.
Alternatively, methods described also includes：
If not matching similar entry corresponding to newly-increased entry, newly-increased entry is created in the default clause storehouse
Class, and the newly-increased entry is write into the newly-increased entry class；
The newly-increased entry class is established to the mapping relations of the specified entry class in the default clause storehouse, generation is with described
The manipulation clause of newly-increased entry.
Alternatively, the judgement can be matched in default clause storehouse similar entry corresponding to the newly-increased entry it
Before, methods described also includes：
Judge whether the occurrence number of the newly-increased entry is less than default frequency value；
If the occurrence number of the newly-increased entry is not less than default frequency value, then perform the judgement can be in default sentence
The step of similar entry corresponding to the newly-increased entry is matched in formula storehouse.
Alternatively, entry will be extended corresponding to original entry and original entry described, is defined as increasing newly
After entry, methods described also includes：
Clustering processing is carried out to the newly-increased entry, obtains at least one newly-increased entry class；
Then, can the judgement match similar entry corresponding to the newly-increased entry in default clause storehouse, including：Sentence
It is disconnected that similar entry class corresponding to the newly-increased entry class can be matched in default clause storehouse.
Second aspect, the disclosure provide one kind and apply control method, for the behaviour using the generation of first aspect methods described
Control clause include using manipulation, methods described：
Obtain user and be directed to the human-machine interaction data for treating manipulation application input, the human-machine interaction data is described for controlling
Treat that manipulation application performs and specify function；
The human-machine interaction data is identified, obtains text message, and utilize the text message and the manipulation
Clause is matched；
If matching manipulation clause corresponding to the text message, by described in the corresponding manipulation clause control
Treat that manipulation application performs and specify function.
The third aspect, the disclosure provide a kind of manipulation clause generating means, and described device includes：
Original entry acquisition module, original entry corresponding to the function of manipulation application support is treated for obtaining, it is described original
Entry includes original entity and/or original intention；
Newly-increased entry determining module, for entry will to be extended corresponding to original entry and original entry, really
It is set to newly-increased entry；
First judge module, for judging that similar word corresponding to the newly-increased entry can be matched in default clause storehouse
Bar；
Entry merging module, for when matching similar entry corresponding to the newly-increased entry, by the newly-increased entry
It is merged into the entry class belonging to the similar entry；
Clause generation module is manipulated, in the entry class according to belonging to the similar entry and the default clause storehouse
Mapping relations between other entry classes, manipulation clause of the generation with the newly-increased entry.
Alternatively, described device also includes：
Entry output module is extended, for the input using original entry as entry extended model, through the entry
After extended model processing, extension entry corresponding to original entry is exported.
Alternatively, described device also includes：
Human-machine interaction data labeling module, for obtaining training human-machine interaction data, and it is every man-machine friendship of training
Growth data corresponding to mutual data mark, training human-machine interaction data include training and used with manipulation clause and/or training
Entry；
Topological structure determining module, for determining the topological structure of the entry extended model；
Model training module, for being obtained using the training human-machine interaction data and the topological structure, training
The entry extended model, the growth data of the entry extended model output is set to comprise at least the growth data of mark.
Alternatively, the newly-increased entry determining module, for calculating between the extension entry and corresponding original entry
The first similarity, and it is described extension entry and other original entries between the second similarity；Judge that described first is similar
Whether degree is more than second similarity；If first similarity is more than second similarity, and described first similar
Degree and the difference of second similarity are not less than preset difference value, then retain the extension entry；By original entry and
The extension entry of reservation, it is defined as newly-increased entry.
Alternatively, described device also includes：
Entry class creation module, for when not matching similar entry corresponding to newly-increased entry, in the default clause
Newly-increased entry class is created in storehouse, and the newly-increased entry is write into the newly-increased entry class；
Mapping relations establish module, for establishing the newly-increased entry class to the specified entry class in the default clause storehouse
Mapping relations, manipulation clause of the generation with the newly-increased entry.
Alternatively, described device also includes：
Second judge module, for judging whether the occurrence number of the newly-increased entry is less than default frequency value；
First judge module, for when the occurrence number of the newly-increased entry is not less than default frequency value, judging
Similar entry corresponding to the newly-increased entry can be matched in default clause storehouse.
Alternatively, described device also includes：
Clustering processing module, for carrying out clustering processing to the newly-increased entry, obtain at least one newly-increased entry class；
First judge module, for judging to match corresponding to the newly-increased entry class in default clause storehouse
Similar entry class.
Fourth aspect, the disclosure provide one kind and apply actuation means, for the behaviour using the generation of third aspect described device
Control clause include using manipulation, described device：
Human-machine interaction data acquisition module, the human-machine interaction data for treating that manipulation application inputs, institute are directed to for obtaining user
State human-machine interaction data and be used to controlling and described treat that manipulation application performs and specify function；
Clause matching module is manipulated, for the human-machine interaction data to be identified, obtains text message, and utilize institute
Text message is stated to be matched with the manipulation clause；
Functional control module, for match corresponding to the text message manipulate clause when, by described corresponding
Treat that manipulation application performs described in manipulation clause control and specify function.
5th aspect, the disclosure provide a kind of storage medium, wherein being stored with a plurality of instruction, the instruction is added by processor
Carry, perform first aspect methods described the step of.
6th aspect, the disclosure provide a kind of storage medium, wherein being stored with a plurality of instruction, the instruction is added by processor
Carry, perform second aspect methods described the step of.
7th aspect, the disclosure provide a kind of electronic equipment, and the electronic equipment includes；
Storage medium described in 5th aspect；And
Processor, for performing the instruction in the storage medium.
Eighth aspect, the disclosure provide a kind of electronic equipment, it is characterised in that the electronic equipment includes；
Storage medium described in 6th aspect；And
Processor, for performing the instruction in the storage medium.
In disclosure scheme, can obtain treat manipulation application support function corresponding to original entry, and with original word
Extension entry corresponding to bar, as newly-increased entry, carry out manipulating clause generation.Specifically, can be matched in default clause storehouse
Go out the similar entry corresponding with newly-increased entry, and then newly-increased entry is merged into the entry class belonging to similar entry, in this way,
The mapping relations that can have according to the entry class belonging to the similar entry, generate new manipulation clause, i.e., with newly-increased entry
Manipulation clause.Such scheme, it is possible to achieve manipulate automatically generating for clause, no longer limited by new opplication, New function, new saying
It is fixed, flexibility, the scalability of manipulation clause generating process can be improved.
Other feature and advantage of the disclosure will be described in detail in subsequent specific embodiment part.
Brief description of the drawings
Accompanying drawing is for providing further understanding of the disclosure, and a part for constitution instruction, with following tool
Body embodiment is used to explain the disclosure together, but does not form the limitation to the disclosure.In the accompanying drawings：
Fig. 1 is the schematic diagram that clause storehouse is preset in disclosure scheme；
Fig. 2 is the schematic flow sheet that disclosure scheme manipulates clause generation method embodiment 1；
Fig. 3 is the schematic flow sheet that newly-increased entry is obtained in disclosure scheme；
Fig. 4 is the schematic flow sheet that disclosure scheme manipulates clause generation method embodiment 2；
Fig. 5 is the schematic flow sheet that entry extended model is built in disclosure scheme；
Fig. 6 is the schematic flow sheet of disclosure scheme application control method；
Fig. 7 is the composition schematic diagram that disclosure scheme manipulates clause generating means；
Fig. 8 is the composition schematic diagram of disclosure scheme application actuation means；
Fig. 9 is the structural representation of disclosure scheme electronic equipment.
Embodiment
The embodiment of the disclosure is described in detail below in conjunction with accompanying drawing.It should be appreciated that this place is retouched
The embodiment stated is merely to illustrate and explained the disclosure, is not limited to the disclosure.
Before disclosure scheme is introduced, first the default clause storehouse in disclosure scheme is explained.
Disclosure scheme can be based on default clause storehouse generation manipulation clause, in this way, human-computer interaction device can be to pass through
Clause is manipulated, the application run on control device performs corresponding function, for example, manipulation clause is " song for playing Liu De China ",
Music application can be controlled to play the song for specifying singer., can be by being intended to form with entity for manipulation clause, its
In, " broadcasting " is to be intended to, it can be understood as it is desirable that what function application performs；" Liu Dehua " is entity, it can be understood as
The object that function performs, by taking music application as an example, entity can be the name of song, the name of singer, the issuing date of song
Deng.
Default clause storehouse in disclosure scheme can be with as shown in figure 1, including at least one intention class, at least one entity
Class, and can generate manipulation clause intention class and entity class between establish have mapping relations.Such as Fig. 1 examples cited, can wrap
Include：
(1) play and be intended to class, i.e. play in Fig. 1.Such can include at least one intention for representing to play, and e.g., put
Once, play, I wants to listen；
(2) suspend and be intended to class, i.e. pause in Fig. 1.Such can include at least one intention for representing pause, e.g., temporarily
Stop, suspend, stop play etc.；
(3) the music artist in singer's entity class, i.e. Fig. 1.Such can include at least one expression artist name
Entity, e.g., Liu Dehua, schoolmate, Zhou Jielun etc.；
(4) the music song in song entity class, i.e. Fig. 1.Such can include at least one expression song names
Entity, e.g., lustily water, kiss goodbye, east wind is broken etc..
It is to be appreciated that the intention class in default clause storehouse can establish mapping relations with multiple entity class, or, entity
Class also can establish mapping relations with multiple intention classes, and disclosure scheme can be not specifically limited to this, can combine practical application
Depending on situation.For example, (1) can establish mapping relations with (3), and the manipulation sentence of the song of some singer is played for generating
Formula, (1) can also establish mapping relations with (4), and the manipulation clause of some song is played for generating.
As a kind of example, the various ways such as syntax rule, the artificial experiences such as the syntax, dictionary can be combined, create manipulation
Clause storehouse；Or can be combined with practical application and manipulation clause storehouse is updated, disclosure scheme can not do specific limit to this
It is fixed.
The implementation process for manipulating clause generation method to disclosure scheme below is explained.
Referring to Fig. 2, show that disclosure scheme manipulates the schematic flow sheet of clause generation method embodiment 1.It can include
Following steps：
S101, obtains original entry corresponding to the function for the treatment of manipulation application support, and original entry includes original entity
And/or original intention.
As a kind of example, it can be the application run on human-computer interaction device to treat manipulation application, for example, being run on mobile phone
Application, the application that runs on mobile unit etc., disclosure scheme can be not specifically limited to this.
In order to improve the flexibility of disclosure schemes generation manipulation clause, scalability, it can obtain and treat manipulation application branch
Entry corresponding to the function of holding, it is contemplated that in order to generate the higher manipulation clause of quality, support more liberalization, more diversified work(
Energy saying, subsequently can also carry out entry extension, therefore the entry now obtained can be referred to as into original entry.
Exemplified by treating that manipulation application is music application, collect search song, pause, next, a upper head, song title, song
Hand name etc., can be as original entry in disclosure scheme.That is, the original entry obtained can at least be presented as original reality
Body and/or original intention.
In actual application, being also possible to for collecting is original manipulation clause, for example, current interface displaying " whether
Play the song of Liu De China", corresponding to this, original entry can be extracted from original manipulation clause, carry out subsequent treatment,
Disclosure scheme can be not specifically limited to entry extraction process.
In addition, it is necessary to illustrate, original entry in disclosure scheme can be the visible entry of current interface, such as exist
When searching song, current interface shows TOP N search result；It can also be the related entry of backstage caching, such as when searching song, remove
Remaining search result outside the TOP N that current interface is shown.
In addition, it is necessary to illustrate, original entry in disclosure scheme can treat the function pair that manipulation application is supported
All entries answered；Or or treat manipulation application support function corresponding to particial entry, for example, treating manipulation application
There is the original entry of identical with other application, if original entry other application generate manipulation clause when, handled
Finish, then when manipulation application generates manipulation clause, can be weeded out.
S102, by extension entry corresponding to original entry and original entry, it is defined as newly-increased entry.
Such as introduction made above, in order to generate the higher manipulation clause of quality, more liberalization, more diversified function are supported
Saying, entry extension can be carried out on the basis of original entry, obtain extension entry corresponding to each original entry.
For example, original entry is the movie name of a standard, and corresponding extension entry can at least include：Film
Official name, the conventional referred to as, translated name in country variant area of user etc..
As a kind of example, disclosure scheme can build entry extending database in advance, and word is carried out based on the database
Bar extends., will or in order to more fully carry out entry extension, disclosure scheme can also build entry extended model in advance
Input of the original entry as entry extended model, after entry extended model processing, export original entry
Corresponding extension entry.On the building process of entry extended model, reference can be made to introduced at FIG. 5 below, herein wouldn't be detailed
State.
As a kind of example, after carrying out entry extension for different original entries, obtain identical or approximate
Entry is extended, disclosure scheme can also carry out filtration treatment to extension entry, it is ensured that have one between all newly-increased entries
Fixed otherness.
Referring to Fig. 3, the schematic flow sheet that newly-increased entry is obtained in disclosure scheme is shown.It may comprise steps of：
S201, calculate the first similarity between the extension entry and corresponding original entry, and the expansion word
The second similarity between bar and other original entries.
S202, judges whether first similarity is more than second similarity.
S203, if first similarity is more than second similarity, and first similarity and described second
The difference of similarity is not less than preset difference value, then retains the extension entry.
S204, by original entry and the extension entry retained, it is defined as newly-increased entry.
For example, extend to obtain entry A for original entry A1、A2、A3If A1The first similarity between A, no
More than A1, then can be by A with the second similarity between other original entry B1Filter out, not as newly-increased entry.If A2With A it
Between the first similarity, more than A2With the second similarity between other original entry C, but the first similarity is similar to second
The difference of degree is less than preset difference value, i.e., the two otherness it is smaller, then can be by A2Filter out.
Furthermore, it is necessary to explanation, can be according to all newly-increased entries of disclosure schemes generation in actual application
Corresponding manipulation clause；Or the newly-increased entry of low frequency can also be filtered out, only generate manipulation sentence corresponding to the newly-increased entry of high frequency
Formula.Corresponding to this, after obtaining newly-increased entry, it can first judge whether the occurrence number of the newly-increased entry is less than default frequency
Value；If the occurrence number of the newly-increased entry is less than default frequency value, illustrate that the newly-increased entry is low frequency entry, can be by
Filter out.That is, the newly-increased entry in subsequent processes is high frequency entry.
Can S103, judgement match similar entry corresponding to the newly-increased entry in default clause storehouse.
S104, if matching similar entry corresponding to the newly-increased entry, the newly-increased entry is merged into described
In entry class belonging to similar entry.
After obtaining newly-increased entry, it can be matched with the entry in default clause storehouse, can judgement match newly-increased entry
Corresponding similar entry, and then newly-increased entry is merged into the entry class belonging to similar entry, that is, judging to preset
The entry class of newly-increased entry ownership is found in clause storehouse.
As a kind of example, similar entry can be matched for newly-increased entry by Similarity Measure.For example, new epexegesis
Bar can carry out Similarity Measure with the entry in default clause storehouse, can also enter with the class name of the entry class in default clause storehouse
Row Similarity Measure, disclosure scheme can be not specifically limited to this.
For example, it is " singer " to increase entry newly, can carry out Similarity Measure with the class name of " singer's entity class ", also may be used
To carry out Similarity Measure with the entity in " singer's entity class ", certainly, increasing entry newly can also be with other in default clause storehouse
The class name of entry class, other entries carry out Similarity Measure, and then determine similar entry corresponding to newly-increased entry, herein no longer
Illustrate.
In disclosure scheme, the entry that similarity can be more than to preset value is defined as similar word corresponding to newly-increased entry
Bar, wherein, depending on preset value can combine practical application request, disclosure scheme can be not specifically limited.
As a kind of example, in order to simplify processing, after obtaining newly-increased entry, first newly-increased entry can be carried out at cluster
Reason, obtains at least one newly-increased entry class, is then that newly-increased entry class matches similar entry class in default clause storehouse, such as
This, all entries that can include whole newly-increased entry class, is merged into similar entry class.
For example, clustering processing, disclosure scheme can be carried out by k-means clustering algorithms, k nearest neighbor algorithm etc.
This can be not specifically limited.
As a kind of example, the end condition of clustering processing can be in disclosure scheme：Obtained newly-increased entry class
Average distance is more than first threshold between number is not more than the class for specifying number, increasing newly entry class, increases newly average in the class of entry class
Distance is less than Second Threshold etc., disclosure scheme to end condition, specify number, first threshold, Second Threshold etc. can not be done has
Body limits.
As a kind of example, similar entry class can be matched for newly-increased entry class by Similarity Measure.For example, can
Using the class name of the entry class in the class name of newly-increased entry class, default clause storehouse, to carry out Similarity Measure；It can also utilize new
At least one entry that entry class in all entries that epexegesis bar class includes, default clause storehouse includes, carries out Similarity Measure,
Disclosure scheme can be not specifically limited to this.
As a kind of example, if the entry class bag in all entries included using newly-increased entry class, default clause storehouse
At least one entry included, Similarity Measure is carried out, similarity average can be obtained, and similarity average is more than default average
Entry class, be defined as similar entry class corresponding to newly-increased entry class.
As a kind of example, if the entry class bag in all entries included using newly-increased entry class, default clause storehouse
At least one entry included, Similarity Measure is carried out, similarity can be obtained and be more than the entry number N of preset value, and N is more than
The entry class of preset number, it is defined as similar entry class corresponding to newly-increased entry class.
S105, between other entry classes in entry class and the default clause storehouse according to belonging to the similar entry
Mapping relations, manipulation clause of the generation with the newly-increased entry.
As introduced at figure 1 above, after newly-increased entry to be merged into the disclosure scheme entry class belonging to similar entry,
The mapping relations that can have based on the entry class belonging to similar entry, new manipulation clause is generated, i.e., with newly-increased entry
Manipulate clause.
For example, entry is increased newly as " it is first to come one ", after above-mentioned processing, newly-increased entry can be merged into broadcasting and be intended to
In class.In this way, the mapping relations for playing and being intended between class and singer's entity class, song entity class can be combined, generation manipulation sentence
Formula " song for carrying out a first XXX ", wherein XXX can be any entity in singer's entity class；Or generation manipulation clause " comes one
First YYY ", wherein YYY can be any entity in song entity class.
In summary, disclosure scheme can be entered automatically with reference to original entry corresponding to the function for the treatment of manipulation application support
Row manipulation clause generation, is no longer limited by new opplication, New function, new saying, can improve the spirit of manipulation clause generating process
Activity, scalability.
Referring to Fig. 4, show that disclosure scheme manipulates the schematic flow sheet of clause generation method embodiment 2.It can include
Following steps：
S301, obtains original entry corresponding to the function for the treatment of manipulation application support, and original entry includes original entity
And/or original intention.
S302, by extension entry corresponding to original entry and original entry, it is defined as newly-increased entry.
Can S303, judgement match similar entry corresponding to the newly-increased entry in default clause storehouse.
S301~S303 implementation process, it can refer to and introduced at S101~S103 above, here is omitted.
S304, if not matching similar entry corresponding to newly-increased entry, created in the default clause storehouse newly-increased
Entry class, and the newly-increased entry is write into the newly-increased entry class.
S305, the newly-increased entry class is established to the mapping relations of the specified entry class in the default clause storehouse, generates band
There is the manipulation clause of the newly-increased entry.
If not matching similar entry corresponding to newly-increased entry, that is, failing to find in default clause storehouse newly-increased
The entry class of entry ownership, then newly-increased entry class can be created in default clause storehouse, newly-increased entry is write in newly-increased entry class.
Meanwhile in order to generate new manipulation clause, it is also necessary to establish newly-increased entry class to the mapping relations for specifying entry class.
For example, it is " FM103.9 " to increase entry newly, and after above-mentioned processing, newly-increased entry can be write to newly-increased entry class, and " radio station is real
Radio name in body class ", i.e. Fig. 1.At this point it is possible to will play be intended to class, pause is intended to class and is defined as specifying entry class, build
Vertical mapping relations, " AAA FM103.9 ", wherein AAA can be to play any intention being intended in class to generation manipulation clause；Or
Person, " BBB FM103.9 ", wherein BBB can be any intention that pause is intended in class to generation manipulation clause.
In disclosure scheme, it can in several ways determine to specify entry class, be exemplified below：
Mode one, entry class all in default clause storehouse can be defined as specifying entry class, i.e. establish new epexegesis
The full connection of bar class；
It mode two, can dish out to user, determine at least one specified entry class from default clause storehouse by user；
Mode three, it can be determined according to default syntax rule, artificial experience etc. from default clause storehouse at least one
Specify entry class.
Such as introduction made above, disclosure scheme can provide a kind of scheme for building entry extended model, can specifically join
Schematic flow sheet as shown in Figure 5.It may comprise steps of：
S401, training human-machine interaction data is obtained, and to be extended corresponding to every training human-machine interaction data mark
Data, the training include training with manipulation clause and/or training entry with human-machine interaction data.
When carrying out model training, substantial amounts of training human-machine interaction data can be obtained, for example, training manipulation clause
And/or training entry, from described above, training can include training with intention and/or training entity with entry.
As a kind of example, human-machine interaction data can be speech data, and disclosure scheme can be not specifically limited to this.
For example, user wants control music application pause, can be behaviour with phonetic entry " I is not desired to listen song ", namely human-machine interaction data
Control clause；Or can also phonetic entry " pause ", namely human-machine interaction data is entry.
After obtaining training human-machine interaction data, processing can be labeled to it, is every training man-machine interaction number
According to marking out corresponding growth data, it is possible to understand that ground, training are with human-machine interaction data, the growth data marked out with it
Corresponding panel data.
S402, determine the topological structure of the entry extended model.
In disclosure scheme, topological structure can be DNN (English：Deep Neural Networks, Chinese：Depth god
Through network), RNN (English：Recurrent Neural Networks, Chinese：Recognition with Recurrent Neural Network), SVM (English：
Support Vector Machine, Chinese：SVMs) etc..By taking RNN as an example, hidden layer can be 3~7 layers, and node is
2048 or 1024 etc., disclosure scheme can be not specifically limited to this.
S403, using the training human-machine interaction data and the topological structure, training obtains the entry extension
Model, the growth data of the entry extended model output is set to comprise at least the growth data of mark.
After obtaining training human-machine interaction data, topological structure, model training can be carried out using the two, obtain entry expansion
Open up model.In disclosure scheme, the constraints of model training can be：The growth data of entry extended model output at least should
Growth data including mark, in this way, being favorably improved the comprehensive of entry extension.
In disclosure scheme, the training process of model can refer to correlation technique realization, be not detailed herein, for example, can be with
Train to obtain entry extended model using BP algorithm.
If it should be noted that in the model training stage, model training, i.e. model are carried out with manipulation clause using training
Input and output be all the data with clause, and in the model measurement stage, model measurement, i.e. mould are carried out with entry using training
The input and output of type are the data without clause.As in this case, although semantic difference may be made to a certain extent
The similar data of entry can not form panel data, for example, both " I will build annotation ", " I not build annotation " will not turn into flat
Row data, still, entry repugnance of statement and it is context-sensitive when, different sayings can be recalled according to contextual information
Entry, for example, input is " I wants newly-built annotation ", output can be " I thinks one annotation of editor ", the discovery to extending entry
Meaning is bigger.
In addition, the disclosure also provides a kind of manipulation clause based on above method generation, the scheme using manipulation, tool are carried out
Body can be found in schematic flow sheet shown in Fig. 6.It can include：
S501, obtain user and be directed to the human-machine interaction data for treating manipulation application input, the human-machine interaction data is used to control
Treat that manipulation application performs described in system and specify function.
S502, the human-machine interaction data is identified, obtains text message, and using the text message with it is described
Manipulation clause is matched.
S503, if matching manipulation clause corresponding to the text message, pass through the corresponding manipulation clause control
Treat that manipulation application performs described in system and specify function.
When user wants control when the specified function of manipulation application execution, for example, control music application plays Liu De China
Song, corresponding human-machine interaction data can be inputted.In this way, human-machine interaction data is identified, corresponding text envelope is obtained
After breath, all manipulation clause generated in clause storehouse can be preset with disclosure scheme and matched, if text can be matched
Manipulation clause corresponding to information, then by the corresponding manipulation clause music application can be controlled to play the song of Liu De China.
Disclosure scheme can be not detailed to the identification process of human-machine interaction data, be referred to correlation technique realization.
Referring to Fig. 7, show that the disclosure manipulates the composition schematic diagram of clause generating means.Described device can include：
Original entry acquisition module 601, original entry corresponding to the function of manipulation application support, the original are treated for obtaining
There is entry to include original entity and/or original intention；
Newly-increased entry determining module 602, for by expansion word corresponding to original entry and original entry
Bar, it is defined as newly-increased entry；
First judge module 603, for judging that phase corresponding to the newly-increased entry can be matched in default clause storehouse
Like entry；
Entry merging module 604, for when matching similar entry corresponding to the newly-increased entry, by the new epexegesis
Bar is merged into the entry class belonging to the similar entry；
Clause generation module 605 is manipulated, for the entry class according to belonging to the similar entry and the default clause storehouse
In other entry classes between mapping relations, generation with the newly-increased entry manipulation clause.
Alternatively, described device also includes：
Entry output module is extended, for the input using original entry as entry extended model, through the entry
After extended model processing, extension entry corresponding to original entry is exported.
Alternatively, described device also includes：
Human-machine interaction data labeling module, for obtaining training human-machine interaction data, and it is every man-machine friendship of training
Growth data corresponding to mutual data mark, training human-machine interaction data include training and used with manipulation clause and/or training
Entry；
Topological structure determining module, for determining the topological structure of the entry extended model；
Model training module, for being obtained using the training human-machine interaction data and the topological structure, training
The entry extended model, the growth data of the entry extended model output is set to comprise at least the growth data of mark.
Alternatively, the newly-increased entry determining module, for calculating between the extension entry and corresponding original entry
The first similarity, and it is described extension entry and other original entries between the second similarity；Judge that described first is similar
Whether degree is more than second similarity；If first similarity is more than second similarity, and described first similar
Degree and the difference of second similarity are not less than preset difference value, then retain the extension entry；By original entry and
The extension entry of reservation, it is defined as newly-increased entry.
Alternatively, described device also includes：
Entry class creation module, for when not matching similar entry corresponding to newly-increased entry, in the default clause
Newly-increased entry class is created in storehouse, and the newly-increased entry is write into the newly-increased entry class；
Mapping relations establish module, for establishing the newly-increased entry class to the specified entry class in the default clause storehouse
Mapping relations, manipulation clause of the generation with the newly-increased entry.
Alternatively, described device also includes：
Second judge module, for judging whether the occurrence number of the newly-increased entry is less than default frequency value；
First judge module, for when the occurrence number of the newly-increased entry is not less than default frequency value, judging
Similar entry corresponding to the newly-increased entry can be matched in default clause storehouse.
Alternatively, described device also includes：
Clustering processing module, for carrying out clustering processing to the newly-increased entry, obtain at least one newly-increased entry class；
First judge module, for judging to match corresponding to the newly-increased entry class in default clause storehouse
Similar entry class.
Referring to Fig. 8, the composition schematic diagram of disclosure application actuation means is shown, the device can be utilized shown in Fig. 7 and filled
The manipulation clause for putting generation is carried out using manipulation.Described device can include：
Human-machine interaction data acquisition module 701, the human-machine interaction data for treating that manipulation application inputs is directed to for obtaining user,
The human-machine interaction data, which is used to controlling, described to be treated that manipulation application performs and specifies function；
Clause matching module 702 is manipulated, for the human-machine interaction data to be identified, obtains text message, and profit
Matched with the text message with the manipulation clause；
Functional control module 703, for when matching manipulation clause corresponding to the text message, passing through the correspondence
Manipulation clause control described in treat manipulation application perform specify function.
On the device in above-described embodiment, wherein modules perform the concrete mode of operation in relevant this method
Embodiment in be described in detail, explanation will be not set forth in detail herein.
Referring to Fig. 9, the structural representation of electronic equipment 800 is shown.Reference picture 9, electronic equipment 800 include processing component
801, it further comprises one or more processors, and as the storage device resource representated by storage medium 802, for depositing
Storage can be by the instruction of the execution of processing component 801, such as application program.The application program stored in storage medium 802 can wrap
Include it is one or more each correspond to the module of one group of instruction.Refer in addition, processing component 801 is configured as execution
Order, to perform above-mentioned manipulation clause generation method or using control method.
Electronic equipment 800 can also include a power supply module 803, be configured as performing the power supply pipe of electronic equipment 800
Reason；One wired or wireless network interface 806, it is configured as electronic equipment 800 being connected to network；With an input and output
(I/O) interface 805.Electronic equipment 800 can be operated based on the operating system for being stored in storage medium 802, such as Windows
ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM or similar.
The preferred embodiment of the disclosure is described in detail above in association with accompanying drawing, still, the disclosure is not limited to above-mentioned reality
The detail in mode is applied, in the range of the technology design of the disclosure, a variety of letters can be carried out to the technical scheme of the disclosure
Monotropic type, these simple variants belong to the protection domain of the disclosure.
It is further to note that each particular technique feature described in above-mentioned embodiment, in not lance
In the case of shield, can be combined by any suitable means, in order to avoid unnecessary repetition, the disclosure to it is various can
The combination of energy no longer separately illustrates.
In addition, it can also be combined between a variety of embodiments of the disclosure, as long as it is without prejudice to originally
Disclosed thought, it should equally be considered as disclosure disclosure of that.
Claims (20)
1. one kind manipulation clause generation method, it is characterised in that methods described includes：
Obtain and treat original entry corresponding to function that manipulation application is supported, original entry includes original entity and/or original
It is intended to；
By extension entry corresponding to original entry and original entry, it is defined as newly-increased entry；
Can judgement match similar entry corresponding to the newly-increased entry in default clause storehouse；
If matching similar entry corresponding to the newly-increased entry, the newly-increased entry is merged into the similar entry institute
In the entry class of category；
The mapping relations between other entry classes in entry class and the default clause storehouse according to belonging to the similar entry,
Manipulation clause of the generation with the newly-increased entry.
2. according to the method for claim 1, it is characterised in that obtain the mode of extension entry corresponding to original entry
For：
Input using original entry as entry extended model, after entry extended model processing, export the original
There is extension entry corresponding to entry.
3. according to the method for claim 2, it is characterised in that the mode for building the entry extended model is：
Training human-machine interaction data is obtained, and is every training growth data corresponding to human-machine interaction data marks, it is described
Training includes training with manipulation clause and/or training entry with human-machine interaction data；
Determine the topological structure of the entry extended model；
Using the training human-machine interaction data and the topological structure, training obtains the entry extended model, makes institute
The growth data of predicate bar extended model output comprises at least the growth data of mark.
4. according to the method described in any one of claims 1 to 3, it is characterised in that described by original entry and described
Extension entry corresponding to original entry, it is defined as newly-increased entry, including：
Calculate the first similarity between the extension entry and corresponding original entry, and the extension entry and other originals
There is the second similarity between entry；
Judge whether first similarity is more than second similarity；
If first similarity is more than second similarity, and the difference of first similarity and second similarity
Value is not less than preset difference value, then retains the extension entry；
By original entry and the extension entry retained, it is defined as newly-increased entry.
5. according to the method described in any one of claims 1 to 3, it is characterised in that methods described also includes：
If not matching similar entry corresponding to newly-increased entry, newly-increased entry class is created in the default clause storehouse, and
The newly-increased entry is write into the newly-increased entry class；
The newly-increased entry class is established to the mapping relations of the specified entry class in the default clause storehouse, generation is with described newly-increased
The manipulation clause of entry.
6. according to the method described in any one of claims 1 to 3, it is characterised in that can be in default clause storehouse in the judgement
In match similar entry corresponding to the newly-increased entry before, methods described also includes：
Judge whether the occurrence number of the newly-increased entry is less than default frequency value；
If the occurrence number of the newly-increased entry is not less than default frequency value, then perform the judgement can be in default clause storehouse
In the step of matching similar entry corresponding to the newly-increased entry.
7. according to the method described in any one of claims 1 to 3, it is characterised in that described by original entry, Yi Jisuo
Extension entry corresponding to original entry is stated, after being defined as newly-increased entry, methods described also includes：
Clustering processing is carried out to the newly-increased entry, obtains at least one newly-increased entry class；
Then, can the judgement match similar entry corresponding to the newly-increased entry in default clause storehouse, including：Judge energy
It is no that similar entry class corresponding to the newly-increased entry class is matched in default clause storehouse.
8. one kind applies control method, it is characterised in that for the behaviour using the generation of any one of claim 1 to 7 methods described
Control clause include using manipulation, methods described：
Obtain user and be directed to the human-machine interaction data for treating manipulation application input, the human-machine interaction data, which is used to controlling, described to be waited to grasp
Control application, which performs, specifies function；
The human-machine interaction data is identified, obtains text message, and utilize the text message and the manipulation clause
Matched；
If matching manipulation clause corresponding to the text message, by waiting to grasp described in the corresponding manipulation clause control
Control application, which performs, specifies function.
9. one kind manipulation clause generating means, it is characterised in that described device includes：
Original entry acquisition module, original entry corresponding to the function of manipulation application support, original entry are treated for obtaining
Including original entity and/or original intention；
Newly-increased entry determining module, for will extend entry corresponding to original entry and original entry, is defined as
Newly-increased entry；
First judge module, for judging that similar entry corresponding to the newly-increased entry can be matched in default clause storehouse；
Entry merging module, for when matching similar entry corresponding to the newly-increased entry, the newly-increased entry to be merged
Into the entry class belonging to the similar entry；
Clause generation module is manipulated, for other in the entry class according to belonging to the similar entry and the default clause storehouse
Mapping relations between entry class, manipulation clause of the generation with the newly-increased entry.
10. device according to claim 9, it is characterised in that described device also includes：
Entry output module is extended, for the input using original entry as entry extended model, is extended through the entry
After model treatment, extension entry corresponding to original entry is exported.
11. device according to claim 10, it is characterised in that described device also includes：
Human-machine interaction data labeling module, for obtaining training human-machine interaction data, and it is every training man-machine interaction number
According to growth data corresponding to mark, the training includes training with manipulation clause and/or training entry with human-machine interaction data；
Topological structure determining module, for determining the topological structure of the entry extended model；
Model training module, for utilizing the training human-machine interaction data and the topological structure, training obtains described
Entry extended model, the growth data of the entry extended model output is set to comprise at least the growth data of mark.
12. according to the device described in any one of claim 9 to 11, it is characterised in that
The newly-increased entry determining module, it is similar with first between corresponding original entry for calculating the extension entry
The second similarity between degree, and the extension entry and other original entries；Judge whether first similarity is more than
Second similarity；If first similarity is more than second similarity, and first similarity and described the
The difference of two similarities is not less than preset difference value, then retains the extension entry；By original entry and the extension retained
Entry, it is defined as newly-increased entry.
13. according to the device described in any one of claim 9 to 11, it is characterised in that described device also includes：
Entry class creation module, for when not matching similar entry corresponding to newly-increased entry, in the default clause storehouse
Newly-increased entry class is created, and the newly-increased entry is write into the newly-increased entry class；
Mapping relations establish module, for establishing mapping of the newly-increased entry class to the specified entry class in the default clause storehouse
Relation, manipulation clause of the generation with the newly-increased entry.
14. according to the device described in any one of claim 9 to 11, it is characterised in that described device also includes：
Second judge module, for judging whether the occurrence number of the newly-increased entry is less than default frequency value；
First judge module, for when the occurrence number of the newly-increased entry is not less than default frequency value, can judgement
Similar entry corresponding to the newly-increased entry is matched in default clause storehouse.
15. according to the device described in any one of claim 9 to 11, it is characterised in that described device also includes：
Clustering processing module, for carrying out clustering processing to the newly-increased entry, obtain at least one newly-increased entry class；
First judge module, it can be matched for judgement in default clause storehouse similar corresponding to the newly-increased entry class
Entry class.
16. one kind applies actuation means, it is characterised in that for utilizing the generation of any one of claim 9 to 15 described device
Manipulation clause include using manipulation, described device：
Human-machine interaction data acquisition module, the human-machine interaction data for treating that manipulation application inputs, the people are directed to for obtaining user
Machine interaction data, which is used to controlling, described to be treated that manipulation application performs and specifies function；
Clause matching module is manipulated, for the human-machine interaction data to be identified, obtains text message, and utilize the text
This information is matched with the manipulation clause；
Functional control module, for when matching manipulation clause corresponding to the text message, passing through the corresponding manipulation
Treat that manipulation application performs described in clause control and specify function.
17. a kind of storage medium, wherein being stored with a plurality of instruction, it is characterised in that the instruction is loaded by processor, right of execution
Profit requires the step of any one of 1 to 7 methods described.
18. a kind of storage medium, wherein being stored with a plurality of instruction, it is characterised in that the instruction is loaded by processor, right of execution
Profit requires the step of 8 methods described.
19. a kind of electronic equipment, it is characterised in that the electronic equipment includes；
Storage medium described in claim 17；And
Processor, for performing the instruction in the storage medium.
20. a kind of electronic equipment, it is characterised in that the electronic equipment includes；
Storage medium described in claim 18；And
Processor, for performing the instruction in the storage medium.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710781250.7A CN107678309B (en) | 2017-09-01 | 2017-09-01 | Control sentence pattern generation and application control method and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710781250.7A CN107678309B (en) | 2017-09-01 | 2017-09-01 | Control sentence pattern generation and application control method and device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107678309A true CN107678309A (en) | 2018-02-09 |
| CN107678309B CN107678309B (en) | 2021-07-06 |
Family
ID=61135960
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710781250.7A Active CN107678309B (en) | 2017-09-01 | 2017-09-01 | Control sentence pattern generation and application control method and device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107678309B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108829814A (en) * | 2018-06-10 | 2018-11-16 | 张超 | A kind of the knowledge learning Asymptotical Method and device of speech interactive robot |
| CN110516175A (en) * | 2019-08-29 | 2019-11-29 | 秒针信息技术有限公司 | A kind of method, apparatus, equipment and the medium of determining user tag |
| CN111159350A (en) * | 2019-12-30 | 2020-05-15 | 科大讯飞股份有限公司 | User opinion mining and amplification method, device, terminal and storage medium |
Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0954599A (en) * | 1995-08-18 | 1997-02-25 | Meidensha Corp | Intonation control method in speech synthesis by rule and speech synthesizer by rule |
| KR980011006A (en) * | 1996-07-20 | 1998-04-30 | 구자홍 | A Linguistic Model Processing Method Using a Class that Can Process New Registered Words in Speech Recognition System |
| US6192337B1 (en) * | 1998-08-14 | 2001-02-20 | International Business Machines Corporation | Apparatus and methods for rejecting confusible words during training associated with a speech recognition system |
| US20020116194A1 (en) * | 2001-02-21 | 2002-08-22 | International Business Machines Corporation | Method for preserving contextual accuracy in an extendible speech recognition language model |
| CN1437090A (en) * | 2002-02-04 | 2003-08-20 | 松下电器产业株式会社 | Interface equipment, motion control method and image display method |
| CN102636171A (en) * | 2012-04-27 | 2012-08-15 | 深圳市凯立德科技股份有限公司 | Voice navigation method and device |
| CN102682763A (en) * | 2011-03-10 | 2012-09-19 | 北京三星通信技术研究有限公司 | Method, device and terminal for correcting named entity vocabularies in voice input text |
| CN102968987A (en) * | 2012-11-19 | 2013-03-13 | 百度在线网络技术（北京）有限公司 | Speech recognition method and system |
| CN103092928A (en) * | 2012-12-31 | 2013-05-08 | 安徽科大讯飞信息科技股份有限公司 | Voice query method and system |
| CN103337241A (en) * | 2013-06-09 | 2013-10-02 | 北京云知声信息技术有限公司 | Voice recognition method and device |
| CN103472990A (en) * | 2013-08-27 | 2013-12-25 | 小米科技有限责任公司 | Appliance, and method and device for controlling same |
| CN104281609A (en) * | 2013-07-08 | 2015-01-14 | 腾讯科技（深圳）有限公司 | Voice input instruction matching rule configuration method and device |
| CN104700834A (en) * | 2013-12-05 | 2015-06-10 | 联想（新加坡）私人有限公司 | Information processing method and information processing device |
| CN105100460A (en) * | 2015-07-09 | 2015-11-25 | 上海斐讯数据通信技术有限公司 | Method and system for controlling intelligent terminal by use of sound |
| CN105869640A (en) * | 2015-01-21 | 2016-08-17 | 上海墨百意信息科技有限公司 | Method and device for recognizing voice control instruction for entity in current page |
| CN106057199A (en) * | 2016-05-31 | 2016-10-26 | 广东美的制冷设备有限公司 | Control method, control device and terminal |
| CN106571140A (en) * | 2016-11-14 | 2017-04-19 | Tcl集团股份有限公司 | Electrical appliance intelligent control method based on voice meaning and electrical appliance intelligent control system thereof |
| CN106778817A (en) * | 2016-11-25 | 2017-05-31 | 杭州中奥科技有限公司 | A kind of automatic classification method of event |
-
2017
- 2017-09-01 CN CN201710781250.7A patent/CN107678309B/en active Active
Patent Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0954599A (en) * | 1995-08-18 | 1997-02-25 | Meidensha Corp | Intonation control method in speech synthesis by rule and speech synthesizer by rule |
| KR980011006A (en) * | 1996-07-20 | 1998-04-30 | 구자홍 | A Linguistic Model Processing Method Using a Class that Can Process New Registered Words in Speech Recognition System |
| US6192337B1 (en) * | 1998-08-14 | 2001-02-20 | International Business Machines Corporation | Apparatus and methods for rejecting confusible words during training associated with a speech recognition system |
| US20020116194A1 (en) * | 2001-02-21 | 2002-08-22 | International Business Machines Corporation | Method for preserving contextual accuracy in an extendible speech recognition language model |
| CN1437090A (en) * | 2002-02-04 | 2003-08-20 | 松下电器产业株式会社 | Interface equipment, motion control method and image display method |
| CN102682763A (en) * | 2011-03-10 | 2012-09-19 | 北京三星通信技术研究有限公司 | Method, device and terminal for correcting named entity vocabularies in voice input text |
| CN102636171A (en) * | 2012-04-27 | 2012-08-15 | 深圳市凯立德科技股份有限公司 | Voice navigation method and device |
| CN102968987A (en) * | 2012-11-19 | 2013-03-13 | 百度在线网络技术（北京）有限公司 | Speech recognition method and system |
| CN103092928A (en) * | 2012-12-31 | 2013-05-08 | 安徽科大讯飞信息科技股份有限公司 | Voice query method and system |
| CN103337241A (en) * | 2013-06-09 | 2013-10-02 | 北京云知声信息技术有限公司 | Voice recognition method and device |
| CN104281609A (en) * | 2013-07-08 | 2015-01-14 | 腾讯科技（深圳）有限公司 | Voice input instruction matching rule configuration method and device |
| CN103472990A (en) * | 2013-08-27 | 2013-12-25 | 小米科技有限责任公司 | Appliance, and method and device for controlling same |
| CN104700834A (en) * | 2013-12-05 | 2015-06-10 | 联想（新加坡）私人有限公司 | Information processing method and information processing device |
| CN105869640A (en) * | 2015-01-21 | 2016-08-17 | 上海墨百意信息科技有限公司 | Method and device for recognizing voice control instruction for entity in current page |
| CN105100460A (en) * | 2015-07-09 | 2015-11-25 | 上海斐讯数据通信技术有限公司 | Method and system for controlling intelligent terminal by use of sound |
| CN106057199A (en) * | 2016-05-31 | 2016-10-26 | 广东美的制冷设备有限公司 | Control method, control device and terminal |
| CN106571140A (en) * | 2016-11-14 | 2017-04-19 | Tcl集团股份有限公司 | Electrical appliance intelligent control method based on voice meaning and electrical appliance intelligent control system thereof |
| CN106778817A (en) * | 2016-11-25 | 2017-05-31 | 杭州中奥科技有限公司 | A kind of automatic classification method of event |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108829814A (en) * | 2018-06-10 | 2018-11-16 | 张超 | A kind of the knowledge learning Asymptotical Method and device of speech interactive robot |
| CN110516175A (en) * | 2019-08-29 | 2019-11-29 | 秒针信息技术有限公司 | A kind of method, apparatus, equipment and the medium of determining user tag |
| CN111159350A (en) * | 2019-12-30 | 2020-05-15 | 科大讯飞股份有限公司 | User opinion mining and amplification method, device, terminal and storage medium |
| CN111159350B (en) * | 2019-12-30 | 2022-12-06 | 科大讯飞股份有限公司 | User opinion mining and amplification method, device, terminal and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107678309B (en) | 2021-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109918680B (en) | Entity identification method and device and computer equipment | |
| CN104238991B (en) | Phonetic entry matching process and device | |
| CN106951468B (en) | Talk with generation method and device | |
| CN107437415B (en) | Intelligent voice interaction method and system | |
| CN104915340B (en) | Natural language question-answering method and device | |
| CN107833574A (en) | Method and apparatus for providing voice service | |
| CN110444191A (en) | A kind of method, the method and device of model training of prosody hierarchy mark | |
| WO2018019116A1 (en) | Natural language-based man-machine interaction method and system | |
| CN107678309A (en) | Manipulate clause generation, using control method and device, storage medium | |
| CN109410927A (en) | Offline order word parses the audio recognition method combined, device and system with cloud | |
| CN107633005A (en) | A kind of knowledge mapping structure, comparison system and method based on class teaching content | |
| CN106528522A (en) | Scenarized semantic comprehension and dialogue generation method and system | |
| CN109271498A (en) | The natural language interaction method and system of Virtual robot | |
| CN107633079B (en) | Vehicle-mounted natural language man-machine interaction algorithm based on database and neural network | |
| CN107589828A (en) | The man-machine interaction method and system of knowledge based collection of illustrative plates | |
| CN108538294B (en) | Voice interaction method and device | |
| CN111522909B (en) | Voice interaction method and server | |
| JP7300435B2 (en) | Methods, apparatus, electronics, and computer-readable storage media for voice interaction | |
| CN103543979A (en) | Voice outputting method, voice interaction method and electronic device | |
| CN110164415A (en) | A kind of recommended method based on speech recognition, device and medium | |
| CN107665188B (en) | Semantic understanding method and device | |
| CN108304424A (en) | Text key word extracting method and text key word extraction element | |
| CN105973257A (en) | Navigation method and device | |
| CN110428819A (en) | Decoding network generation method, audio recognition method, device, equipment and medium | |
| CN103885924A (en) | Field-adaptive automatic open class subtitle generating system and field-adaptive automatic open class subtitle generating method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |