CN107678309B - Control sentence pattern generation and application control method and device and storage medium - Google Patents
Control sentence pattern generation and application control method and device and storage medium Download PDFInfo
- Publication number
- CN107678309B CN107678309B CN201710781250.7A CN201710781250A CN107678309B CN 107678309 B CN107678309 B CN 107678309B CN 201710781250 A CN201710781250 A CN 201710781250A CN 107678309 B CN107678309 B CN 107678309B
- Authority
- CN
- China
- Prior art keywords
- entry
- newly added
- sentence pattern
- original
- vocabulary
- 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.)
- Active
Links
Images
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 control sentence pattern generation method and device, an application control method and device, a storage medium and an electronic device. The control sentence pattern generation method comprises the following steps: acquiring an original entry corresponding to a function supported by an application to be controlled, wherein the original entry comprises an original entity and/or an original intention; determining the original entry and the expansion entry corresponding to the original entry as a new entry; judging whether similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not; if the similar entry corresponding to the newly added entry is matched, merging the newly added entry into an entry class to which the similar entry belongs; and generating the control sentence pattern with the newly added vocabulary entry according to the mapping relation between the vocabulary entry class to which the similar vocabulary entry belongs and other vocabulary entry classes in the preset sentence pattern library. According to the scheme, the control sentence pattern can be automatically generated, and the flexibility and the expandability of generating the control sentence pattern are improved.
Description
Technical Field
The present disclosure relates to the field of data processing, and in particular, to a method and an apparatus for generating a control sentence pattern, a method and an apparatus for application control, a storage medium, and an electronic device.
Background
With the continuous progress of artificial intelligence technology, human-computer interaction has been developed, and various human-computer interaction devices are also promoted. For example, the human-computer interaction device may be a mobile phone, a smart home, a robot, a vehicle-mounted device, and the like.
Generally, when performing human-computer interaction, it is necessary to construct a corresponding control sentence pattern for a function supported by an application in advance, so that after human-computer interaction data is obtained, it can be determined whether the human-computer interaction data can be matched with the pre-constructed control sentence pattern, and if so, the application is controlled to implement a corresponding function through the control sentence pattern in the matching. For example, for a music application, if the manipulation period in the matching is "switch to the next song", the music application may be controlled to perform song switching by the manipulation period.
The above scheme is only applicable to the application functions for which the manipulation sentence patterns have been constructed, and for the application functions for which the manipulation sentence patterns have not been constructed, for example, functions supported by newly appearing applications, newly appearing functions in applications, new descriptions of existing functions, and the like, because there is no manipulation sentence pattern that can be matched, semantic understanding cannot be performed, and application manipulation is difficult to achieve.
Disclosure of Invention
The disclosure provides a method and an apparatus for generating a manipulation sentence pattern, a method and an apparatus for application manipulation, a storage medium, and an electronic device, which can automatically generate a manipulation sentence pattern and are helpful to improve the flexibility and expandability of generating the manipulation sentence pattern.
In order to achieve the above object, in a first aspect, the present disclosure provides a manipulation sentence generation method, including:
acquiring an original entry corresponding to a function supported by an application to be controlled, wherein the original entry comprises an original entity and/or an original intention;
determining the original entry and the expansion entry corresponding to the original entry as a new entry;
judging whether similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not;
if the similar entry corresponding to the newly added entry is matched, merging the newly added entry into an entry class to which the similar entry belongs;
and generating the control sentence pattern with the newly added vocabulary entry according to the mapping relation between the vocabulary entry class to which the similar vocabulary entry belongs and other vocabulary entry classes in the preset sentence pattern library.
Optionally, the manner of obtaining the extension entry corresponding to the original entry is as follows:
and taking the original entry as the input of an entry expansion model, and outputting the expansion entry corresponding to the original entry after the entry expansion model processes the entry.
Optionally, the manner of constructing the entry expansion model is as follows:
acquiring human-computer interaction data for training, and labeling corresponding extension data for each piece of human-computer interaction data for training, wherein the human-computer interaction data for training comprises control sentences for training and/or entries for training;
determining a topological structure of the entry expansion model;
and training to obtain the vocabulary entry extension model by utilizing the training human-computer interaction data and the topological structure, so that the extension data output by the vocabulary entry extension model at least comprise labeled extension data.
Optionally, the determining the original entry and the extended entry corresponding to the original entry as a new entry includes:
calculating a first similarity between the expansion entry and the corresponding original entry and a second similarity between the expansion entry and other original entries;
judging whether the first similarity is greater than the second similarity;
if the first similarity is greater than the second similarity, and the difference value between the first similarity and the second similarity is not less than a preset difference value, the expansion entry is reserved;
and determining the original entry and the reserved expansion entry as new entries.
Optionally, the method further comprises:
if the similar vocabulary entry corresponding to the newly added vocabulary entry is not matched, creating a newly added vocabulary entry class in the preset sentence pattern library, and writing the newly added vocabulary entry into the newly added vocabulary entry class;
and establishing a mapping relation from the newly added vocabulary entry class to the specified vocabulary entry class of the preset sentence pattern library, and generating the control sentence pattern with the newly added vocabulary entry.
Optionally, before the determining whether the similar vocabulary entry corresponding to the newly added vocabulary entry can be matched in the preset sentence pattern library, the method further includes:
judging whether the occurrence frequency of the newly added entry is smaller than a preset frequency value or not;
and if the occurrence frequency of the newly added entry is not less than the preset frequency value, executing the step of judging whether the similar entry corresponding to the newly added entry can be matched in a preset sentence pattern library or not.
Optionally, after the determining the original entry and the extended entry corresponding to the original entry as the new entry, the method further includes:
clustering the newly added entries to obtain at least one newly added entry class;
then, the judging whether the similar vocabulary entry corresponding to the newly added vocabulary entry can be matched in a preset sentence pattern library includes: and judging whether the similar entry class corresponding to the newly added entry class can be matched in a preset sentence pattern library.
In a second aspect, the present disclosure provides an application manipulation method for performing application manipulation using a manipulation sentence pattern generated by the method in the first aspect, the method including:
acquiring human-computer interaction data input by a user aiming at an application to be controlled, wherein the human-computer interaction data is used for controlling the application to be controlled to execute a specified function;
identifying the human-computer interaction data to obtain text information, and matching the text information with the control sentence pattern;
and if the control sentence pattern corresponding to the text information is matched, controlling the application to be controlled to execute the specified function through the corresponding control sentence pattern.
In a third aspect, the present disclosure provides a manipulation sentence pattern generation apparatus, the apparatus comprising:
the system comprises an original entry acquisition module, a control module and a display module, wherein the original entry acquisition module is used for acquiring an original entry corresponding to a function supported by an application to be controlled, and the original entry comprises an original entity and/or an original intention;
a newly added entry determining module, configured to determine the original entry and an extended entry corresponding to the original entry as a newly added entry;
the first judging module is used for judging whether the similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not;
the entry merging module is used for merging the newly added entry into an entry class to which the similar entry belongs when the similar entry corresponding to the newly added entry is matched;
and the control sentence pattern generating module is used for generating the control sentence pattern with the newly added vocabulary entry according to the mapping relation between the vocabulary entry class to which the similar vocabulary entry belongs and other vocabulary entry classes in the preset sentence pattern library.
Optionally, the apparatus further comprises:
and the extended entry output module is used for taking the original entry as the input of an entry extended model, and outputting the extended entry corresponding to the original entry after the entry extended model processes the original entry.
Optionally, the apparatus further comprises:
the human-computer interaction data labeling module is used for acquiring human-computer interaction data for training and labeling corresponding extension data for each piece of human-computer interaction data for training, wherein the human-computer interaction data for training comprises control sentences for training and/or entries for training;
the topological structure determining module is used for determining the topological structure of the entry expansion model;
and the model training module is used for training to obtain the entry extension model by utilizing the training human-computer interaction data and the topological structure, so that the extension data output by the entry extension model at least comprises labeled extension data.
Optionally, the newly added entry determining module is configured to calculate a first similarity between the extended entry and the corresponding original entry, and a second similarity between the extended entry and another original entry; judging whether the first similarity is greater than the second similarity; if the first similarity is greater than the second similarity, and the difference value between the first similarity and the second similarity is not less than a preset difference value, the expansion entry is reserved; and determining the original entry and the reserved expansion entry as new entries.
Optionally, the apparatus further comprises:
the vocabulary entry class creating module is used for creating a new vocabulary entry class in the preset sentence pattern library and writing the new vocabulary entry into the new vocabulary entry class when the similar vocabulary entry corresponding to the new vocabulary entry is not matched;
and the mapping relation establishing module is used for establishing the mapping relation from the newly added vocabulary entry class to the specified vocabulary entry class of the preset sentence pattern library and generating the control sentence pattern with the newly added vocabulary entry.
Optionally, the apparatus further comprises:
the second judgment module is used for judging whether the occurrence frequency of the newly added entry is smaller than a preset frequency value or not;
and the first judging module is used for judging whether the similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not when the occurrence frequency of the newly added entries is not less than a preset frequency value.
Optionally, the apparatus further comprises:
the clustering processing module is used for clustering the newly added entry to obtain at least one newly added entry class;
and the first judging module is used for judging whether the similar entry class corresponding to the newly added entry class can be matched in a preset sentence pattern library.
In a fourth aspect, the present disclosure provides an application manipulation device for performing application manipulation using a manipulation sentence pattern generated by the device in the third aspect, the device including:
the system comprises a human-computer interaction data acquisition module, a control module and a display module, wherein the human-computer interaction data acquisition module is used for acquiring human-computer interaction data input by a user aiming at an application to be controlled, and the human-computer interaction data is used for controlling the application to be controlled to execute a specified function;
the control sentence pattern matching module is used for identifying the human-computer interaction data to obtain text information and matching the text information with the control sentence pattern;
and the function control module is used for controlling the application to be controlled to execute the specified function through the corresponding control sentence pattern when the control sentence pattern corresponding to the text information is matched.
In a fifth aspect, the present disclosure provides a storage medium having stored therein a plurality of instructions, which are loaded by a processor, for performing the steps of the method of the first aspect.
In a sixth aspect, the present disclosure provides a storage medium having stored therein a plurality of instructions, which are loaded by a processor, for performing the steps of the method of the second aspect.
In a seventh aspect, the present disclosure provides an electronic device comprising;
the storage medium of the fifth aspect; and
a processor to execute the instructions in the storage medium.
In an eighth aspect, the present disclosure provides an electronic device, comprising;
the storage medium of the sixth aspect; and
a processor to execute the instructions in the storage medium.
In the scheme, the original entry corresponding to the function supported by the application to be controlled can be obtained, and the expansion entry corresponding to the original entry is used as a new entry to generate the control sentence pattern. Specifically, similar entries corresponding to the newly added entries may be matched in a preset sentence pattern library, and the newly added entries are merged into entry classes to which the similar entries belong, so that a new control sentence pattern, that is, a control sentence pattern with the newly added entries, may be generated according to a mapping relationship of the entry classes to which the similar entries belong. According to the scheme, the automatic generation of the control sentence pattern can be realized, the limitation of new application, new functions and new language methods is avoided, and the flexibility and the expandability of the control sentence pattern generation process can be improved.
Additional features and advantages of the disclosure will be set forth in the detailed description which follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure without limiting the disclosure. In the drawings:
FIG. 1 is a diagram illustrating a predefined sentence pattern library in accordance with the disclosed embodiments;
FIG. 2 is a schematic flow chart of embodiment 1 of the sentence generation control method according to the present disclosure;
FIG. 3 is a schematic flow chart of obtaining new entries according to the present disclosure;
FIG. 4 is a schematic flow chart of embodiment 2 of the sentence generation control method according to the present disclosure;
FIG. 5 is a schematic flow chart of the construction of the entry expansion model according to the present disclosure;
FIG. 6 is a schematic flow chart of an application control method according to the present disclosure;
FIG. 7 is a schematic diagram of a sentence pattern generator controlled by the present disclosure;
FIG. 8 is a schematic diagram of an operation device according to the present disclosure;
fig. 9 is a schematic structural diagram of an electronic device according to the present disclosure.
Detailed Description
The following detailed description of specific embodiments of the present disclosure is provided in connection with the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present disclosure, are given by way of illustration and explanation only, not limitation.
Before introducing the present disclosure, a preset sentence library in the present disclosure is explained.
The scheme disclosed by the disclosure can generate the control sentence pattern based on the preset sentence pattern library, so that the man-machine interaction device can control the application running on the device to execute the corresponding function by controlling the sentence pattern, for example, the control sentence pattern is 'playing songs of Liudebua', and the music application can be controlled to play songs of a designated singer. For manipulation sentences, it can be composed of intents and entities, where "play" is an intention, and it can be understood what function an application is expected to perform; "Liu de Hua" is an entity, and can be understood as an object for executing a function, and taking a music application as an example, the entity can be the name of a song, the name of a singer, the release time of the song, and the like.
The preset sentence pattern library in the present disclosure may be as shown in fig. 1, and includes at least one intention class and at least one entity class, and a mapping relationship is established between the intention class and the entity class capable of generating the manipulation sentence pattern. As illustrated in fig. 1, may include:
(1) the intent class, play in FIG. 1. The class may include at least one intent to represent play, e.g., put, play, i want to hear, etc.;
(2) pause intention class, pause in FIG. 1. The class may include at least one intent to indicate a pause, e.g., pause, stop playing, etc.;
(3) singer entity class, namely music artist in FIG. 1. The class may include at least one entity representing the name of the singer, e.g., Liudebua, Zhang schoolfellow, Zhouygeren, etc.;
(4) the song entity class, music song in FIG. 1. The class may include at least one entity representing the name of the song, such as forgetting water, kissing, Dongfeng, etc.
It should be understood that the intent classes in the preset sentence library may establish a mapping relationship with a plurality of entity classes, or the entity classes may also establish a mapping relationship with a plurality of intent classes. For example, (1) a mapping relationship may be established with (3) for generating a manipulation sentence for playing a song of a singer, and (1) a mapping relationship may also be established with (4) for generating a manipulation sentence for playing a song.
As an example, the sentence manipulation library can be created by combining grammatical rules such as grammars and dictionaries, artificial experiences and other modes; alternatively, the manipulation sentence pattern library may also be updated in combination with the actual application, and the scheme of the present disclosure may not be specifically limited to this.
The following explains the implementation process of the manipulation sentence generation method in the solution of the present disclosure.
Referring to fig. 2, a flow chart of embodiment 1 of the method for generating a manipulation sentence according to the present disclosure is shown. May include the steps of:
s101, original entries corresponding to functions supported by the application to be controlled are obtained, and the original entries comprise original entities and/or original intentions.
As an example, the application to be controlled may be an application running on the human-computer interaction device, for example, an application running on a mobile phone, an application running on a vehicle-mounted device, and the like, which may not be specifically limited in this disclosure.
In order to improve the flexibility and the expandability of generating the control sentence pattern in the scheme, entries corresponding to functions supported by an application to be controlled can be obtained, and in order to generate the control sentence pattern with higher quality, more free and diversified functional expressions are supported, and entry expansion is carried out subsequently, so that the entries obtained at the moment can be called as original entries.
Taking the example that the application to be controlled is a music application, the collected song search, pause, next, previous, song title, singer name, etc. can be used as the original entries in the scheme of the disclosure. That is, the obtained original entry may be at least embodied as an original entity and/or an original intention.
In the actual application process, the collected sentence patterns may be original control sentence patterns, for example, the current interface shows "play songs of liu de hua? Corresponding to this, the original vocabulary entry may be extracted from the original control sentence pattern for subsequent processing.
In addition, it should be noted that the original entry in the present disclosure may be an entry visible on the current interface, for example, when searching for a song, the current interface displays the search result of TOP N; or related entries cached in the background, such as the rest of the search results except for TOP N displayed on the current interface when searching songs.
In addition, it should be noted that the original entries in the present disclosure may be all entries corresponding to functions supported by the application to be controlled; or, the entry may also be a partial entry corresponding to a function supported by the application to be controlled, for example, the application to be controlled has the same original entry as other applications, and if the original entry is already processed when the other applications generate the control sentence pattern, the original entry may be removed when the application to be controlled generates the control sentence pattern.
S102, determining the original entry and the expansion entry corresponding to the original entry as a new entry.
As described above, in order to generate a higher-quality control sentence pattern and support a more free and diversified functional expression, the entry may be expanded based on the original entries to obtain the expanded entry corresponding to each original entry.
For example, the original entry is a standard movie name, and the corresponding extension entry at least includes: official name of the movie, common abbreviation of the user, translation names of different countries and regions, and the like.
As an example, the disclosed solution may pre-construct an entry expansion database, and perform entry expansion based on the database. Or, in order to perform entry expansion more comprehensively, the disclosed scheme may also pre-construct an entry expansion model, take the original entry as an input of the entry expansion model, and output an expansion entry corresponding to the original entry after being processed by the entry expansion model. For the construction process of the term expansion model, reference is made to the description in fig. 5 below, and details thereof will not be provided here.
As an example, in order to avoid obtaining the same or similar expanded entries after performing entry expansion on different original entries, the scheme of the present disclosure may further perform filtering processing on the expanded entries, so as to ensure that all the newly added entries have a certain difference.
Referring to fig. 3, a schematic flow chart of obtaining a new entry in the present disclosure is shown. May include the steps of:
s201, calculating a first similarity between the extension entry and the corresponding original entry, and a second similarity between the extension entry and other original entries.
S202, judging whether the first similarity is larger than the second similarity.
S203, if the first similarity is greater than the second similarity and the difference between the first similarity and the second similarity is not less than a preset difference, the expansion entry is reserved.
And S204, determining the original entry and the reserved expansion entries as new entries.
For example, for the original entryA expansion obtains entry A1、A2、A3If A is1A first similarity with A of not more than A1A second similarity with other original entry B, then A can be determined1And filtering the new entry not serving as the new entry. If A2A first similarity with A, greater than A2The second similarity with other original entry C, but the difference between the first similarity and the second similarity is smaller than the preset difference, namely the difference between the first similarity and the second similarity is smaller, then A can be used2And filtering out.
In addition, it should be noted that, in the practical application process, all the control sentence patterns corresponding to the newly added vocabulary entries can be generated according to the scheme disclosed herein; or, the low-frequency newly added vocabulary entry can be filtered out, and only the control sentence pattern corresponding to the high-frequency newly added vocabulary entry is generated. Correspondingly, after the newly added entry is obtained, whether the occurrence frequency of the newly added entry is smaller than a preset frequency value or not can be judged; and if the occurrence frequency of the newly added entry is smaller than the preset frequency value, the newly added entry is a low-frequency entry and can be filtered. That is, the newly added entries in the subsequent processing are all high-frequency entries.
And S103, judging whether the similar vocabulary entry corresponding to the newly added vocabulary entry can be matched in a preset sentence pattern library.
And S104, if the similar entry corresponding to the newly added entry is matched, merging the newly added entry into the entry class to which the similar entry belongs.
After the new entry is obtained, the new entry can be matched with the entries in the preset sentence pattern library, whether the similar entry corresponding to the new entry can be matched is judged, and then the new entry is merged into the entry class to which the similar entry belongs, that is, whether the entry class to which the new entry belongs can be found in the preset sentence pattern library is judged.
As an example, similar entries may be matched for the newly added entry through similarity calculation. For example, the newly added vocabulary entry may perform similarity calculation with a vocabulary entry in the preset sentence pattern library, or may perform similarity calculation with a class name of a vocabulary entry class in the preset sentence pattern library, which is not specifically limited in the present disclosure.
For example, the newly added entry is a singer, and may perform similarity calculation with the class name of the singer entity class, or may perform similarity calculation with the entity in the singer entity class, and of course, the newly added entry may also perform similarity calculation with the class names of other entry classes and other entries in the preset sentence library, so as to determine the similar entry corresponding to the newly added entry, which is not illustrated here.
In the scheme disclosed by the disclosure, the entry with the similarity greater than the preset value can be determined as the similar entry corresponding to the newly added entry, wherein the preset value can be determined by combining with the actual application requirements, and the scheme disclosed by the disclosure may not be specifically limited.
As an example, in order to simplify the processing, after the new entries are obtained, clustering processing may be performed on the new entries to obtain at least one new entry class, and then similar entry classes are matched for the new entry classes in a preset sentence pattern library, so that all entries included in the whole new entry class may be merged into the similar entry classes.
For example, the clustering process may be performed by a K-means clustering algorithm, a K nearest neighbor algorithm, and the like, which is not particularly limited in the present disclosure.
As an example, the termination condition of the clustering process in the present disclosure may be: the number of the newly added entry classes is not more than the specified number, the average distance between the newly added entry classes is greater than a first threshold, the average distance in the newly added entry classes is less than a second threshold, and the like.
As an example, similar vocabulary entry classes can be matched for the newly added vocabulary entry classes through similarity calculation. For example, the similarity calculation may be performed by using the class name of the newly added entry class and the class name of the entry class in the preset period library; similarity calculation can also be performed by using all the entries included in the newly added entry class and at least one entry included in the entry class in the preset sentence pattern library, which is not specifically limited in the present disclosure.
As an example, if similarity calculation is performed using all the entries included in the new entry class and at least one entry included in the entry class in the preset period library, a similarity mean may be obtained, and the entry class whose similarity mean is greater than the preset mean is determined as the similar entry class corresponding to the new entry class.
As an example, if similarity calculation is performed by using all the entries included in the new entry class and at least one entry included in the entry class in the preset sentence library, the number N of entries with similarity greater than the preset value may be obtained, and the entry class with N greater than the preset number is determined as the similar entry class corresponding to the new entry class.
And S105, generating the control sentence pattern with the newly added vocabulary entry according to the mapping relation between the vocabulary entry class to which the similar vocabulary entry belongs and other vocabulary entry classes in the preset sentence pattern library.
As described in fig. 1, the new entry is merged into the entry class to which the similar entry belongs according to the present disclosure, and then a new control sentence pattern, that is, a control sentence pattern with the new entry, is generated based on the mapping relationship between the entry classes to which the similar entry belongs.
For example, the newly added entry is "come one", and after the above processing, the newly added entry can be merged into the playing intention class. Thus, the mapping relation between the playing intention class and the singer entity class and the song entity class can be combined to generate a control sentence pattern 'Laiyang of XXX', wherein XXX can be any entity in the singer entity class; alternatively, a manipulation sentence pattern "Laiyy" is generated, where YYY may be any entity in the entity class of songs.
In summary, the scheme of the present disclosure can automatically generate the control sentence pattern in combination with the original vocabulary entry corresponding to the function supported by the application to be controlled, and is not limited by the new application, the new function, and the new grammar, so that the flexibility and the expandability of the control sentence pattern generation process can be improved.
Referring to fig. 4, a flow chart of embodiment 2 of the method for generating a manipulation sentence according to the present disclosure is shown. May include the steps of:
s301, original entries corresponding to functions supported by the application to be controlled are obtained, and the original entries comprise original entities and/or original intentions.
S302, determining the original entry and the expansion entry corresponding to the original entry as a new entry.
And S303, judging whether the similar vocabulary entry corresponding to the newly added vocabulary entry can be matched in a preset sentence pattern library.
The implementation processes of S301 to S303 can refer to the descriptions of S101 to S103, and are not described herein again.
S304, if the similar vocabulary entry corresponding to the newly added vocabulary entry is not matched, creating a newly added vocabulary entry class in the preset sentence pattern library, and writing the newly added vocabulary entry into the newly added vocabulary entry class.
S305, establishing a mapping relation from the newly added vocabulary entry class to the specified vocabulary entry class of the preset sentence pattern library, and generating the control sentence pattern with the newly added vocabulary entry.
If the similar entry corresponding to the newly added entry is not matched, that is, the entry class to which the newly added entry belongs cannot be found in the preset sentence pattern library, the newly added entry class can be created in the preset sentence pattern library, and the newly added entry is written into the newly added entry class.
Meanwhile, in order to generate a new control sentence pattern, a mapping relationship from the newly added entry class to the specified entry class needs to be established. For example, the newly added entry is "FM 103.9", and after the above processing, the newly added entry may be written into the newly added entry class "station entity class", i.e., the radio name in fig. 1. At this time, the playing intention class and the pause intention class can be determined as the specified entry class, a mapping relation is established, and a control sentence pattern "AAA FM 103.9" is generated, wherein AAA can be any intention in the playing intention class; alternatively, a maneuver period "BBB FM 103.9" is generated, where BBB may be any intent in the pause intent class.
In the present disclosure, the specified entry class may be determined in various ways, which is exemplified as follows:
in the first mode, all the vocabulary entry classes in the preset sentence pattern library can be determined as the appointed vocabulary entry classes, namely, the full connection of the newly added vocabulary entry classes is established;
in a second mode, a user can be given out, and the user determines at least one specified vocabulary entry class from the preset sentence pattern library;
and thirdly, determining at least one appointed vocabulary entry class from the preset sentence pattern library according to preset grammar rules, human experience and the like.
As described above, the present disclosure may provide a scheme for constructing an entry expansion model, and refer to the flowchart shown in fig. 5 specifically. May include the steps of:
s401, acquiring training human-computer interaction data, and labeling corresponding extension data for each piece of training human-computer interaction data, wherein the training human-computer interaction data comprises a training control sentence pattern and/or a training vocabulary entry.
In model training, a large amount of training human-computer interaction data, such as training control sentences and/or training vocabulary entries, may be obtained, and as is clear from the above description, the training vocabulary entries may include training intents and/or training entities.
As an example, the human-computer interaction data may be voice data, which may not be particularly limited by the present disclosure. For example, a user may input "i do not want to listen to a song" by voice, that is, man-machine interaction data is a control sentence pattern, in order to control the music application to pause; or, the voice input 'pause' can be used, that is, the man-machine interaction data is used as an entry.
After the training human-computer interaction data is obtained, the training human-computer interaction data can be labeled, corresponding extension data is labeled for each piece of training human-computer interaction data, and understandably, the training human-computer interaction data and the labeled extension data are corresponding parallel data.
S402, determining the topological structure of the vocabulary entry expansion model.
In the present disclosure, the topology may be DNN (Deep Neural Networks, chinese), RNN (Recurrent Neural Networks, chinese), SVM (Support Vector Machine, chinese), and the like. Taking RNN as an example, the hidden layer may be 3-7 layers, and the node is 2048 or 1024, etc., which is not specifically limited in this disclosure.
And S403, training to obtain the vocabulary entry extension model by using the training human-computer interaction data and the topological structure, so that the extension data output by the vocabulary entry extension model at least comprise labeled extension data.
After the human-computer interaction data and the topological structure for training are obtained, the two data can be used for model training to obtain an entry expansion model. In the scheme of the present disclosure, the constraint conditions of model training may be: the extension data output by the entry extension model at least comprises tagged extension data, so that the comprehensiveness of entry extension is improved.
In the present disclosure, the training process of the model may be implemented with reference to related technologies, which are not described in detail herein, for example, the entry expansion model may be obtained by training with a BP algorithm.
It should be noted that, if in the model training stage, the model training is performed by using the control sentence pattern, i.e. the input and output of the model are data with sentence pattern, and in the model testing stage, the model testing is performed by using the training vocabulary, i.e. the input and output of the model are data without sentence pattern. In this case, although data with different semantics but similar terms may not form parallel data to some extent, for example, "i want to build an annotation" and "i don't build an annotation" may not become parallel data, when the terms are inconsistent and context-dependent, terms with different descriptions may be recalled according to context information, for example, "i want to build an annotation", and "i want to edit an annotation" may be output, which is more meaningful for finding an expanded term.
In addition, the present disclosure also provides a scheme for performing application control based on the control sentence pattern generated by the above method, which may specifically refer to the flow diagram shown in fig. 6. The method can comprise the following steps:
s501, human-computer interaction data input by a user aiming at an application to be controlled are obtained, and the human-computer interaction data are used for controlling the application to be controlled to execute a specified function.
And S502, identifying the human-computer interaction data to obtain text information, and matching the text information with the control sentence pattern.
And S503, if the control sentence pattern corresponding to the text information is matched, controlling the application to be controlled to execute the specified function through the corresponding control sentence pattern.
When a user wants to control the application to be controlled to execute a specified function, for example, control a music application to play a song in liudeluxe, corresponding human-computer interaction data can be input. Therefore, after the man-machine interaction data is identified and the corresponding text information is obtained, the man-machine interaction data can be matched with all the control sentence patterns generated in the preset sentence pattern library of the scheme, and if the control sentence patterns corresponding to the text information can be matched, the music application can be controlled to play the songs in Liu De Hua through the corresponding control sentence patterns.
The identification process of the human-computer interaction data by the scheme of the disclosure can be realized by referring to related technologies without detailed description.
Referring to fig. 7, a schematic diagram of the composition of the manipulation sentence generating apparatus of the present disclosure is shown. The apparatus may include:
an original entry obtaining module 601, configured to obtain an original entry corresponding to a function supported by an application to be controlled, where the original entry includes an original entity and/or an original intention;
a newly added entry determining module 602, configured to determine the original entry and an extended entry corresponding to the original entry as a newly added entry;
a first judging module 603, configured to judge whether a similar entry corresponding to the newly added entry can be matched in a preset sentence pattern library;
an entry merging module 604, configured to merge the newly added entry into an entry class to which the similar entry belongs when the similar entry corresponding to the newly added entry is matched;
and a control sentence pattern generating module 605, configured to generate a control sentence pattern with the newly added entry according to a mapping relationship between the entry class to which the similar entry belongs and the other entry classes in the preset sentence pattern library.
Optionally, the apparatus further comprises:
and the extended entry output module is used for taking the original entry as the input of an entry extended model, and outputting the extended entry corresponding to the original entry after the entry extended model processes the original entry.
Optionally, the apparatus further comprises:
the human-computer interaction data labeling module is used for acquiring human-computer interaction data for training and labeling corresponding extension data for each piece of human-computer interaction data for training, wherein the human-computer interaction data for training comprises control sentences for training and/or entries for training;
the topological structure determining module is used for determining the topological structure of the entry expansion model;
and the model training module is used for training to obtain the entry extension model by utilizing the training human-computer interaction data and the topological structure, so that the extension data output by the entry extension model at least comprises labeled extension data.
Optionally, the newly added entry determining module is configured to calculate a first similarity between the extended entry and the corresponding original entry, and a second similarity between the extended entry and another original entry; judging whether the first similarity is greater than the second similarity; if the first similarity is greater than the second similarity, and the difference value between the first similarity and the second similarity is not less than a preset difference value, the expansion entry is reserved; and determining the original entry and the reserved expansion entry as new entries.
Optionally, the apparatus further comprises:
the vocabulary entry class creating module is used for creating a new vocabulary entry class in the preset sentence pattern library and writing the new vocabulary entry into the new vocabulary entry class when the similar vocabulary entry corresponding to the new vocabulary entry is not matched;
and the mapping relation establishing module is used for establishing the mapping relation from the newly added vocabulary entry class to the specified vocabulary entry class of the preset sentence pattern library and generating the control sentence pattern with the newly added vocabulary entry.
Optionally, the apparatus further comprises:
the second judgment module is used for judging whether the occurrence frequency of the newly added entry is smaller than a preset frequency value or not;
and the first judging module is used for judging whether the similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not when the occurrence frequency of the newly added entries is not less than a preset frequency value.
Optionally, the apparatus further comprises:
the clustering processing module is used for clustering the newly added entry to obtain at least one newly added entry class;
and the first judging module is used for judging whether the similar entry class corresponding to the newly added entry class can be matched in a preset sentence pattern library.
Referring to fig. 8, a schematic diagram of an application control device of the present disclosure is shown, which can use the control sentence pattern generated by the device shown in fig. 7 to perform application control. The apparatus may include:
a human-computer interaction data obtaining module 701, configured to obtain human-computer interaction data input by a user for an application to be controlled, where the human-computer interaction data is used to control the application to be controlled to execute a specified function;
a control sentence pattern matching module 702, configured to identify the human-computer interaction data to obtain text information, and match the text information with the control sentence pattern;
and the function control module 703 is configured to, when the control sentence pattern corresponding to the text information is matched, control the application to be controlled to execute the specified function through the corresponding control sentence pattern.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
Referring to fig. 9, a schematic diagram of a structure of an electronic device 800 is shown. Referring to fig. 9, an electronic device 800 includes a processing component 801 that further includes one or more processors, and storage device resources, represented by storage medium 802, for storing instructions, such as application programs, that are executable by the processing component 801. The application stored in storage medium 802 may include one or more modules that each correspond to a set of instructions. Further, the processing component 801 is configured to execute instructions to execute the above-described manipulation sentence generation method or application manipulation method.
The electronic device 800 may further include a power component 803 configured to perform power management of the electronic device 800; a wired or wireless network interface 806 configured to connect the electronic device 800 to a network; and an input/output (I/O) interface 805. The electronic device 800 may operate based on an operating system stored on the storage medium 802, such as Windows Server, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, or the like.
The preferred embodiments of the present disclosure are described in detail with reference to the accompanying drawings, however, the present disclosure is not limited to the specific details of the above embodiments, and various simple modifications may be made to the technical solution of the present disclosure within the technical idea of the present disclosure, and these simple modifications all belong to the protection scope of the present disclosure.
It should be noted that, in the foregoing embodiments, various features described in the above embodiments may be combined in any suitable manner, and in order to avoid unnecessary repetition, various combinations that are possible in the present disclosure are not described again.
In addition, any combination of various embodiments of the present disclosure may be made, and the same should be considered as the disclosure of the present disclosure, as long as it does not depart from the spirit of the present disclosure.
Claims (19)
1. A method for generating a manipulation sentence, the method comprising:
acquiring an original entry corresponding to a function supported by an application to be controlled, wherein the original entry comprises an original entity and/or an original intention;
determining the original entry and the expansion entry corresponding to the original entry as a new entry;
judging whether similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not;
if the similar entry corresponding to the newly added entry is matched, merging the newly added entry into an entry class to which the similar entry belongs; if the similar vocabulary entry corresponding to the newly added vocabulary entry is not matched, creating a newly added vocabulary entry class in the preset sentence pattern library, and writing the newly added vocabulary entry into the newly added vocabulary entry class;
generating a control sentence pattern with the added vocabulary entry according to the mapping relation between the vocabulary entry class to which the similar vocabulary entry belongs and other vocabulary entry classes in the preset sentence pattern library or establishing the mapping relation from the added vocabulary entry class to the appointed vocabulary entry class in the preset sentence pattern library; the generating of the control sentence pattern with the newly added vocabulary entry specifically comprises: and generating a control sentence pattern with the newly added vocabulary entry based on the mapping relation between the intention type vocabulary entry and the entity type vocabulary entry.
2. The method of claim 1, wherein the manner of obtaining the extended entry corresponding to the original entry is as follows:
and taking the original entry as the input of an entry expansion model, and outputting the expansion entry corresponding to the original entry after the entry expansion model processes the entry.
3. The method of claim 2, wherein the term expansion model is constructed by:
acquiring human-computer interaction data for training, and labeling corresponding extension data for each piece of human-computer interaction data for training, wherein the human-computer interaction data for training comprises control sentences for training and/or entries for training;
determining a topological structure of the entry expansion model;
and training to obtain the vocabulary entry extension model by utilizing the training human-computer interaction data and the topological structure, so that the extension data output by the vocabulary entry extension model at least comprise labeled extension data.
4. The method according to any one of claims 1 to 3, wherein the determining the original entry and the extension entry corresponding to the original entry as the new entry comprises:
calculating a first similarity between the expansion entry and the corresponding original entry and a second similarity between the expansion entry and other original entries;
judging whether the first similarity is greater than the second similarity;
if the first similarity is greater than the second similarity, and the difference value between the first similarity and the second similarity is not less than a preset difference value, the expansion entry is reserved;
and determining the original entry and the reserved expansion entry as new entries.
5. The method according to any one of claims 1 to 3, wherein before the determining whether the similar vocabulary entry corresponding to the newly added vocabulary entry can be matched in the preset sentence pattern library, the method further comprises:
judging whether the occurrence frequency of the newly added entry is smaller than a preset frequency value or not;
and if the occurrence frequency of the newly added entry is not less than the preset frequency value, executing the step of judging whether the similar entry corresponding to the newly added entry can be matched in a preset sentence pattern library or not.
6. The method according to any one of claims 1 to 3, wherein after determining the original entry and the extension entry corresponding to the original entry as new entries, the method further comprises:
clustering the newly added entries to obtain at least one newly added entry class;
then, the judging whether the similar vocabulary entry corresponding to the newly added vocabulary entry can be matched in a preset sentence pattern library includes: and judging whether the similar entry class corresponding to the newly added entry class can be matched in a preset sentence pattern library.
7. An application manipulation method for performing application manipulation using a manipulation sentence pattern generated by the method of any one of claims 1 to 6, the method comprising:
acquiring human-computer interaction data input by a user aiming at an application to be controlled, wherein the human-computer interaction data is used for controlling the application to be controlled to execute a specified function;
identifying the human-computer interaction data to obtain text information, and matching the text information with the control sentence pattern;
and if the control sentence pattern corresponding to the text information is matched, controlling the application to be controlled to execute the specified function through the corresponding control sentence pattern.
8. A manipulation sentence generation apparatus, comprising:
the system comprises an original entry acquisition module, a control module and a display module, wherein the original entry acquisition module is used for acquiring an original entry corresponding to a function supported by an application to be controlled, and the original entry comprises an original entity and/or an original intention;
a newly added entry determining module, configured to determine the original entry and an extended entry corresponding to the original entry as a newly added entry;
the first judging module is used for judging whether the similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not;
the entry merging module is used for merging the newly added entry into an entry class to which the similar entry belongs when the similar entry corresponding to the newly added entry is matched;
the vocabulary entry class creating module is used for creating a new vocabulary entry class in the preset sentence pattern library and writing the new vocabulary entry into the new vocabulary entry class when the similar vocabulary entry corresponding to the new vocabulary entry is not matched;
a control sentence pattern generation module, configured to generate a control sentence pattern with the new vocabulary entry according to a mapping relationship between the vocabulary entry class to which the similar vocabulary entry belongs and other vocabulary entry classes in the preset sentence pattern library, or according to an established mapping relationship from the new vocabulary entry class to an assigned vocabulary entry class in the preset sentence pattern library; the generating of the control sentence pattern with the newly added vocabulary entry specifically comprises: and generating a control sentence pattern with the newly added vocabulary entry based on the mapping relation between the intention type vocabulary entry and the entity type vocabulary entry.
9. The apparatus of claim 8, further comprising:
and the extended entry output module is used for taking the original entry as the input of an entry extended model, and outputting the extended entry corresponding to the original entry after the entry extended model processes the original entry.
10. The apparatus of claim 9, further comprising:
the human-computer interaction data labeling module is used for acquiring human-computer interaction data for training and labeling corresponding extension data for each piece of human-computer interaction data for training, wherein the human-computer interaction data for training comprises control sentences for training and/or entries for training;
the topological structure determining module is used for determining the topological structure of the entry expansion model;
and the model training module is used for training to obtain the entry extension model by utilizing the training human-computer interaction data and the topological structure, so that the extension data output by the entry extension model at least comprises labeled extension data.
11. The apparatus according to any one of claims 8 to 10,
the newly added entry determining module is used for calculating a first similarity between the extension entry and the corresponding original entry and a second similarity between the extension entry and other original entries; judging whether the first similarity is greater than the second similarity; if the first similarity is greater than the second similarity, and the difference value between the first similarity and the second similarity is not less than a preset difference value, the expansion entry is reserved; and determining the original entry and the reserved expansion entry as new entries.
12. The apparatus of any one of claims 8 to 10, further comprising:
and the mapping relation establishing module is used for establishing the mapping relation from the newly added vocabulary entry class to the specified vocabulary entry class of the preset sentence pattern library and generating the control sentence pattern with the newly added vocabulary entry.
13. The apparatus of any one of claims 8 to 10, further comprising:
the second judgment module is used for judging whether the occurrence frequency of the newly added entry is smaller than a preset frequency value or not;
and the first judging module is used for judging whether the similar entries corresponding to the newly added entries can be matched in a preset sentence pattern library or not when the occurrence frequency of the newly added entries is not less than a preset frequency value.
14. The apparatus of any one of claims 8 to 10, further comprising:
the clustering processing module is used for clustering the newly added entry to obtain at least one newly added entry class;
and the first judging module is used for judging whether the similar entry class corresponding to the newly added entry class can be matched in a preset sentence pattern library.
15. An application manipulation device for manipulating an application using a manipulation sentence generated by the device of any one of claims 8 to 14, the device comprising:
the system comprises a human-computer interaction data acquisition module, a control module and a display module, wherein the human-computer interaction data acquisition module is used for acquiring human-computer interaction data input by a user aiming at an application to be controlled, and the human-computer interaction data is used for controlling the application to be controlled to execute a specified function;
the control sentence pattern matching module is used for identifying the human-computer interaction data to obtain text information and matching the text information with the control sentence pattern;
and the function control module is used for controlling the application to be controlled to execute the specified function through the corresponding control sentence pattern when the control sentence pattern corresponding to the text information is matched.
16. A storage medium having stored thereon a plurality of instructions, wherein the instructions are loadable by a processor and adapted to cause execution of the steps of the method according to any of claims 1 to 6.
17. A storage medium having stored thereon a plurality of instructions, wherein the instructions are loaded by a processor to perform the steps of the method of claim 7.
18. An electronic device, characterized in that the electronic device comprises;
the storage medium of claim 16; and
a processor to execute the instructions in the storage medium.
19. An electronic device, characterized in that the electronic device comprises;
the storage medium of claim 17; and
a processor to execute the instructions 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 CN107678309A (en) | 2018-02-09 |
| CN107678309B true 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) |
Families Citing this family (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 |
| CN110516175B (en) * | 2019-08-29 | 2022-05-17 | 秒针信息技术有限公司 | Method, device, equipment and medium for determining user label |
| 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 |
Citations (7)
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| US7103533B2 (en) * | 2001-02-21 | 2006-09-05 | International Business Machines Corporation | Method for preserving contextual accuracy in an extendible speech recognition language model |
| US20030191629A1 (en) * | 2002-02-04 | 2003-10-09 | Shinichi Yoshizawa | Interface apparatus and task control method for assisting in the operation of a device using recognition technology |
| 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 |
| CN103092928B (en) * | 2012-12-31 | 2015-12-23 | 安徽科大讯飞信息科技股份有限公司 | Voice inquiry method and system |
| CN104281609B (en) * | 2013-07-08 | 2020-03-17 | 腾讯科技（深圳）有限公司 | Configuration method and device for voice input instruction matching rule |
| 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 |
| US10770060B2 (en) * | 2013-12-05 | 2020-09-08 | Lenovo (Singapore) Pte. Ltd. | Adaptively learning vocabulary for completing speech recognition commands |
| CN105869640B (en) * | 2015-01-21 | 2019-12-31 | 上海墨百意信息科技有限公司 | Method and device for recognizing voice control instruction aiming at 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 |
| CN106571140B (en) * | 2016-11-14 | 2021-06-22 | Tcl科技集团股份有限公司 | Intelligent electric appliance control method and system based on voice semantics |
-
2017
- 2017-09-01 CN CN201710781250.7A patent/CN107678309B/en active Active
Patent Citations (7)
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107678309A (en) | 2018-02-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109145123B (en) | Knowledge graph model construction method, intelligent interaction method and system and electronic equipment | |
| US9865264B2 (en) | Selective speech recognition for chat and digital personal assistant systems | |
| CN108694940B (en) | Voice recognition method and device and electronic equipment | |
| CN106205615B (en) | Control method and system based on voice interaction | |
| CN107678309B (en) | Control sentence pattern generation and application control method and device and storage medium | |
| CN103903611B (en) | A kind of recognition methods of voice messaging and equipment | |
| CN109243468B (en) | Voice recognition method and device, electronic equipment and storage medium | |
| EP3977445B1 (en) | System and method for registering device for voice assistant service | |
| CN107515857B (en) | Semantic understanding method and system based on customization technology | |
| US11830482B2 (en) | Method and apparatus for speech interaction, and computer storage medium | |
| JP2015511746A (en) | Interactive information retrieval system and method | |
| CN110428819B (en) | Decoding network generation method, voice recognition method, device, equipment and medium | |
| CN108304424B (en) | Text keyword extraction method and text keyword extraction device | |
| CN105973257A (en) | Navigation method and device | |
| CN111949240A (en) | Interaction method, storage medium, service program, and device | |
| CN104714954A (en) | Information searching method and system based on context understanding | |
| WO2021135561A1 (en) | Skill voice wake-up method and apparatus | |
| CN115132209B (en) | Speech recognition method, apparatus, device and medium | |
| CN110008326A (en) | Knowledge abstraction generating method and system in conversational system | |
| CN111414512A (en) | Resource recommendation method and device based on voice search and electronic equipment | |
| CN110795547B (en) | Text recognition method and related product | |
| CN111427444B (en) | Control method and device of intelligent device | |
| CN109725798B (en) | Intelligent role switching method and related device | |
| CN111312230A (en) | Voice interaction monitoring method and device for voice conversation platform | |
| CN113470649A (en) | Voice interaction method and device |
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 |