CN110413758B - Session framework construction method and device based on machine learning - Google Patents
Session framework construction method and device based on machine learning Download PDFInfo
- Publication number
- CN110413758B CN110413758B CN201910694635.9A CN201910694635A CN110413758B CN 110413758 B CN110413758 B CN 110413758B CN 201910694635 A CN201910694635 A CN 201910694635A CN 110413758 B CN110413758 B CN 110413758B
- Authority
- CN
- China
- Prior art keywords
- model
- node
- state
- session
- flow
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Abstract
The invention provides a session framework construction method and device based on machine learning, wherein the method comprises the following steps: sending the identifications of the plurality of preconfigured process nodes to a client for user configuration selection; acquiring a conversation process fed back by a client; acquiring corresponding process node information according to a plurality of process node identifiers in the session process; and obtaining a finite state machine model table of the session flow according to the flow node information acquired by the node flow organization in the session flow, wherein the finite state machine model table is used as a session frame, the session frame can be quickly generated by pushing a plurality of preconfigured alternative nodes to a user, and in addition, the preconfigured alternative nodes can be multiplexed aiming at different sessions, so that the development time and development resources are greatly saved, and the development speed is improved.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a session framework construction method and device based on machine learning.
Background
As technology develops, more and more devices will have networking capabilities, and how these devices interact with people will become a challenge.
Natural language becomes a novel interaction mode adapting to the trend, and the conversation robot is expected to replace the past websites and the current APP and occupies a new generation of human-computer interaction air ports. A conversation bot is essentially a computer program that simulates a human conversation or chat. On a technical level, the conversational robots may be divided into three categories: a chatting robot, a question and answer robot and a task robot.
With the development of artificial intelligence and natural language processing technology, the conversation robot is gradually applied to financial services, home life and personal assistants.
At present, the market demands for robots are diversified, and corresponding dialogue robots are required to be created according to different application demands so as to realize personalized functions, but the construction of the dialogue robots with different functions from planning and implementation to online is complex in involved links, long in time consumption of iterative optimization, high in creation threshold, low in engineering scheduling efficiency, and limited in wide application of the dialogue robots.
Disclosure of Invention
In view of the problems in the prior art, the present invention provides a session framework construction method and apparatus based on machine learning, an electronic device, and a computer-readable storage medium, which can at least partially solve the problems in the prior art.
In order to achieve the purpose, the invention adopts the following technical scheme:
in a first aspect, a session framework construction method based on machine learning is provided, which includes:
sending an identification of a plurality of preconfigured process nodes to a client for user configuration selection, the plurality of preconfigured process nodes respectively executing a plurality of different jobs, each preconfigured process node information comprising a plurality of states, each state comprising: present state, condition, action, and substate;
obtaining a conversation process fed back by a client, wherein the conversation process comprises a plurality of process node identifications and node flow directions;
acquiring corresponding process node information according to a plurality of process node identifiers in the session process;
and obtaining a finite state machine model table of the conversation process according to the process node information acquired by the node flow direction organization in the conversation process, and using the finite state machine model table as a conversation frame.
Further, the current state is the current state of the dialog; triggering an action or executing a state transition when the condition is met, wherein the state transition is a new state to be transited to after the condition is met;
the finite state machine model table of the session flow is obtained according to the flow node information acquired by the node flow organization in the session flow, and the finite state machine model table is used as a session frame and comprises the following steps:
sequencing the acquired process node information according to the node flow direction in the session process;
and configuring the secondary state of each state in the sequenced flow node information so as to carry out node jump according to the flow direction of the nodes in the conversation flow.
Further, the secondary state includes: the address bar is used for indicating a new state to be migrated after the condition is met;
the configuration of the secondary state of each state in the sorted flow node information includes:
and inserting the jump address of the next node into an address column in the secondary state of the last state of the current node.
Further, the action includes: a rule template identification and/or a process model identification,
the session framework construction method further comprises the following steps:
calling a corresponding rule template from a database according to the rule template identification included in each action in the finite state machine model table;
and calling the corresponding processing model from the database according to the processing model identification included in each action in the finite state machine model table.
Further, the process model includes: a word segmentation model, a text classification model or an element extraction model.
Further, still include:
calling a model training process of the model from a model training engine according to the processing model identification;
obtaining labeled model training data;
the processing model is trained according to the labeled model training data and the model training flow.
Further, still include:
obtaining annotated model test data;
the process model is tested using the labeled model test data.
In a second aspect, a session framework building apparatus based on machine learning is provided, including:
the standby node display module sends the identifications of a plurality of preconfigured process nodes to the client for configuration and selection of a user, the plurality of preconfigured process nodes respectively execute a plurality of different jobs, each preconfigured process node information comprises a plurality of states, and each state comprises: present state, condition, action, and substate;
the session flow receiving module is used for acquiring a session flow fed back by the client, and the session flow comprises a plurality of flow node identifications and node flow directions;
the flow node information acquisition module acquires corresponding flow node information according to a plurality of flow node identifiers in the session flow;
and the session frame construction module is used for obtaining a finite state machine model table of the session flow as a session frame according to the flow node information acquired by the node flow organization in the session flow.
Further, the current state is the current state of the dialog; triggering an action or executing a state transition when the condition is met, wherein the state transition is a new state to be transited to after the condition is met;
the session framework building module comprises:
the architecture organization unit sequences the acquired process node information according to the node flow direction in the session process;
and the configuration unit is used for configuring the secondary state of each state in the sequenced process node information so as to carry out node jump according to the flow direction of the nodes in the conversation process.
Further, the secondary state includes: the address bar is used for indicating a new state to be migrated after the condition is met;
the configuration unit includes:
and the jump configuration subunit inserts the jump address of the next node into an address column in the secondary state of the last state of the current node.
Further, the action includes: a rule template identification and/or a process model identification,
the session framework construction device further comprises:
the rule template calling module is used for calling the corresponding rule template from the database according to the rule template identification included in each action in the finite state machine model table;
and the processing model calling module calls the corresponding processing model from the database according to the processing model identification included in each action in the finite state machine model table.
Further, the process model includes: a word segmentation model, a text classification model or an element extraction model.
Further, still include:
the model training flow acquisition module calls a model training flow of the model from a model training engine according to the processing model identification;
a training sample acquisition module for acquiring labeled model training data;
and the model training module is used for training the processing model according to the labeled model training data and the model training process.
Further, still include:
a test sample acquisition module for acquiring the labeled model test data;
and the model testing module is used for testing the processing model by using the marked model testing data.
In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the steps of the above-mentioned session framework building method based on machine learning when executing the program.
In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the above-mentioned machine learning based session framework construction method.
The invention provides a session framework construction method, a session framework construction device, electronic equipment and a computer-readable storage medium based on machine learning, wherein the method comprises the following steps: sending an identification of a plurality of preconfigured process nodes to a client for user configuration selection, the plurality of preconfigured process nodes respectively executing a plurality of different jobs, each preconfigured process node information comprising a plurality of states, each state comprising: present state, condition, action, and substate; obtaining a conversation process fed back by a client, wherein the conversation process comprises a plurality of process node identifications and node flow directions; acquiring corresponding process node information according to a plurality of process node identifiers in the session process; the method comprises the steps of obtaining a finite state machine model table of a session flow according to flow node information obtained by a node flow direction organization in the session flow, using the finite state machine model table as a session frame, wherein the session frame can be quickly generated by pushing a plurality of pre-configured alternative nodes to a user, in addition, the pre-configured alternative nodes can be reused aiming at different sessions, development time and development resources are greatly saved, development speed is improved, the user only needs to drag a point-line connection and inject training or test data into a server, a conversation robot meeting requirements can be created and trained, the conversation robot can be directly on-line in a project, the robot can become more clever and intelligent along with continuous learning of conversations with people, flexible customization and efficient scheduling are realized, and the conversation robot can be quickly and simply produced in batches.
In order to make the aforementioned and other objects, features and advantages of the invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts. In the drawings:
FIG. 1 is a schematic diagram of an architecture between a server S1 and a client device B1 according to an embodiment of the present invention;
FIG. 2 is a block diagram of the server S1, the client device B1 and the database server S2 according to an embodiment of the present invention;
FIG. 3 is a first flowchart illustrating a session framework building method based on machine learning according to an embodiment of the present invention;
FIG. 4 shows a session framework constructed by the session framework construction method based on machine learning provided by the embodiment of the invention;
fig. 5 shows the specific steps of step S400 in fig. 3;
FIG. 6 is a block diagram of a session framework building apparatus based on machine learning according to an embodiment of the present invention;
fig. 7 is a block diagram of an electronic device according to an embodiment of the invention.
Detailed Description
In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only partial embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
It should be noted that the terms "comprises" and "comprising," and any variations thereof, in the description and claims of this application and the above-described drawings, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
At present, the market demands for robots are diversified, and corresponding dialogue robots are required to be created according to different application demands so as to realize personalized functions, but the construction of the dialogue robots with different functions from planning and implementation to online is complex in involved links, long in time consumption of iterative optimization, high in creation threshold, low in engineering scheduling efficiency, and limited in wide application of the dialogue robots.
In order to at least partially solve the technical problems in the prior art, embodiments of the present invention provide a session framework construction method based on machine learning, which can quickly generate a session framework by pushing a plurality of preconfigured candidate nodes to a user, and in addition, can reuse the preconfigured candidate nodes for different sessions, thereby greatly saving development time and development resources, improving development speed, and allowing the user to create and train a session robot meeting requirements by dragging a dotted-line connection and injecting training or test data into a server, which can be directly brought online in a project, and allow the session robot to become smarter and smarter along with continuous learning of a session with people, thereby achieving flexible customization and efficient scheduling, and being capable of quickly and simply producing the session robot in batches.
In view of the above, the present application provides a session framework building apparatus based on machine learning, which may be a server S1, see fig. 1, where the server S1 may be communicatively connected to at least one client device B1, the server S1 sends to the client device B1 an identification of a plurality of preconfigured process nodes for user configuration selection, the plurality of preconfigured process nodes respectively execute a plurality of different jobs, each preconfigured process node information includes a plurality of states, and each state includes: present state, condition, action, and substate; the client device B1 may send a selected session flow to the server S1, where the session flow includes multiple flow node identifiers and node flow directions, and the server S1 may receive the session flow online and obtain corresponding flow node information according to the multiple flow node identifiers in the session flow; and obtaining a finite state machine model table of the conversation process according to the process node information acquired by the node flow direction organization in the conversation process, and using the finite state machine model table as a conversation frame.
In addition, referring to fig. 2, the server S1 may further be communicatively connected to at least one database server S2, the database server S2 is configured to store rule templates, processing models, and training procedures of the respective models, for the server S1 to call, and the server S1 may receive labeled model training data online, and then perform model training on a desired processing model according to the labeled model training data.
Based on the above, the server S1 may receive the labeled model training data online and then test the process model using the labeled model test data.
It is understood that the client device B1 may include a smart phone, a tablet electronic device, a portable computer, a desktop computer, a Personal Digital Assistant (PDA), or the like.
In practical applications, the part of the session framework construction can be performed on the side of the server S1 as described above, i.e. the architecture shown in fig. 1, all operations can be completed in the client device B1, and the client device B1 can be directly connected to the database server S2 in communication. Specifically, the selection may be performed according to the processing capability of the client device B1, the limitation of the user usage scenario, and the like. This is not a limitation of the present application. If all the operations are completed in the client device B1, the client device B1 may further include a processor for performing a specific process of session framework construction.
The server and the client device may communicate using any suitable network protocol, including network protocols not yet developed at the filing date of this application. The network protocol may include, for example, a TCP/IP protocol, a UDP/IP protocol, an HTTP protocol, an HTTPS protocol, or the like. Of course, the network Protocol may also include, for example, an RPC Protocol (Remote Procedure Call Protocol), a REST Protocol (Representational State Transfer Protocol), and the like used above the above Protocol.
Fig. 3 is a first flowchart of a session framework building method based on machine learning in an embodiment of the present invention. As shown in fig. 3, the session framework building method based on machine learning may include the following:
step S100: sending an identification of a plurality of preconfigured process nodes to a client for user configuration selection, the plurality of preconfigured process nodes respectively executing a plurality of different jobs, each preconfigured process node information comprising a plurality of states, each state comprising: present state, condition, action, and substate;
specifically, each process node is used for executing a job, the state of which is configured by the user in advance, and the model or algorithm to be called in the action is also configured by the user in advance.
In addition, the flow node not only comprises a node for executing logic processing, but also comprises a reply dialect node, wherein the specific dialect replied in the reply dialect node is configured by the user according to a specific task when the conversation flow is created, and the configured dialect is returned according to the intention of the user.
Step S200: and acquiring a conversation process fed back by the client, wherein the conversation process comprises a plurality of process node identifications and node flow directions.
Wherein the node flow direction indicates the direction of the flow.
Specifically, the user may implement the session flow configuration by dragging the nodes needed for the dotted line connection on the display of the client.
It is worth mentioning that the session flow can be understood as a flow chart for implementing a task, see fig. 4.
As shown in fig. 4, the session flow is a session framework configured for the user to perform the task of adjusting the credit line.
The session flow required by the session framework is as follows:
firstly, identity authentication is carried out; when the identity authentication fails, skipping to the voice navigation of the previous node; judging whether the identity authentication is successful or not; if yes, obtaining the credit card limit, if the condition is met, replying the dialogue "confirm the quota", if the condition is not met, jumping to a voice navigation node or transferring to manual work or ending the conversation after replying according to specific customer intention.
And dragging a dotted line on a display of the client to connect the required nodes according to the flow to realize conversation flow configuration, and configuring the conversations required by each answering conversation node.
Step S300: acquiring corresponding process node information according to a plurality of process node identifiers in the session process;
specifically, in the process of feeding back the process node to the client so that the client configures the process, only the process node identifier is transmitted for reducing data transmission, and the implementation program of the process node does not need to be transmitted to the client, so that the transmission speed can be improved on one hand, and on the other hand, the node program can be prevented from being stolen and copied on the other hand.
Based on the method, the session flow configured by the user comprises a plurality of flow node identifications and node flow directions. The process node information corresponding to the process node identifier needs to be called from the database according to the process node identifier.
The node attribute configured by the user may include: node name, maximum turn of conversation, whether to edit a conversation, match word slots, whether to break, etc.
Step S400: and obtaining a finite state machine model table of the conversation process according to the process node information acquired by the node flow direction organization in the conversation process, and using the finite state machine model table as a conversation frame.
It can be understood by those skilled in the art that after the information of each flow node is obtained, because the information of each flow node is independently prefabricated and thus has no relationship with each other, at this time, the flow node information of each flow node obtained according to the flow direction organization of the nodes is needed to obtain a session framework for implementing a specific task, and the session framework can be represented by a finite state machine model table (see table 1 and table 2), which state transition function is determined to be executed according to the current state of the system and an event that occurs, and a conditional branch technique is used to implement automatic change of the system state.
TABLE 1
TABLE 2
The finite-state machine model table comprises a present state, a condition, an action and a secondary state. The present state and condition are the cause, and the action and the next state are the effect. Wherein, in the session framework:
the current state is as follows: the current state of the dialog.
Conditions are as follows: when a condition is satisfied (an intent is recognized), an action is triggered or a state transition is performed.
The actions are as follows: and after the condition is met, the executed action can be transferred to a new state or still keep the original state. In a conversational robot, the robot performs a reply to different utterances (actions) according to different recognized intentions (conditions).
The next state: and migrating to a new state after the condition is met. The secondary state, relative to the present state, once activated, transitions to a new present state.
According to the technical scheme, the session framework can be quickly generated by pushing a plurality of preconfigured alternative nodes to a user, in addition, the preconfigured alternative nodes can be reused for different sessions, development time and development resources are greatly saved, development speed is improved, the user only needs to drag a dotted line for connection and inject training or test data into a server, a conversation robot meeting requirements can be created and trained, the conversation robot can be directly on-line in a project, the conversation robot becomes smarter and smarter along with continuous learning of conversations with people, flexible customization and efficient scheduling are realized, and the conversation robot can be quickly and simply produced in batches.
Fig. 5 shows the specific steps of step S400 in fig. 3; as shown in fig. 5, this step S400 may include the following:
step S410: sequencing the acquired process node information according to the node flow direction in the session process;
it is worth mentioning that the node flow direction may include a straight line flow direction and a multi-branch flow direction, and the straight line flow direction is, for example: node 1 → node 2 → node 3 → node 4, with no bifurcation at each node. Multi-pronged flow referring to fig. 4, the nodes may or may not be bifurcated.
Specifically, the process node information corresponding to each node is organized according to the node flow.
Step S420: and configuring the secondary state of each state in the sequenced flow node information so as to carry out node jump according to the flow direction of the nodes in the conversation flow.
Wherein the current state is the current state of the dialog; and triggering an action or executing one state migration when the condition is met, wherein the secondary state is a new state to be migrated after the condition is met.
It should be noted that the configuration process mainly organizes the jump relationship of each flow node according to the node flow direction, and thus mainly configures the sub-state of each state in the flow node information corresponding to each node.
Specifically, the secondary state may include: the address bar is used for indicating a new state to be migrated after the condition is met;
configuring the secondary state of each state in the sequenced process node information as follows: and inserting the jump address of the next node into an address column in the secondary state of the last state of the current node.
By adopting the technical scheme, after the operation of the node is executed, each node can jump to the next node according to the flow configured by the user.
In an optional embodiment, the actions corresponding to each state include: a rule template identification and/or a process model identification for performing certain actions.
The session framework construction method further comprises the following steps:
calling a corresponding rule template from a database according to the rule template identification included in each action in the finite state machine model table;
and calling the corresponding processing model from the database according to the processing model identification included in each action in the finite state machine model table.
Wherein the process model comprises: a computer model commonly used in a conversation, such as a word segmentation model, a text classification model or an element extraction model.
As will be understood by those skilled in the art, rule templates and processing models are generally complex and occupy a large space, and therefore, corresponding rule templates or processing models are not stored in each node, and only rule template identifications and/or processing model identifications are configured by a user, so that the rule templates and the processing models used for pairing need to be called in a database according to the identifications.
It is to be noted that the identification mentioned in the present application may be a name or a number, etc.
In an optional embodiment, the session framework building method based on machine learning may further include:
calling a model training process of the model from a model training engine according to the processing model identification;
obtaining labeled model training data;
and training the processing model according to the labeled model training data and the model training process.
The model training process is pre-stored in a model training engine, and in practical application, the model training process of the model is called from the model training engine according to the processing model identification and is used for training the model.
By adopting the technical scheme, the modeling training can be quickly realized by using a machine learning method, the dialogue robot can be quickly constructed, the intelligent level of the robot can be improved, the online cost of the project can be saved, the repeated development can be prevented, and the development manpower and resources can be saved.
In an optional embodiment, the session framework building method based on machine learning may further include:
obtaining annotated model test data;
testing the process model using the labeled model test data.
After model training is finished, the labeled model test data is input into the trained model, the output of the model is used as a test result, the test result is compared with a label pre-labeled by the model test data, whether the model precision meets the requirement or not is judged, if yes, the trained model is issued to a node, and if not, the training data is replaced to retrain the model.
In an optional embodiment, after training and testing data input by a user is acquired, naming the data and setting browsing permission by the session framework construction method based on machine learning.
The data naming is to select a corresponding sample data set in the model training process.
The data authority setting is to ensure that only persons related to a certain scene can access or modify data when a plurality of machine scenes are jointly developed.
In an optional embodiment, the session framework building method based on machine learning may further include: and setting data display parameters.
Specifically, the display parameters include: the line number of data display is previewed, and the data display viewing mode, such as data source, data line number, uploading date and the like, is viewed by a user or a developer after data import, so that the data import is ensured to be correct.
In an optional embodiment, the session framework building method based on machine learning may further include:
and numbering the acquired session frames.
In particular, for the case where multiple robots or one robot can implement multiple sessions, the session frames generated are labeled to distinguish between different session frames.
In an optional embodiment, the session framework building method based on machine learning may further include: and (5) incremental data annotation.
Specifically, new added session data are clustered by a clustering algorithm every day, the user intention is preliminarily recognized, and the user intention is corrected manually, so that the corrected data are used for performing supplementary training on the processing model.
In an alternative embodiment, the machine learning model is release controlled through a release control and release agent. And uniformly issuing a request for controlling the Publishmaster to interface the workstations. The heterogeneous publishing Agent is a publishing logic control node, receives the information of the publishMaster, and realizes publishing or upgrading of the heterogeneous model based on k8s Gateway.
The machine learning model service management manages and monitors the online pre-estimated service, the service management realizes the online and offline management of the service by calling a K8S interface, and the model version subscription and version state monitoring in the service are realized by registering zookeeper. The unified estimation service is realized based on PAAS cloud.
It should be noted that a variety of models such as a word segmentation model (jieba, etc.), a text classification model (Fasttext, RNN, BERT, etc.), and an element extraction model (CRF, LSTM, etc.) are prestored in the database. By calling the model, the robot is led to recognize the user intention and extract key elements. For example, the user says: i would transfer 100 money to xiaoming. The intent is identified by the model as: and (7) transferring the funds. Identifying key elements: xiaoming, 100 blocks.
The model training engine schedules YARN using big data resources, packages model training engines such as open source H2O training engine, Python model training engine, etc.
Based on the same inventive concept, the embodiment of the present application further provides a session framework building apparatus based on machine learning, which can be used to implement the methods described in the above embodiments, as described in the following embodiments. Because the principle of solving the problems of the session framework building device based on machine learning is similar to that of the method, the implementation of the session framework building device based on machine learning can refer to the implementation of the method, and repeated details are not repeated. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
Fig. 6 is a block diagram i of the structure of the session framework building device based on machine learning in the embodiment of the present invention. As shown in fig. 6, the session framework building apparatus based on machine learning specifically includes: the system comprises an alternative node display module 10, a session flow receiving module 20, a flow node information acquisition module 30 and a session framework construction module 40.
The candidate node presentation module 10 sends, to the client, identifiers of a plurality of preconfigured process nodes for user configuration selection, where the plurality of preconfigured process nodes respectively execute a plurality of different jobs, and each piece of preconfigured process node information includes a plurality of states, and each state includes: present state, condition, action, and substate;
specifically, each process node is used for executing a job, the state of which is configured by the user in advance, and the model or algorithm to be called in the action is also configured by the user in advance.
In addition, the flow node not only comprises a node for executing logic processing, but also comprises a reply dialect node, wherein the specific dialect replied in the reply dialect node is configured by the user according to a specific task when the conversation flow is created, and the configured dialect is returned according to the intention of the user.
The session flow receiving module 20 obtains a session flow fed back by the client, where the session flow includes multiple flow node identifiers and node flow directions;
wherein the node flow direction indicates the direction of the flow.
Specifically, the user can implement the configuration of the conversation process by dragging the point-line connection required node on the display of the client, and configure the conversation required to be answered by each answering conversation node.
It is worth mentioning that the session flow can be understood as a flow chart for implementing a task
The process node information obtaining module 30 obtains corresponding process node information according to the multiple process node identifiers in the session process;
specifically, in the process of feeding back the process node to the client so that the client configures the process, only the process node identifier is transmitted for reducing data transmission, and the implementation program of the process node does not need to be transmitted to the client, so that the transmission speed can be improved on one hand, and on the other hand, the node program can be prevented from being stolen and copied on the other hand.
Based on the method, the session flow configured by the user comprises a plurality of flow node identifications and node flow directions. The process node information corresponding to the process node identifier needs to be called from the database according to the process node identifier.
The node attribute configured by the user may include: node name, maximum turn of conversation, whether to edit a conversation, match word slots, whether to break, etc.
The session framework building module 40 obtains a finite state machine model table of the session flow as a session framework according to the flow node information acquired by the node flow organization in the session flow.
It can be understood by those skilled in the art that after the information of each flow node is obtained, because the information of each flow node is independently prefabricated and thus has no relationship with each other, at this time, the flow node information of each flow node obtained according to the flow direction organization of the nodes is needed to obtain a session frame for implementing a specific task, and the session frame can be represented by a finite state machine model table, which state transition function is determined to be executed according to the current state of the system and an event that occurs, and a conditional branching technique is used to implement automatic change of the system state.
The finite-state machine model table comprises a present state, a condition, an action and a secondary state. The present state and condition are the cause, and the action and the next state are the effect. Wherein, in the session framework:
the current state is as follows: the current state of the dialog.
Conditions are as follows: when a condition is satisfied (an intent is recognized), an action is triggered or a state transition is performed.
The actions are as follows: and after the condition is met, the executed action can be transferred to a new state or still keep the original state. In a conversational robot, the robot performs a reply to different utterances (actions) according to different recognized intentions (conditions).
The next state: and migrating to a new state after the condition is met. The secondary state, relative to the present state, once activated, transitions to a new present state.
According to the technical scheme, the session framework construction device based on machine learning provided by the embodiment of the invention can quickly generate the session framework by pushing a plurality of preconfigured alternative nodes to a user, in addition, the preconfigured alternative nodes can be reused aiming at different sessions, the development time and development resources are greatly saved, the development speed is improved, the user only needs to drag a point-line connection and inject training or test data into a server, a session robot meeting requirements can be created and trained, the session robot can be directly on-line in a project, the session robot becomes smarter and smarter along with continuous learning of the session with people, flexible customization and efficient scheduling are realized, and the session robot can be quickly and simply produced in batches.
In an optional embodiment, the current state is a current state of the dialog; triggering an action or executing one state migration when the condition is met, wherein the secondary state is a new state to be migrated after the condition is met;
the session framework building module 40 includes: an architecture organization unit and a configuration unit.
The architecture organization unit sequences the acquired process node information according to the node flow direction in the session process;
it is worth mentioning that the node flow direction may include a straight line flow direction and a multi-branch flow direction, and the straight line flow direction is, for example: node 1 → node 2 → node 3 → node 4, with no bifurcation at each node. Multi-pronged flow referring to fig. 4, the nodes may or may not be bifurcated.
Specifically, the process node information corresponding to each node is organized according to the node flow.
The configuration unit configures the secondary state of each state in the sequenced flow node information so as to carry out node jump according to the flow direction of the nodes in the conversation flow.
Wherein the current state is the current state of the dialog; and triggering an action or executing one state migration when the condition is met, wherein the secondary state is a new state to be migrated after the condition is met.
It should be noted that the configuration process mainly organizes the jump relationship of each flow node according to the node flow direction, and thus mainly configures the sub-state of each state in the flow node information corresponding to each node.
In an alternative embodiment, the secondary state comprises: the address bar is used for indicating a new state to be migrated after the condition is met;
the configuration unit includes: and the jump configuration subunit inserts the jump address of the next node into an address column in the secondary state of the last state of the current node.
By adopting the technical scheme, after the operation of the node is executed, each node can jump to the next node according to the flow configured by the user.
In an alternative embodiment, the actions include: a rule template identification and/or a process model identification,
the session framework construction device further comprises: the rule template calling module and the processing model calling module.
The rule template calling module calls a corresponding rule template from a database according to a rule template identifier included in each action in the finite state machine model table;
and the processing model calling module calls the corresponding processing model from the database according to the processing model identification included in each action in the finite state machine model table.
Wherein the process model comprises: a computer model commonly used in a conversation, such as a word segmentation model, a text classification model or an element extraction model.
As will be understood by those skilled in the art, rule templates and processing models are generally complex and occupy a large space, and therefore, corresponding rule templates or processing models are not stored in each node, and only rule template identifications and/or processing model identifications are configured by a user, so that the rule templates and the processing models used for pairing need to be called in a database according to the identifications.
It is to be noted that the identification mentioned in the present application may be a name or a number, etc.
In an optional embodiment, the session framework building apparatus based on machine learning may further include: the device comprises a model training flow acquisition module, a training sample acquisition module and a model training module.
The model training flow obtaining module calls a model training flow of the model from a model training engine according to the processing model identification;
a training sample acquisition module acquires labeled model training data;
and the model training module trains the processing model according to the labeled model training data and the model training process.
The model training process is pre-stored in a model training engine, and in practical application, the model training process of the model is called from the model training engine according to the processing model identification and is used for training the model.
By adopting the technical scheme, the modeling training can be quickly realized by using a machine learning method, the dialogue robot can be quickly constructed, the intelligent level of the robot can be improved, the online cost of the project can be saved, the repeated development can be prevented, and the development manpower and resources can be saved.
In an optional embodiment, the session framework building apparatus based on machine learning may further include: the device comprises a test sample acquisition module and a model test module.
A test sample acquisition module acquires the labeled model test data;
a model test module tests the process model using the labeled model test data.
After model training is finished, the labeled model test data is input into the trained model, the output of the model is used as a test result, the test result is compared with a label pre-labeled by the model test data, whether the model precision meets the requirement or not is judged, if yes, the trained model is issued to a node, and if not, the training data is replaced to retrain the model.
In an optional embodiment, the session framework building device based on machine learning may further include a naming authority setting module, configured to name and set browsing authority for training and testing data input by a user after the data is obtained.
The data naming is to select a corresponding sample data set in the model training process.
The data authority setting is to ensure that only persons related to a certain scene can access or modify data when a plurality of machine scenes are jointly developed.
In an optional embodiment, the session framework building apparatus based on machine learning may further include: and the display parameter setting module is used for setting data display parameters.
Specifically, the display parameters include: the line number of data display is previewed, and the data display viewing mode, such as data source, data line number, uploading date and the like, is viewed by a user or a developer after data import, so that the data import is ensured to be correct.
In an optional embodiment, the session framework building apparatus based on machine learning may further include:
and the session frame numbering module is used for numbering the acquired session frames.
In particular, for the case where multiple robots or one robot can implement multiple sessions, the session frames generated are labeled to distinguish between different session frames.
In an optional embodiment, the session framework building apparatus based on machine learning may further include: and the incremental data labeling module is used for clustering the newly added session data every day through a clustering algorithm to preliminarily identify the user intention.
After the user intention is recognized, the correction is carried out manually, so that the processing model is subjected to supplementary training by using the corrected data.
In an alternative embodiment, the machine learning model is release controlled through a release control and release agent. And uniformly issuing a request for controlling the Publishmaster to interface the workstations. The heterogeneous publishing Agent is a publishing logic control node, receives the information of the publishMaster, and realizes publishing or upgrading of the heterogeneous model based on k8s Gateway.
The machine learning model service management manages and monitors the online pre-estimated service, the service management realizes the online and offline management of the service by calling a K8S interface, and the model version subscription and version state monitoring in the service are realized by registering zookeeper. The unified estimation service is realized based on PAAS cloud.
It should be noted that a variety of models such as a word segmentation model (jieba, etc.), a text classification model (Fasttext, RNN, BERT, etc.), and an element extraction model (CRF, LSTM, etc.) are prestored in the database. By calling the model, the robot is led to recognize the user intention and extract key elements. For example, the user says: i would transfer 100 money to xiaoming. The intent is identified by the model as: and (7) transferring the funds. Identifying key elements: xiaoming, 100 blocks.
The model training engine schedules YARN using big data resources, packages model training engines such as open source H2O training engine, Python model training engine, etc.
The apparatuses, modules or units illustrated in the above embodiments may be implemented by a computer chip or an entity, or implemented by a product with certain functions. A typical implementation device is an electronic device, which may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
In a typical example, the electronic device specifically includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor implements the following steps when executing the program:
sending an identification of a plurality of preconfigured process nodes to a client for user configuration selection, the plurality of preconfigured process nodes respectively executing a plurality of different jobs, each preconfigured process node information comprising a plurality of states, each state comprising: present state, condition, action, and substate;
obtaining a conversation process fed back by a client, wherein the conversation process comprises a plurality of process node identifications and node flow directions;
acquiring corresponding process node information according to a plurality of process node identifiers in the session process;
and obtaining a finite state machine model table of the conversation process according to the process node information acquired by the node flow direction organization in the conversation process, and using the finite state machine model table as a conversation frame.
As can be seen from the above description, the electronic device provided in the embodiment of the present invention may be used to construct a session framework, and a plurality of preconfigured candidate nodes are pushed to a user, so that the session framework can be quickly generated, and in addition, the preconfigured candidate nodes can be reused for different sessions, so that development time and development resources are greatly saved, and development speed is improved.
Referring now to FIG. 7, shown is a schematic diagram of an electronic device 600 suitable for use in implementing embodiments of the present application.
As shown in fig. 7, the electronic apparatus 600 includes a Central Processing Unit (CPU)601 that can perform various appropriate works and processes according to a program stored in a Read Only Memory (ROM)602 or a program loaded from a storage section 608 into a Random Access Memory (RAM)) 603. In the RAM603, various programs and data necessary for the operation of the system 600 are also stored. The CPU601, ROM602, and RAM603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
The following components are connected to the I/O interface 605: an input portion 606 including a keyboard, a mouse, and the like; an output portion 607 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section 608 including a hard disk and the like; and a communication section 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the internet. The driver 610 is also connected to the I/O interface 605 as needed. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 610 as necessary, so that a computer program read out therefrom is mounted as necessary on the storage section 608.
In particular, according to an embodiment of the present invention, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, an embodiment of the invention includes a computer-readable storage medium having a computer program stored thereon, which when executed by a processor, performs the steps of:
sending an identification of a plurality of preconfigured process nodes to a client for user configuration selection, the plurality of preconfigured process nodes respectively executing a plurality of different jobs, each preconfigured process node information comprising a plurality of states, each state comprising: present state, condition, action, and substate;
obtaining a conversation process fed back by a client, wherein the conversation process comprises a plurality of process node identifications and node flow directions;
acquiring corresponding process node information according to a plurality of process node identifiers in the session process;
and obtaining a finite state machine model table of the conversation process according to the process node information acquired by the node flow direction organization in the conversation process, and using the finite state machine model table as a conversation frame.
As can be seen from the above description, the computer-readable storage medium provided in the embodiments of the present invention may be used to construct a session framework, and push a plurality of preconfigured candidate nodes to a user, so as to quickly generate the session framework, and in addition, the preconfigured candidate nodes can be reused for different sessions, so as to greatly save development time and development resources, and improve development speed.
In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 609, and/or installed from the removable medium 611.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (14)
1. A session framework construction method based on machine learning is characterized by comprising the following steps:
sending an identification of a plurality of preconfigured process nodes to a client for user configuration selection, the plurality of preconfigured process nodes respectively executing a plurality of different jobs, each preconfigured process node information comprising a plurality of states, each state comprising: a present state, a condition, an action and a next state, wherein the action comprises: processing the model identification;
obtaining a conversation process fed back by a client, wherein the conversation process comprises a plurality of process node identifications and node flow directions;
acquiring corresponding process node information according to a plurality of process node identifiers in the session process;
obtaining a finite state machine model table of the conversation process according to the process node information acquired by the node flow direction organization in the conversation process, and using the finite state machine model table as a conversation frame;
calling a corresponding processing model from a database according to a processing model identifier included in each action in the state machine model table;
calling a model training process of the model from a model training engine according to the processing model identification;
obtaining labeled model training data;
and training the processing model according to the labeled model training data and the model training process.
2. The machine learning-based session framework construction method according to claim 1, wherein the current state is a current state of a dialog; triggering an action or executing one state migration when the condition is met, wherein the secondary state is a new state to be migrated after the condition is met;
the obtaining of the finite state machine model table of the session flow according to the flow node information obtained by the node flow direction organization in the session flow, as a session frame, includes:
sequencing the acquired process node information according to the node flow direction in the session process;
and configuring the secondary state of each state in the sequenced flow node information so as to carry out node jump according to the flow direction of the nodes in the conversation flow.
3. The machine-learning based session framework building method of claim 2, wherein the secondary state comprises: the address bar is used for indicating a new state to be migrated after the condition is met;
the configuring the secondary state of each state in the sorted flow node information includes:
and inserting the jump address of the next node into an address column in the secondary state of the last state of the current node.
4. The machine learning-based session framework building method according to claim 1, wherein the actions include: the identity of the rule template is identified,
the session framework construction method further comprises the following steps:
and calling the corresponding rule template from the database according to the rule template identification included in each action in the finite state machine model table.
5. The machine-learning based session framework building method of claim 4, wherein the processing model comprises: a word segmentation model, a text classification model or an element extraction model.
6. The machine learning-based session framework building method according to claim 1, further comprising:
obtaining annotated model test data;
testing the process model using the labeled model test data.
7. A session framework building apparatus based on machine learning, comprising:
the standby node display module sends the identifications of a plurality of preconfigured process nodes to the client for configuration and selection of a user, the plurality of preconfigured process nodes respectively execute a plurality of different jobs, each preconfigured process node information comprises a plurality of states, and each state comprises: a present state, a condition, an action and a next state, wherein the action comprises: processing the model identification;
the session flow receiving module is used for acquiring a session flow fed back by the client, and the session flow comprises a plurality of flow node identifications and node flow directions;
the flow node information acquisition module acquires corresponding flow node information according to a plurality of flow node identifiers in the session flow;
the session framework construction module is used for obtaining a finite state machine model table of the session flow as a session framework according to the flow node information acquired by the node flow direction organization in the session flow;
the processing model calling module is used for calling the corresponding processing model from the database according to the processing model identification included in each action in the finite state machine model table;
the model training flow acquisition module calls a model training flow of the model from a model training engine according to the processing model identification;
a training sample acquisition module for acquiring labeled model training data;
and the model training module is used for training the processing model according to the labeled model training data and the model training process.
8. The machine-learning based session framework building apparatus according to claim 7, wherein the current state is a state in which a dialog is currently located; triggering an action or executing one state migration when the condition is met, wherein the secondary state is a new state to be migrated after the condition is met;
the session framework building module comprises:
the architecture organization unit sequences the acquired process node information according to the node flow direction in the session process;
and the configuration unit is used for configuring the secondary state of each state in the sequenced process node information so as to carry out node jump according to the flow direction of the nodes in the conversation process.
9. The machine-learning based session framework building apparatus of claim 8, wherein the secondary state comprises: the address bar is used for indicating a new state to be migrated after the condition is met;
the configuration unit includes:
and the jump configuration subunit inserts the jump address of the next node into an address column in the secondary state of the last state of the current node.
10. The machine-learning based session framework building apparatus of claim 7, wherein the actions comprise: the identity of the rule template is identified,
the session framework construction device further comprises:
and the rule template calling module calls the corresponding rule template from the database according to the rule template identification included in each action in the finite state machine model table.
11. The machine-learning based session framework building apparatus of claim 10, wherein the process model comprises: a word segmentation model, a text classification model or an element extraction model.
12. The machine-learning based session framework building apparatus of claim 7, further comprising:
a test sample acquisition module for acquiring the labeled model test data;
and the model testing module is used for testing the processing model by using the marked model testing data.
13. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the machine learning based session framework construction method according to any one of claims 1 to 6 when executing the program.
14. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the machine learning based session framework construction method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910694635.9A CN110413758B (en) | 2019-07-30 | 2019-07-30 | Session framework construction method and device based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910694635.9A CN110413758B (en) | 2019-07-30 | 2019-07-30 | Session framework construction method and device based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110413758A CN110413758A (en) | 2019-11-05 |
| CN110413758B true CN110413758B (en) | 2022-03-25 |
Family
ID=68364196
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910694635.9A Active CN110413758B (en) | 2019-07-30 | 2019-07-30 | Session framework construction method and device based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110413758B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112767930A (en) * | 2019-11-06 | 2021-05-07 | 苏州思必驰信息科技有限公司 | Voice skill dialog manager configuration method, man-machine dialog method and system |
| CN111241257B (en) * | 2020-01-03 | 2023-07-21 | 联想(北京)有限公司 | Information processing method and electronic device |
| CN111309603A (en) * | 2020-02-06 | 2020-06-19 | 北京奇艺世纪科技有限公司 | Task flow information storage method and device, electronic equipment and storage medium |
| CN111563151B (en) * | 2020-05-07 | 2024-02-02 | 腾讯科技（深圳）有限公司 | Information acquisition method, session configuration method, device and storage medium |
| CN111638871A (en) * | 2020-05-28 | 2020-09-08 | 深圳市元征科技股份有限公司 | Equipment information processing method, device, equipment and medium |
| CN112231027A (en) * | 2020-09-27 | 2021-01-15 | 中国建设银行股份有限公司 | Task type session configuration method and system |
| CN112363869B (en) * | 2020-11-19 | 2022-03-01 | 中国核动力研究设计院 | Nuclear DCS (distributed control System) backtracking method and system based on three-state variable finite state machine |
| CN113379590B (en) * | 2021-06-07 | 2023-06-30 | 腾讯科技（上海）有限公司 | Animation data processing method, device, computer equipment and storage medium |
| CN113609831A (en) * | 2021-08-05 | 2021-11-05 | 北京金堤科技有限公司 | Method and device for generating product delivery flow template |
| CN114221991B (en) * | 2021-11-08 | 2023-07-04 | 北京智优集品科技有限公司 | Session recommendation feedback processing method based on big data and deep learning service system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109514586A (en) * | 2019-01-30 | 2019-03-26 | 第四范式（北京）技术有限公司 | Realize the method and system of intelligent customer service robot |
| CN109743346A (en) * | 2018-11-19 | 2019-05-10 | 众安信息技术服务有限公司 | Session process configuration method, device and computer readable storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190220774A1 (en) * | 2015-01-23 | 2019-07-18 | Conversica, Inc. | Systems and methods for configuring message exchanges in machine learning conversations |
-
2019
- 2019-07-30 CN CN201910694635.9A patent/CN110413758B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109743346A (en) * | 2018-11-19 | 2019-05-10 | 众安信息技术服务有限公司 | Session process configuration method, device and computer readable storage medium |
| CN109514586A (en) * | 2019-01-30 | 2019-03-26 | 第四范式（北京）技术有限公司 | Realize the method and system of intelligent customer service robot |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110413758A (en) | 2019-11-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110413758B (en) | Session framework construction method and device based on machine learning | |
| US10949748B2 (en) | Deep learning of bots through examples and experience | |
| US20180025726A1 (en) | Creating coordinated multi-chatbots using natural dialogues by means of knowledge base | |
| CN108804643A (en) | Flow dispensing unit, method and the intelligent robot interactive system of more wheel sessions | |
| CN106372059A (en) | Information input method and information input device | |
| CN109983460A (en) | For developing the service of dialogue drive-type application program | |
| US10956480B2 (en) | System and method for generating dialogue graphs | |
| US11888941B2 (en) | Function-as-a-service for two-way communication systems | |
| CN104813311A (en) | System and methods for virtual agent recommendation for multiple persons | |
| CN110956956A (en) | Voice recognition method and device based on policy rules | |
| KR20210157848A (en) | Computer-implemented conference reservation method and apparatus, device, and medium | |
| EP3543875A1 (en) | Conversation context management in a conversation agent | |
| US20170131978A1 (en) | Automated offline application (app) generation system and method therefor | |
| CN111679811B (en) | Web service construction method and device | |
| US20220366913A1 (en) | Systems and method for third party natural language understanding service integration | |
| CN112231027A (en) | Task type session configuration method and system | |
| CN112802478B (en) | Hybrid cloud fusion scheduling method and device | |
| CN109621406A (en) | A kind of generation method and device of configuration file | |
| US11188548B2 (en) | Profile data store automation via bots | |
| CN111176622B (en) | Implementation method, management system and management method for unified management of voice skills of solution | |
| CN110852799A (en) | User screening method and device based on intention label, electronic equipment and medium | |
| KR102233235B1 (en) | Intelligence collaboration assistant system based on analyzing user intent from message | |
| CN109933314A (en) | The generating means of more wheel question answering systems | |
| CN112487170A (en) | Scene configuration-oriented human-computer interaction dialogue robot system | |
| KR20210062189A (en) | Collaboration assistant method based on analyzing user intent from message |
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 |