CN109146081A - It is a kind of for quickly creating the method and device of model item in machine learning platform - Google Patents
It is a kind of for quickly creating the method and device of model item in machine learning platform Download PDFInfo
- Publication number
- CN109146081A CN109146081A CN201710502054.1A CN201710502054A CN109146081A CN 109146081 A CN109146081 A CN 109146081A CN 201710502054 A CN201710502054 A CN 201710502054A CN 109146081 A CN109146081 A CN 109146081A
- Authority
- CN
- China
- Prior art keywords
- model item
- model
- machine learning
- node
- component
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Stored Programmes (AREA)
Abstract
This application discloses a kind of method and apparatus for quickly creating model item in machine learning platform, wherein the described method includes: creation machine learning running environment；Component needed for adding model item into the machine learning running environment；Component needed for the model item is established into input and output link according to the sequence that model item is set to create model item；Component needed for model item to the creation carries out parameter configuration；Run the model item of the creation.User can substantially be shortened using method provided by the present application and create model terms object time in machine learning platform.
Description
Technical field
This application involves machine learning fields, and in particular to one kind is for quickly creating model item in machine learning platform
Method and device.
Background technique
With the rapid development of artificial intelligence in recent years, machine learning as artificial intelligence main implementation method also
Rapid development, major Internet company are proposed the machine learning platform of oneself one after another.Pass through machine learning platform, Wo Menke
To obtain reasonable model item according to existing data, it not only can solve some of scientific research using these model items
Problem, also can be applied to the every field in real life, our life is energetically instructed the activities such as to produce.Machine learning
Platform is even repeatedly iteratively repeated this process to improve, verify model terms by creation and the experiment of moving model project training
Mesh is until obtain satisfied model item, to finally be created that model item.
It needs first to create experiment when the machine learning platform of mainstream creates model item at present, then the demand according to experiment
It is pulled in the component experiment from component tray, after then carrying out line and parameter configuration to component, then runs the experiment.It is aforementioned
When the method for experiment creates experiment every time, component needed for individually pulling experiment one by one from corresponding component tray is required, it is non-
It is often time-consuming and inconvenient, also cause the ease for use of machine learning platform poor.
Summary of the invention
The application provides the method for quickly creating model item in machine learning platform, to solve existing machine learning
The above problem of platform creation model item.The application also provides in a kind of machine learning platform simultaneously and quickly creates model item
Device.
A kind of method for quickly creating model item in machine learning platform provided by the present application, comprising:
Create machine learning running environment；
Component needed for adding model item into the machine learning running environment；
Required component is established into input and output link according to the sequence that model item is set to create model item；
Parameter configuration is carried out to component needed for the model item；
Run the model item of the creation.
Optionally, the creation machine learning running environment includes:
Model item panel is created in the machine learning platform；
Identification code is distributed for the model item panel of the creation.
Optionally, the component needed for addition model item into the machine learning running environment includes:
Component is added into the model item panel；
The attribute information of the component of addition is inserted into model item faceplate formation table；
Wherein, the attribute information includes model item panel identification code where the component of the addition, described to add
Position letter of the component of the component Name, component recognition code and the addition that add in the model item panel
Breath.
Optionally, the component includes algorithm assembly and data source component.
Optionally, it is described by required component according to the sequence that model item is set establish input and output link include:
It is encoded for model item distribution model project name to be run and model item；
Component needed for selecting the model item；
Data or order input are established between the component needed for the model item according to the sequence that model item is set
Output link.
Optionally, it is described by required component according to the sequence that model item is set establish input and output link include:
Component needed for preference pattern project；
Judge whether component to be selected includes father node or parallel node, and the automatic distribution model entry name when not including
Title and identification code；
Data or order input are established between the component needed for the model item according to the sequence that model item is set
Output link；Wherein,
Various components in model item constitute model terms destination node, and father node is a upper node for selected node, and
Row node is the node that selected node has common parent.
Optionally, component needed for preference pattern project includes that selection adds into the machine learning model project operation environment
The component added, and/or
It is already present in model item running environment or has been run other model items in environment and used
Component.
Optionally, further includes: model item information and model item panel identification code are inserted into model item information table；
Wherein, the model item information includes needed for model item title, model item coding and the model item
Component and its attribute information.
It optionally, further include that each nodal information for constituting the model item is inserted into informational table of nodes；
Wherein, the nodal information includes node location, node correlation father node and child node information.
Optionally, carrying out parameter configuration to component needed for the model item includes:
The parameter information of algorithm assembly needed for configuring the model item；
Parameter information described above, node identification code and model item coding are inserted into node parameter table.
Optionally, the model item of the operation creation includes
Model item to be run is selected, and starts moving model project；
It is encoded according to the model item, reads all node identification codes in the model item；
According to set membership of the node identification code in node table, node sequencing is given；
Node is sequentially carried out according to the node sequencing.
Optionally, further includes: node is inserted into node execution journal table after executing and executes record；And it is run in model item
State table insertion model item, which is executed, to model item after end executes status data.
Optionally, described to sequentially carry out node according to the node sequencing and include:
For node described in each, according to model item coding and the node identification code from the node parameter
The parameter information of the node is obtained in table；
Judge that the node belongs to the data source nodes or the algorithm node；
Data are obtained into the correspondence database of machine learning platform when judging result is the data source nodes；
The parameter of this node is obtained from node parameter table when judging result is the algorithm node and is joined according to algorithm
Number executes algorithm.
Optionally, the component needed for addition model item into the machine learning running environment specifically:
The component needed for addition same model project into the machine learning running environment；Or
It is described add different model items into the machine learning running environment needed for component.
In addition, the application also provides one kind for quickly creating model terms destination device in machine learning platform, comprising:
Creating unit, for creating machine learning running environment；
Component adding unit, for component needed for adding model item into the machine learning running environment；
Connection unit, for required component to be established input and output link according to the sequence that model item is set to create
Model item；
Parameter configuration unit, for carrying out parameter configuration to component needed for the model item；
Running unit, for running the model item of the creation.
Optionally, the creating unit includes:
Panel creating unit, for creating model item panel in the machine learning platform；
Identity assignments unit, for distributing identification code for the model item panel of the creation.
Optionally, the component adding unit includes:
Assembly unit, for adding component into the model item panel；
Attribute is inserted into unit, for the attribute information of the component of addition to be inserted into model item faceplate formation table
In；
Wherein, the attribute information includes model item panel identification code where the component of the addition, described to add
Location information of the component of the component Name and the addition that add in the model item panel.
Optionally, the connection unit includes:
Model item allocation unit, for being encoded for model item distribution model project name to be run and model item；
Selecting unit, for component needed for selecting the model item；
Link unit, the sequence for setting according to model item establish number between the component needed for the model item
According to or order input and output link.
In addition, the application also provides a kind of electronic equipment comprising:
Pocessor and storage media；
The storage medium stores or carries the method for quickly creating model item in machine learning platform；This sets
It is standby to be powered and by the program of the processor operation method for being used in machine learning platform quickly create model item
Afterwards, operations described below is executed:
Create machine learning running environment；
Component needed for adding model item into the machine learning running environment；
Required component is established into input and output link according to the sequence that model item is set to create model item；
Parameter configuration is carried out to component needed for the model item；
Run the model item of the creation.
In addition, the application also provides a kind of method for quickly creating model item in machine learning platform, comprising:
Create machine learning running environment；
Component needed for adding model item into the machine learning running environment；
In the academic environment, the component based on addition creates model item；
Run the model item of the creation.
Compared with prior art, the one aspect of the application has the advantage that embodiments herein is created first
A machine learning running environment is built, component needed for then adding model item into the environment needs to create model terms every time
When purpose, the sequence that component needed for model item is set according to model item, which is directly carried out line, can create model terms
Mesh can moving model project after then carrying out parameter configuration to component needed for model item again.Compared with prior art,
This method is due to adding component into machine learning running environment first, as long as creating model item every time later for model item
The sequence that required component is set according to model item, which carries out line, can create model item, and need not be by pulling mould one by one
Component needed for type project adds component needed for model item, highly shortened creation model terms object time, while also mentioning
The ease for use of machine learning platform is risen.
Detailed description of the invention
Fig. 1 is a kind of for quickly creating the stream of the embodiment of the method for model item in machine learning platform of the application
Cheng Tu；
Fig. 2 is the method flow diagram of creation machine learning running environment provided in this embodiment；
Fig. 3 is the method flow of the component provided in this embodiment needed for addition model item into machine learning running environment
Figure；
Fig. 4 is that provided in this embodiment establish component needed for model item according to the sequence that model item is set inputs
Export the method flow diagram of link；
Fig. 5 is the method flow diagram provided in this embodiment that parameter configuration is carried out to component needed for model item；
Fig. 6 is the method flow diagram of the created model item of operation provided in this embodiment；
Fig. 7 is the method flow diagram provided in this embodiment that node is executed according to the node sequence；
Fig. 8 shows the schematic diagram of the panel created on machine learning platform；
Fig. 9 shows the process of node line and the content of relevant parameter information table；
Figure 10 is the flow chart that the model item of embodiments herein is run；
Figure 11 is the algorithm node execution flow chart of embodiments herein；
Figure 12 is the Installation practice for quickly creating model item in machine learning platform of the embodiment of the present application
Schematic diagram；
Figure 13 is a kind of model item that the quick creation model item method using the embodiment of the present application is established in platform
Component annexation figure.
Specific embodiment
In the following description, numerous specific details are set forth in order to facilitate a full understanding of the present invention.But the present invention can be with
Much it is different from other way described herein to implement, those skilled in the art can be without prejudice to intension of the present invention the case where
Under do similar popularization, therefore the present invention is not limited to the specific embodiments disclosed below.
By excavating valuable data in mass data, it is of great significance to science and social progress.With shifting
The fast development of dynamic internet, human society have inevitably entered the epoch of big data.Compared with traditional data, greatly
Data have magnanimity, isomery, the data for repeating even conflict, this make traditional data mining method no matter from data shape or
From computation model, or can not all continue to be competent at new demand from computational efficiency.Based on this, cloud computing, cloud service rapidly at
For research hotspot.Machine learning cloud platform is also come into being.
Machine learning platform passes through distributed cloud computing platform, realizes the functions such as data mining, modeling, prediction.For user
The one-stop algorithm services such as algorithm development, sharing, model training, deployment, monitoring are provided, in some machine learning platforms, are used
Family can also be operated by visual operation interface entirely establishes model experiment process, while also supporting command mode, allows
User is by order line come operation experiments modeling.By machine learning platform, user or developer's exploitation and portion can be greatly reduced
Affix one's name to the threshold of distributed machines learning system and related application.It is provided herein a kind of for quickly being created in machine learning platform
The method of established model project.In this way, user or associated developer can quickly create its desired computation model project or
Person's experimental model project, and fast verification its feasibility.Machine learning platform is efficiently quickly utilized to easily facilitate developer
Development model project.A kind of method for quickly creating model item in machine learning platform of the application includes: creation machine
Device learning model project operation environment；Component is added into the machine learning model project operation environment；By model item institute
The component needed is established input and output according to the sequence that model item is set and is linked to create model item；To the model item
Required component carries out parameter configuration；Run the model item of the creation.It is further explained below.
Referring to FIG. 1, its flow chart for the embodiment of the application method.Being used for for the application is fast in machine learning platform
The method of speed creation model item includes the following steps:
Step 101: creation machine learning model project operation environment.
The model item refer to that machine learning platform user builds on machine learning platform by several or a series of
The data workflow or data application of component composition.Model item is also referred to as computation model project, also includes for verifying spy
The experimental model project determining function and temporarily or for a long time creating.Embodiments herein is said by taking experimental model project as an example
It is bright.The rule between available data can be found out by model item, is finally created that the model item for reflecting the rule.
The machine learning running environment is to be used to create the virtual environment of simultaneously moving model project in machine learning platform,
Model item can be created in the present context, and runs created model item.
The method of the creation machine learning model project operation environment, includes the following steps:
Step 101-1: model item panel is created on the machine learning platform.
The model item panel refers to the machine learning model project operation environment of panel-form, can be on the panel
It creates and runs the model item.Unlike traditional model item interface in machine learning platform, the model
Item panel can also store component on it in directly visible form.It can be by clicking on the machine learning platform
The button of creation model item panel the model item panel is created on the machine learning platform.User can be to institute
The model item panel of creation, which is named, rename etc., to be operated.
The model item panel is provided with scroll bar, in order to check, operate entire model item panel.
Multiple model item panels can be created simultaneously.
It can be created simultaneously on the same model item panel and run multiple model items；And when same
It, can be on back-end server according to the use of number of users and component when the multiple model items of Shi Yunhang lead to page degradation
Frequency carries out dilatation.
User may also select whether to be stored on the model item panel run successfully the model item and its
The model terms can be set when selection saves the model item of the model item and its generation in the model item of generation
Mesh or its preservation title, position and the life cycle of preservation of the model item that generate etc. save parameter.
Step 101-2: for the model item panel distribution model Item panel identification code of the creation.
The model item panel identification code refers to the coding for identifying the model item panel.The model item
Panel identification code and the model item panel correspond, and are unique encodings.
When creating model item panel success on the machine learning platform, background program will be created
The model item panel distributes the model item panel identification code.
As Fig. 8 shows the schematic diagram of the panel created on machine learning platform.As shown in figure 8, the experimental model of creation
" experiment operation panel " in Item panel such as figure.Below it is panel zone, experimental model item can be created by pulling component
Mesh, and simulated.Wherein, experimental model project (referred to as testing in Fig. 8 into Figure 11) refers to that machine learning platform user is taken
The data workflow or data application built；User needs first to establish an experiment embodiment, then builds in test panel
Data flow.
With continued reference to FIG. 1, step 102: component needed for adding model item into the machine learning running environment.
Specifically, in this step, according to the needs of model item or project etc. into the model item running environment
Add component.
The component refers to can call what is executed to be used to represent various algorithms or number on the machine learning platform
According to the operating unit in source.Such as data exporting, data processing, data analysis, model item training or prediction.
The component for representing algorithm is algorithm assembly, and the algorithm assembly has parameter configuration column for configuring each calculation
The parameter of method at runtime, and there is input, output button to be used for connection and algorithm between each algorithm on the algorithm assembly
The output of operation result is checked.
The component for representing data source is data source component, and the data source component is mainly used to that table name is arranged, once
After table name is arranged in data source component, system will go in background data base to read the data of corresponding table name, be supplied to subsequent calculation
Method component uses.
The component includes the algorithm assembly and the data source component.
The method of the component needed for addition model item into the machine learning running environment, includes the following steps:
Step 102-1: component is added into the model item panel.Wherein, the component can be same model item
Purpose component, or the component of different model items.The present embodiment is said for adding same project model component
It is bright.But it should be recognized that the application is not limited thereto, the component of different model items can be simultaneously or sequentially added, and
In the next steps, the component for belonging to same model project is established into input and output link according to sequencing.
Pass through the component for pulling the corresponding assembly in component tray one by one into the model item panel to be pulled
It is added in the model item panel.
Step 102-2: the attribute information of the component of addition is inserted into model item faceplate formation table；
Wherein, the attribute information includes model item panel identification code where the component of the addition, described to add
Location information of the component of the component Name and the addition that add in the model item panel.
The model item faceplate formation table, which refers to, records the described above of all components in the model item panel
The database table of attribute information.
When adding component success into the model item panel, background program is by the place of the added component
Attribute information described in location information in model item panel identification code, component Name and its described model item panel etc. is inserted
Enter into model item faceplate formation table to record the added component.
Step 103: by component needed for model item according to the sequence that model item is set establish input and output link with
Create model item.
Component needed for the model item refers to the algorithm assembly and data that model item to be created needs
The summation of source component.
The sequence of the model item setting refers to the sequencing of the execution step of the model item.
The model terms that the sequencing for the step of input and output link refers to according to the model item carries out
Line needed for mesh between algorithm assembly and data source component.The input and output for building the data or order between component are logical
Road.
It is described that component needed for model item is established into input and output link according to the sequence that model item is set to create
The method of model item, includes the following steps:
Step 103-1: it is encoded for model item distribution model project name to be run and model item.
The model item to be run, which refers to, sets component needed for all model items according to the model item
Several companies in component needed for the model item or the model item that the fixed creation being linked in sequence together is completed
The model item created being connected together.
The model item coding refers to the coding for identifying the model item.Model item coding with it is described
Model item corresponds, and is unique encodings.
For each model item to be run, background program is assigned with one with regard to automated randomized when all it is created
A model item title and corresponding model item coding.
Step 103-2: component needed for selecting the model item.
Component needed for component needed for the selection model item refers to two model items of selection according to
The model item setting sequence carries out line.
Component needed for above-mentioned two model item for carrying out line is according to the model item setting sequence
It is followed successively by parent component and sub-component.
Component needed for the preference pattern project includes adding from the machine learning model project operation environment
It selects, or is selected from the component being already present in the model item running environment, Huo Zhecong in the component
Selected in the used component of the institute of model item described in other.Or in a certain project model project, simultaneously
From it is aforementioned all or any two from select component, to meet needed for model item or project.
Step 103-3: judge whether selected component includes father node or parallel node, and distributed automatically when not including
Model item title and model item coding；
Wherein, each component in the model item constitutes model item node, abbreviation node；The father node
For the upper one model item node of the selected model item node；The parallel node is the selected model terms
Mesh node has the model item node of the common father node.
Judge whether selected component includes that father node or parallel node refer to the father judged in the step 103-2
Whether component includes father node or parallel node.
If the parent component does not include the father node or the parallel node, show the sub-component and its
Line creates a new model project, and background program is automated randomized its model of distribution of new model project for being created at this time
Project name and model item coding；
If the parent component includes the father node or the parallel node, show to connect the sub-component with it
Line is to add the component comprising model item to be run described in the parent component, and there is no creation new models at this time
Project.
Step 103-4: model item information and the model item panel identification code are inserted into model item information table
In；
The model item information includes the model item title, model item coding, model item institute
The attribute information of component needed for the component and the model item that need.
The model item information table refers to the model item described above letter for recording all model items to be run
The tables of data of breath and the model item panel identification code where it.
When creating a new model project in step 103-3 described above, the new model project created also at
It, then will be where the affiliated model item information for the new model project that created and the model item for model item to be run
Model item panel identification code is also inserted into the model item information table.
Step 103-5: each nodal information for constituting the model item is inserted into informational table of nodes.
The nodal information includes the location information of the node, the relevant father node of the node and child node information.
The child node of the node is model item node described in the next stage of the model item node.
The relevant father node of the node refers to all father nodes of the model item node.
The relevant child node of the node refers to all child nodes of the model item node.
The informational table of nodes refers to the upper of all nodes for recording that all model items to be run are included
State the tables of data of the nodal information.
When creating a new model project in step 103-3 described above, step described above is not only executed
The nodal information described above for constituting each model item node of the model item will be also inserted by 103-4 simultaneously
In the informational table of nodes.
Step 103-6: data are established between the component needed for the model item according to the sequence that model item is set
Or order input and output link
This step refers to aforementioned four step 103-2,103-3,103-4,103-5 described in repetition, successively by model item
Together with the sequence line that required whole algorithm assemblies and data source are set according to model item, to create model item.
Step 104: parameter configuration is carried out to component needed for the model item；
The method for carrying out parameter configuration to component needed for model item described above, includes the following steps:
Step 104-1: the parameter information of algorithm assembly needed for configuring the model item.
The parameter configuration column of algorithm assembly needed for carrying out the model item of parameter configuration described in opening, choosing
Required parameter is configured when selecting algorithm operation.If such as a certain algorithm node is random forests algorithm node, configuration
It sets number, the parameters such as tree depth, training column, target column.
Step 104-2；Parameter information described above, node identification code, nodename and model item identification code are inserted into
Into node parameter table.
The parameter information of algorithm assembly needed for the model item includes the parameter name and ginseng of its all parameter
Number Configuration Values.
The node parameter table refers to the node identification code for recording the node that the model item is included, node name
The tables of data of the parameter information of title, associated model item identification code and the corresponding component of the node.
After the parameter configuration of algorithm assembly needed for completing the model item by step 104-1 described above, 104-2
Step background program executes automatically.
Fig. 9 shows the process of node line and the content of relevant parameter information table, above-mentioned steps 103 are related to 104
Content can also can refer to Fig. 9.
Through the above steps, that is, the creation of model item is completed, further, the dry run model can be passed through
Project checks the operation result of the model item to simulate, and whether the model item built can achieve expected purpose.Please after
It is continuous to refer to Fig. 1.
Step 105: running the model item of the creation.
The method of the model item of the operation creation, includes the following steps:
Step 105-1: the selection model item to be run, and start moving model project.
After user selects the model item to be run in such a way that frame selects, by pressing on the machine learning platform
The button of preset moving model project starts moving model project.
Since the machine learning platform supports multiple model items to run simultaneously, user can select with time-frame and multiple transport
Row model item makes multiple model items of institute's frame choosing while running.
If user does not have frame to select the model item to be run, the default choice machine learning model project operation
All model items in environment.Certainly, platform can choose the model item mode of default, not have for example, it is also possible to default
It is framed not run any model item when selecting.
Step 105-2: according to the model item identification code, all node identification codes in the model item are read.
When the model item that the step 105-1 institute's frame described above that brings into operation selects, background program reads the institute of operation first
There is the model item of model item to encode, then encoded according to the read model item, reads the model item and compile
The node identification code of all nodes in the corresponding model item of code.
Step 105-3: according to set membership of the node identification code in node table, node sequencing is given；
Then step 105-2 described above reads the node identification code each read in the informational table of nodes
In all father nodes and child node, to determine the father and son between all nodes that run model item includes
Relationship, and all node sequencings for giving run model item to include according to this set membership.
Step 105-4: node is sequentially carried out according to the node sequencing.
Then step 105-3 described above, after all node sequencings for including to the model item run,
This step will be executed.
The method for sequentially carrying out node according to the node sequencing, includes the following steps:
Step 105-4-1: be directed to each described node, according to the model item coding and the node identification code from
The parameter information of the node is obtained in the node parameter table；
When one node of every execution, background program first will be according to performed node identification code mould corresponding with its
Type project code reads the parameter information of the node from the node parameter table.
Step 105-4-2: judge that the node belongs to the data source nodes or the algorithm node；
The data source nodes refer to that component corresponding to node is data source component.The algorithm node refers to node institute
Corresponding component is algorithm assembly.Judged the node for data source section according to the node parameter information read from node parameter table
Point or algorithm node.
Step 105-4-3: it when judging result is the data source nodes, is obtained from the database of machine learning platform
Data；
If the judging result of step 105-4-2 described above is data source nodes, from the machine learning platform
Data are obtained in corresponding database.For machine learning platform, which is broadly divided into three layers, and first layer is
The interface Web UI, the second layer are IDST algorithm layers, and the last layer is database platform layer.This step has in machine learning platform
Body refers to the reading table data from corresponding database.
Step 105-4-4: the parameter of this node is obtained from node parameter table when judging result is the algorithm node
And algorithm is executed according to algorithm parameter.
If the judging result of step 105-4-2 described above is algorithm node, according to step 105-4-1 described above
The parameter information of the algorithm node read executes the corresponding algorithm of the algorithm node.
Specifically, parameter information is spliced into machine learning order line form according to the format of agreement.Background program will
This machine learning order is sent to Database Systems by scheduling system, and Database Systems are according to parsing machine learning command format
In parameter, obtain algorithm where memory space and algorithm title and algorithm parameter configuration.Database Systems calling refers to
Determine the target algorithm in algorithm memory space and algorithm is executed according to algorithm parameter.Algorithm returns to the result of algorithm after running successfully
Table name or the algorithm model project name of generation, as a result table is stored in the corresponding database project space of user's current system
In (simple understand be exactly database), algorithm model project pmml file is stored in the ODPS volumn that background system is specified.
Algorithm node execution flow chart can be referring to attached drawing 11.
Step 105-5: node is inserted into node execution journal table after executing and executes record, and in model item end of run
Backward model item executes state table insertion model item and executes status data.
The node execution journal table refers to the database table for recording node health；
The model item executes state table and refers to the database table for recorder model item operation result.
The model item operation result includes the runing time of model item, operation result.
One node of every execution, background program all will be into the node execution journal tables in step 105-4 described above
A data is inserted into record the practice condition of node.
If the node runs succeeded, the execution journal of the node is recorded；
If the node executes failure, the execution failure log of the node is recorded；At the same time, described to execute mistake
Model item corresponding to the node lost will stop running, and execute to model item and be inserted into the model item fortune in state table
Capable result data；
If all nodes of the model item of institute's frame choosing run succeeded, the model item of institute's frame choosing is run successfully, and
Also the operation result data that the model item is inserted into state table are executed to model item.
The process of experimental model project operation can also be referring to attached drawing 10.
In the present embodiment, the model item, model item information, module information and the component parameter configuration that are generated in panel
After be stored in the corresponding table of mysql, can also be stored in other kinds of table such as rds, sql server etc. or other
Storage system such as lds, cache etc..When the model item operation generated in the present embodiment, the memory space of algorithm packet and data is
The project space of database.Wherein running environment may be ECS, and storage system is also possible to OSS etc..Here not reinflated opinion
It states.
Illustrate the quick creation model item method of the application below by a specific example, as shown in figure 13,
It illustrates the model item component connections that a kind of quick creation model item method using above-described embodiment is established in platform
Relational graph.In order to be predicted for haze weather, after obtaining weather forecasting data, can be carried out by data prediction component
Pretreatment, then by treated, data carry out model item training, and the model item and test data that train is used to predict
Weather conditions and assessment prediction result.It is specific as follows: by turning input data source (weather forecasting data package) by type
It is split as two class data i.e. training data and test data after changing (data prediction), training data is input to random forest
Two kinds of engineerings of (random forest -1 in attached drawing 13) and logistic regression two classification (classification of logistic regression two -1 in attached drawing 13)
It practises in component, obtains training pattern project respectively；Then the test data using training pattern project and after splitting is as prediction group
The input of part predicts whether the corresponding weather of every data line is haze weather in test data；And it is assessed by evaluation component
Prediction effect；
For above-mentioned haze weather prognostic experiment model item, can method through the foregoing embodiment establish such as Figure 13
Shown in experiment flow, component is pulled by panel according to experiment demand, and connection is established according to experiment flow respectively, that is, established
Interim experiment.If the output model project of two classification component of random forest or logistic regression is only needed to carry out other work,
If model item is disposed online, online weather prediction service of opening to the outside world, so that it may be built in test environment (i.e. test panel)
Vertical interim experiment, operation pilot scale study save experimental model project；It simultaneously can be according to the model of random forest and logistic regression
Which experimental model project the selection of project forecast assessment result saves；Without establishing experiment in test environment, and can adjust
Try experiment parameter, contrast and experiment；It is haze weather prediction in this experiment simultaneously, is predicted for two classification of others, we
It only needs to modify data source component, shares the remaining components in this experiment；User can be according to random forest and logic
The superiority and inferiority for returning the experimental result of two classification, which modifies to experiment, (only to be saved the forecast assessment branch of random forest or only saves
The forecast assessment branch that logistic regression two is classified) then save experiment.
It, will be real by creating the running environment for quickly creating in machine learning platform in embodiments herein
The component tested is put into the experimental situation, the analog running experiment in experimental situation, and is quickly obtained experimental result in time.It is flat
Platform system is running environment allocation identification region (i.e. with the panel of identification code), and experiment package is built on the panel, backstage
To record experiment information, parameter information, nodal information and mutual related information by the parameter list of different dimensions.It is real
When testing operation, by parameter list and relevant data and order are called, the operation of experiment can be not only realized with fast and flexible, and
Component can be multiplexed on panel, and same component can be used in different experiments, pass through experimental identification code determination component and associated section
Point is called by which experiment, multiple can be tested in panel and not interfere with each other, concurrently runs.
Above-described embodiment of the application can save user's a large amount of time and physical strength, and user only needs moving model project
When create a model item trial operation panel, pull algorithm assembly into panel, these algorithms can be used always
Component forms all kinds of temporary pattern projects, and can in the panel trial operation model item, debugging model project, check model terms
Mesh effect, just chosen after model item is run successfully whether preservation model project or save temporary pattern project generate model
Project.It only needs to protect if user carries out functions, the users such as subsequent data prediction it is only necessary to the model item of generation
Model item is deposited, without additional creation model item, moving model project generates model item.Each user wants to appoint
When what model item, it is only necessary to " trial operation is once " in the panel.
Furthermore after creating panel in embodiments herein, the algorithm assembly that all possibility are used all is drawn to panel
On, when algorithm assembly is more, the page may can create multiple panels, each panel than comparatively dense confusion with user
Identification code it is unique, and the component for including in each panel can be pulled arbitrarily from component tray." panel " contains rolling simultaneously
Dynamic item, allows panel that very big region is presented.In the same panel, while when running multiple model items, the page
Can be possible poor, server back end can carry out dilatation according to number of users and component frequency of use thus；
In addition, the temporary pattern project in the application can be with preservation model project after trial operation success, it is only necessary to set
Model item title, save location and model item life cycle are set, can be obtained the successful model item of operation；When
When only needing model item, user only needs the corresponding temporary pattern project of trial operation primary, without creating model
The cumbersome process such as project；
In addition, the algorithm assembly in panel will not be varied, user because of the success or failure that model item is run
These algorithm assemblies can be repeatedly used, no longer need to be pulled from the algorithm assembly column in the machine learning page.
In addition, can establish temporary pattern project in " panel " carries out smoke test, in addition to this it is possible to establish interim
Model item is only for input output model project data, pretreated model project data, acquisition model item model item etc..
In addition, can choose whether the algorithm groups in change " panel " when the algorithm assembly in component tray is changed
Part, such as new and old algorithm assembly, add newly-increased algorithm assembly etc., have very strong flexibility.
In the above-described embodiment, a kind of method for quickly creating model item in machine learning platform is provided,
Corresponding, the application also provides a kind of for quickly creating model terms destination device in machine learning platform.The device
It is corresponding with the embodiment of the above method.
Embodiment as described above provides a kind of method for quickly creating model item in machine learning platform.This
A kind of method for quickly creating model item in machine learning platform of application can also be realized by the following method: creation
Machine learning running environment；Component needed for adding model item into the machine learning running environment；In the academic environment, base
Model item is created in the component of addition；Run the model item of the creation.Here not reinflated discussion.
Figure 12 is please referred to, is embodiments herein for quickly creating the dress of model item in machine learning platform
Set the schematic diagram of embodiment.Since Installation practice is substantially similar to embodiment of the method, so describe fairly simple, correlation
Place illustrates referring to the part of embodiment of the method.Installation practice described below is only schematical.The present embodiment
One kind is for quickly creating model terms destination device in machine learning platform, comprising:
Creating unit 201, for creating machine learning running environment；
Component adding unit 202, for component needed for adding model item into the machine learning running environment；
Connection unit 203, it is defeated for component needed for model item to be established input according to the sequence that model item is set
It is linked out to create model item；
Parameter configuration unit 204, for carrying out parameter configuration to component needed for the model item；
Running unit 205, for running the model item of the creation.
Optionally, the creating unit includes:
Panel creating unit, for creating model item panel in the machine learning platform；
Identity assignments unit, for distributing identification code for the model item panel of the creation.
Optionally, the component adding unit includes:
Assembly unit, for adding component into the model item panel；
Attribute is inserted into unit, for the attribute information of the component of addition to be inserted into model item faceplate formation table
In；
Wherein, the attribute information includes model item panel identification code where the component of the addition, described to add
Location information of the component of the component Name and the addition that add in the model item panel.
Optionally, the connection unit includes:
Model item allocation unit, for being encoded for model item distribution model project name to be run and model item；
Selecting unit, for component needed for selecting the model item；
Link unit, the sequence for setting according to model item establish number between the component needed for the model item
According to or order input and output link.
Optionally, the connection unit includes:
Selected unit, for component needed for preference pattern project；
Judging unit, for judging whether component to be selected includes father node or parallel node, and it is automatic when not including
Distribution model project name and identification code；
Link unit, the sequence for setting according to model item establish number between the component needed for the model item
According to or order input and output link；Wherein,
Various components in model item constitute model terms destination node, and father node is a upper node for selected node, and
Row node is the node that selected node has common parent.
Optionally, further includes: model item information recording unit, for model item information and model item panel to be known
Other code is inserted into model item information table；
Wherein, the model item information includes needed for model item title, model item coding and the model item
Component and its attribute information.
It optionally, further include nodal information recording unit, for each nodal information of the model item will to be constituted
It is inserted into informational table of nodes；
Wherein, the nodal information includes node location, node correlation father node and child node information.
Optionally, the parameter configuration unit includes:
Algorithm parameter configuration unit, the parameter information for algorithm assembly needed for configuring the model item；
Node parameter recording unit is used for parameter information described above, node identification code, nodename and model terms
Mesh coding is inserted into node parameter table.
Optionally, the running unit includes
Start unit selects model item to be run, and starts moving model project；
Identification code reading unit is read in the model item according to the coding of the model item in node parameter table
All node identification codes；
Node sequencing unit gives node sequencing according to set membership of the node identification code in node table；
Execution unit sequentially carries out node according to the node sequencing.
Optionally, further includes: log unit executes record to the insertion of node execution journal table after executing for node；Shape
State table generation unit is used to and executes state table insertion model item to model item after model item end of run execute shape
State data.
Optionally, the execution unit includes:
Node parameter transfers unit, for node described in each, according to model item coding and the node mark
Know the parameter information that code obtains the node from the node parameter table；
Nodal community judging unit judges that the node belongs to the data source nodes or the algorithm node；
DEU data execution unit, for judging result be the data source nodes when to machine learning platform corresponding data
Data are obtained in library；
Command executing unit, for executing algorithm according to algorithm parameter when judging result is the algorithm node.
In addition, the application also provides a kind of electronic equipment comprising: pocessor and storage media；
The storage medium stores or carries the method for quickly creating model item in machine learning platform；This sets
It is standby to be powered and by the program of the processor operation method for being used in machine learning platform quickly create model item
Afterwards, operations described below is executed:
Create machine learning running environment；Component is added into the machine learning running environment；It will be needed for model item
Component according to the sequence that model item is set establish input and output link to create model item；To the model item institute
The component needed carries out parameter configuration；Run the model item of the creation.
Although the present invention is disclosed as above with preferred embodiment, it is not for limiting the present invention, any this field skill
Art personnel without departing from the spirit and scope of the present invention, can make possible variation and modification, therefore guarantor of the invention
Shield range should be subject to the range that the claims in the present invention are defined.
In a typical configuration, calculating equipment includes one or more processors (CPU), input/output interface, net
Network interface and memory.
Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM) and/or
The forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM).Memory is computer-readable medium
Example.
1, computer-readable medium can be by any side including permanent and non-permanent, removable and non-removable media
Method or technology realize that information stores.Information can be computer readable instructions, data structure, the module of program or other numbers
According to.The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory
(SRAM), dynamic random access memory (DRAM), other kinds of random access memory (RAM), read-only memory
(ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory techniques, CD-ROM are read-only
Memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or
Other magnetic storage devices or any other non-transmission medium, can be used for storage can be accessed by a computing device information.According to
Herein defines, and computer-readable medium does not include non-temporary computer readable media (transitory media), such as modulates
Data-signal and carrier wave.
2, it will be understood by those skilled in the art that embodiments herein can provide as the production of method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the application
Form.It can be used moreover, the application can be used in the computer that one or more wherein includes computer usable program code
The computer program product implemented on storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Form.
Claims (20)
1. a kind of method for quickly creating model item in machine learning platform characterized by comprising
Create machine learning running environment；
Component needed for adding model item into the machine learning running environment；
Required component is established into input and output link according to the sequence that model item is set to create model item；
Parameter configuration is carried out to component needed for the model item；
Run the model item of the creation.
2. the method according to claim 1 for quickly creating model item in machine learning platform, which is characterized in that
The creation machine learning running environment includes:
Model item panel is created in the machine learning platform；
Identification code is distributed for the model item panel of the creation.
3. the method according to claim 2 for quickly creating model item in machine learning platform, it is characterised in that:
It is described into the machine learning running environment add model item needed for component include:
Component is added into the model item panel；
The attribute information of the component of addition is inserted into model item faceplate formation table；
Wherein, the attribute information includes model item panel identification code where the component of the addition, the addition
The location information of the component Name, component recognition code and the component of the addition in the model item panel.
4. the method according to claim 3 for quickly creating model item in machine learning platform, it is characterised in that:
The component includes algorithm assembly and data source component.
5. the method according to claim 3 for quickly creating model item in machine learning platform, it is characterised in that:
It is described by required component according to the sequence that model item is set establish input and output link include:
It is encoded for model item distribution model project name to be run and model item；
Component needed for selecting the model item；
Data or order input and output are established between the component needed for the model item according to the sequence that model item is set
Link.
6. the method according to claim 3 for quickly creating model item in machine learning platform, it is characterised in that:
It is described by required component according to the sequence that model item is set establish input and output link include:
Component needed for preference pattern project；
Judge whether component to be selected includes father node or parallel node, and when not including automatic distribution model project name and
Identification code；
Data or order input and output are established between the component needed for the model item according to the sequence that model item is set
Link；Wherein,
Various components in model item constitute model terms destination node, and father node is a upper node for selected node, parallel to save
Point has the node of common parent for selected node.
7. special according to any method for quickly creating model item in machine learning platform of claim 5 or 6
Sign is that component needed for preference pattern project includes the group for selecting to add into the machine learning model project operation environment
Part, and/or
It is already present in model item running environment or has been run used group of other model items institute in environment
Part.
8. special according to any method for quickly creating model item in machine learning platform of claim 5 or 6
Sign is, further includes: model item information and model item panel identification code are inserted into model item information table；
Wherein, the model item information include model item title, model item coding and the model item needed for group
Part and its attribute information.
9. the method according to claim 8 for quickly creating model item in machine learning platform, which is characterized in that
It further include that each nodal information for constituting the model item is inserted into informational table of nodes；
Wherein, the nodal information includes node location, node correlation father node and child node information.
10. the method according to claim 9 for quickly creating model item in machine learning platform, feature exist
In carrying out parameter configuration to component needed for the model item includes:
The parameter information of algorithm assembly needed for configuring the model item；
Parameter information described above, node identification code and model item coding are inserted into node parameter table.
11. the method according to claim 10 for quickly creating model item in machine learning platform, feature exist
In: the model item of the operation creation includes
Model item to be run is selected, and starts moving model project；
It is encoded according to the model item, reads all node identification codes in the model item；
According to set membership of the node identification code in node table, node sequencing is given；
Node is sequentially carried out according to the node sequencing.
12. the method according to claim 11 for quickly creating model item in machine learning platform, feature exist
In, further includes: node is inserted into node execution journal table after executing and executes record；And to model after model item end of run
Project implementation state table is inserted into model item and executes status data.
13. the method according to claim 11 for quickly creating model item in machine learning platform, feature exist
In: it is described to sequentially carry out node according to the node sequencing and include:
For node described in each, according to model item coding and the node identification code from the node parameter table
Obtain the parameter information of the node；
Judge that the node belongs to the data source nodes or the algorithm node；
Data are obtained into the correspondence database of machine learning platform when judging result is the data source nodes；
The parameter of this node is obtained from node parameter table when judging result is the algorithm node and is held according to algorithm parameter
Row algorithm.
14. the method according to claim 1 for quickly creating model item in machine learning platform, feature exist
In the component needed for addition model item into the machine learning running environment specifically:
The component needed for addition same model project into the machine learning running environment；Or
It is described add different model items into the machine learning running environment needed for component.
15. one kind is for quickly creating model terms destination device in machine learning platform characterized by comprising
Creating unit, for creating machine learning running environment；
Component adding unit, for component needed for adding model item into the machine learning running environment；
Connection unit, for required component to be established input and output link according to the sequence that model item is set to create model
Project；
Parameter configuration unit, for carrying out parameter configuration to component needed for the model item；
Running unit, for running the model item of the creation.
16. according to claim 15 for quickly creating model terms destination device in machine learning platform, feature exists
In the creating unit includes:
Panel creating unit, for creating model item panel in the machine learning platform；
Identity assignments unit, for distributing identification code for the model item panel of the creation.
17. according to claim 15 for quickly creating model terms destination device in machine learning platform, feature exists
In: the component adding unit includes:
Assembly unit, for adding component into the model item panel；
Attribute is inserted into unit, for the attribute information of the component of addition to be inserted into model item faceplate formation table；
Wherein, the attribute information includes model item panel identification code where the component of the addition, the addition
The location information of the component Name and the component of the addition in the model item panel.
18. according to claim 15 for quickly creating model terms destination device in machine learning platform, feature exists
In: the connection unit includes:
Model item allocation unit, for being encoded for model item distribution model project name to be run and model item；
Selecting unit, for component needed for selecting the model item；
Link unit, the sequence for being set according to model item established between the component needed for the model item data or
Order input and output link.
19. a kind of electronic equipment, characterized by comprising:
Pocessor and storage media；
The storage medium stores or carries the method for quickly creating model item in machine learning platform；The equipment is logical
After the program of electricity and the method by being used to quickly create model item in machine learning platform described in processor operation, hold
Row operations described below:
Create machine learning running environment；
Component needed for adding model item into the machine learning running environment；
Required component is established into input and output link according to the sequence that model item is set to create model item；
Parameter configuration is carried out to component needed for the model item；
Run the model item of the creation.
20. a kind of method for quickly creating model item in machine learning platform characterized by comprising
Create machine learning running environment；
Component needed for adding model item into the machine learning running environment；
In the academic environment, the component based on addition creates model item；
Run the model item of the creation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710502054.1A CN109146081B (en) | 2017-06-27 | 2017-06-27 | Method and device for creating model project in machine learning platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710502054.1A CN109146081B (en) | 2017-06-27 | 2017-06-27 | Method and device for creating model project in machine learning platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109146081A true CN109146081A (en) | 2019-01-04 |
| CN109146081B CN109146081B (en) | 2022-04-29 |
Family
ID=64805125
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710502054.1A Active CN109146081B (en) | 2017-06-27 | 2017-06-27 | Method and device for creating model project in machine learning platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109146081B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107451663A (en) * | 2017-07-06 | 2017-12-08 | 阿里巴巴集团控股有限公司 | Algorithm assembly, based on algorithm assembly modeling method, device and electronic equipment |
| CN109948804A (en) * | 2019-03-15 | 2019-06-28 | 北京清瞳时代科技有限公司 | Cross-platform towed deep learning modeling and training method and device |
| CN110276456A (en) * | 2019-06-20 | 2019-09-24 | 山东大学 | A kind of machine learning model auxiliary construction method, system, equipment and medium |
| CN110414187A (en) * | 2019-07-03 | 2019-11-05 | 北京百度网讯科技有限公司 | Model safety delivers the system and method for automation |
| CN111367891A (en) * | 2020-03-30 | 2020-07-03 | 中国建设银行股份有限公司 | Method, device and equipment for calling modeling intermediate data and readable storage medium |
| CN111861020A (en) * | 2020-07-27 | 2020-10-30 | 深圳壹账通智能科技有限公司 | Model deployment method, device, equipment and storage medium |
| CN112331348A (en) * | 2020-10-21 | 2021-02-05 | 北京医准智能科技有限公司 | Analysis method and system for set marking, data, project management and non-programming modeling |
| CN113609098A (en) * | 2021-07-31 | 2021-11-05 | 云南电网有限责任公司信息中心 | Visual modeling platform based on data mining process |
| CN114168446A (en) * | 2022-02-10 | 2022-03-11 | 浙江大学 | Simulation evaluation method and device for mobile terminal operation algorithm model |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708404A (en) * | 2012-02-23 | 2012-10-03 | 北京市计算中心 | Machine learning based method for predicating parameters during MPI (message passing interface) optimal operation in multi-core environments |
| US20150220589A1 (en) * | 2014-02-03 | 2015-08-06 | Software Ag | Systems and methods for managing graphical model consistency |
| CN105868019A (en) * | 2016-02-01 | 2016-08-17 | 中国科学院大学 | Automatic optimization method for performance of Spark platform |
| CN106021211A (en) * | 2016-05-18 | 2016-10-12 | 山东达创网络科技股份有限公司 | Intelligent form system and generation method thereof |
| CN106779088A (en) * | 2016-12-06 | 2017-05-31 | 北京物思创想科技有限公司 | Perform the method and system of machine learning flow |
-
2017
- 2017-06-27 CN CN201710502054.1A patent/CN109146081B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708404A (en) * | 2012-02-23 | 2012-10-03 | 北京市计算中心 | Machine learning based method for predicating parameters during MPI (message passing interface) optimal operation in multi-core environments |
| US20150220589A1 (en) * | 2014-02-03 | 2015-08-06 | Software Ag | Systems and methods for managing graphical model consistency |
| CN105868019A (en) * | 2016-02-01 | 2016-08-17 | 中国科学院大学 | Automatic optimization method for performance of Spark platform |
| CN106021211A (en) * | 2016-05-18 | 2016-10-12 | 山东达创网络科技股份有限公司 | Intelligent form system and generation method thereof |
| CN106779088A (en) * | 2016-12-06 | 2017-05-31 | 北京物思创想科技有限公司 | Perform the method and system of machine learning flow |
Non-Patent Citations (1)
| Title |
|---|
| 开发者社区: "阿里云机器学***台", 《HTTPS://DEVELOPER.ALIYUN.COM/ARTICLE/57677》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107451663A (en) * | 2017-07-06 | 2017-12-08 | 阿里巴巴集团控股有限公司 | Algorithm assembly, based on algorithm assembly modeling method, device and electronic equipment |
| CN109948804A (en) * | 2019-03-15 | 2019-06-28 | 北京清瞳时代科技有限公司 | Cross-platform towed deep learning modeling and training method and device |
| CN110276456A (en) * | 2019-06-20 | 2019-09-24 | 山东大学 | A kind of machine learning model auxiliary construction method, system, equipment and medium |
| CN110414187A (en) * | 2019-07-03 | 2019-11-05 | 北京百度网讯科技有限公司 | Model safety delivers the system and method for automation |
| CN110414187B (en) * | 2019-07-03 | 2021-09-17 | 北京百度网讯科技有限公司 | System and method for model safety delivery automation |
| CN111367891A (en) * | 2020-03-30 | 2020-07-03 | 中国建设银行股份有限公司 | Method, device and equipment for calling modeling intermediate data and readable storage medium |
| CN111861020A (en) * | 2020-07-27 | 2020-10-30 | 深圳壹账通智能科技有限公司 | Model deployment method, device, equipment and storage medium |
| CN112331348A (en) * | 2020-10-21 | 2021-02-05 | 北京医准智能科技有限公司 | Analysis method and system for set marking, data, project management and non-programming modeling |
| CN112331348B (en) * | 2020-10-21 | 2021-06-25 | 北京医准智能科技有限公司 | Analysis method and system for set marking, data, project management and non-programming modeling |
| CN113609098A (en) * | 2021-07-31 | 2021-11-05 | 云南电网有限责任公司信息中心 | Visual modeling platform based on data mining process |
| CN114168446A (en) * | 2022-02-10 | 2022-03-11 | 浙江大学 | Simulation evaluation method and device for mobile terminal operation algorithm model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109146081B (en) | 2022-04-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109146081A (en) | It is a kind of for quickly creating the method and device of model item in machine learning platform | |
| CN109857668A (en) | UI automated function test method, test device, test equipment and storage medium | |
| US9858544B2 (en) | Expert exchange during UML design updates | |
| CN106156186A (en) | A kind of data model managing device, server and data processing method | |
| CN109086071A (en) | A kind of method and server of management software version information | |
| CN110287097A (en) | Batch testing method, device and computer readable storage medium | |
| Hegedüs et al. | Back-annotation of simulation traces with change-driven model transformations | |
| Mitsyuk et al. | Generating event logs for high-level process models | |
| JP6045134B2 (en) | Parallel workload simulation for application performance testing | |
| CN108345532A (en) | A kind of automatic test cases generation method and device | |
| CN101946261A (en) | Automated model generation for computer based business process | |
| CN105094818A (en) | Construction method and system for comprehensive application of natural resources based on SOA | |
| KR100672894B1 (en) | Apparatus and method for product-line architecture description and verification | |
| Campagna et al. | Product and production process modeling and configuration | |
| CN115099793B (en) | Dynamic model assembling method, server and storage medium for task scene | |
| CN105446952A (en) | Method and system for processing semantic fragments | |
| Arrotéia et al. | BIM critical success factors in different lifecycle phases: a review exploring the technology, people, and process BIM categories | |
| CN102053825A (en) | Method and system for processing software design conflicts | |
| CN111897725A (en) | Automatic test method, medium, equipment and system for middle platform service | |
| CN102594851A (en) | Method for dynamically building marine application service chain | |
| Jacob et al. | An Evolutionary Software Paradigm for Developing Internet of Things (IoT) Based Systems | |
| CN102591779B (en) | Establishing method for workflow-based universal software testing process model | |
| CN107122849B (en) | Spark R-based product detection total completion time minimization method | |
| CN109582347A (en) | A kind of method and device obtaining front-end code | |
| CN104267967A (en) | Method and device for generating program codes of software |
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 |