WO2022142015A1 - 流程迁移方法、装置、设备及介质 - Google Patents

流程迁移方法、装置、设备及介质 Download PDF

Info

Publication number
WO2022142015A1
WO2022142015A1 PCT/CN2021/090527 CN2021090527W WO2022142015A1 WO 2022142015 A1 WO2022142015 A1 WO 2022142015A1 CN 2021090527 W CN2021090527 W CN 2021090527W WO 2022142015 A1 WO2022142015 A1 WO 2022142015A1
Authority
WO
WIPO (PCT)
Prior art keywords
domain
text
recognition
node
model
Prior art date
Application number
PCT/CN2021/090527
Other languages
English (en)
French (fr)
Inventor
蔡辉
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2022142015A1 publication Critical patent/WO2022142015A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
    • G06F16/275—Synchronous replication

Definitions

  • the present application relates to the technical field of data synchronization of big data, and in particular, to a process migration method, apparatus, device and medium.
  • the present application provides a process migration method, apparatus, computer equipment and storage medium, which improve the efficiency and quality of process migration.
  • a process migration method including:
  • the process set includes multiple processes, each of which includes process document information, process nodes and process node relationship;
  • the process node and the process node relationship in the process to be migrated with the same domain object by using the domain design model to obtain an aggregation result corresponding to the domain object;
  • the aggregation result includes the entity corresponding to the process node, the value object associated with the entity, and the aggregate root;
  • the entity corresponding to the process node in the process to be migrated the entity data table, the value object and the aggregate root associated with the entity, in the target
  • the system reconstructs each target process corresponding to each process to be migrated, reconstructs all the target processes, and determines that the system process migration request is completed.
  • a process migration device comprising:
  • the receiving module is used to respond to the system process migration request, and obtain the process set of the target system, the source system and the domain migration list in the system process migration request;
  • the process set includes a plurality of processes, and each of the processes includes a process Document information, process nodes and process node relationships;
  • the extraction module is used to input all the processes into the domain design model, extract the domain semantic features of the process document information and the process nodes in each of the processes through the domain design model, and identify and The domain object corresponding to the process described above;
  • a matching module configured to filter out the process corresponding to the domain object matching the domain to be migrated in the domain migration list from the process set, and determine the filtered process as the process to be migrated;
  • an aggregation module configured to aggregate the process node and the process node relationship in the process to be migrated with the same domain object through the domain design model to obtain an aggregation result corresponding to the domain object;
  • the aggregation The result includes an entity corresponding to the process node, a value object associated with the entity, and an aggregate root;
  • a creation module configured to create an entity data table associated with each entity in the target system according to the entity in each of the aggregation results and the value object associated with the entity;
  • Reconstruction module configured to calculate according to the process to be migrated, the entity corresponding to the process node in the process to be migrated, and the entity data table, the value object and the aggregate associated with the entity At the root, each target process corresponding to each process to be migrated is reconstructed in the target system, all the target processes are reconstructed, and it is determined that the system process migration request is completed.
  • a computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, the processor implementing the following steps when executing the computer-readable instructions:
  • the process set includes multiple processes, each of which includes process document information, process nodes and process node relationship;
  • the process node and the process node relationship in the process to be migrated with the same domain object by using the domain design model to obtain an aggregation result corresponding to the domain object;
  • the aggregation result includes the entity corresponding to the process node, the value object associated with the entity, and the aggregate root;
  • the entity corresponding to the process node in the process to be migrated the entity data table, the value object and the aggregate root associated with the entity, in the target
  • the system reconstructs each target process corresponding to each process to be migrated, reconstructs all the target processes, and determines that the system process migration request is completed.
  • One or more readable storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the following steps:
  • the process set includes multiple processes, each of which includes process document information, process nodes and process node relationship;
  • the process node and the process node relationship in the process to be migrated with the same domain object by using the domain design model to obtain an aggregation result corresponding to the domain object;
  • the aggregation result includes the entity corresponding to the process node, the value object associated with the entity, and the aggregate root;
  • the entity corresponding to the process node in the process to be migrated the entity data table, the value object and the aggregate root associated with the entity, in the target
  • the system reconstructs each target process corresponding to each process to be migrated, reconstructs all the target processes, and determines that the system process migration request is completed.
  • the process migration method, device, computer equipment and storage medium based on the domain design model provided by the present application saves the storage space of the process and improves the efficiency and quality of the process migration.
  • FIG. 1 is a schematic diagram of an application environment of a process migration method in an embodiment of the present application
  • step S20 of the process migration method in an embodiment of the present application is a flowchart of step S20 of the process migration method in an embodiment of the present application
  • step S40 of the process migration method in an embodiment of the present application is a flowchart of step S40 of the process migration method in an embodiment of the present application.
  • FIG. 5 is a flowchart of step S50 of the process migration method in an embodiment of the present application.
  • FIG. 6 is a schematic block diagram of a conversion module of a process migration device in an embodiment of the present application.
  • FIG. 7 is a schematic diagram of a computer device in an embodiment of the present application.
  • the process migration method provided by the present application can be applied in the application environment as shown in FIG. 1 , wherein the client (computer device) communicates with the server through the network.
  • the client computer equipment
  • the server includes but is not limited to various personal computers, notebook computers, smart phones, tablet computers, cameras and portable wearable devices.
  • the server can be implemented as an independent server or a server cluster composed of multiple servers.
  • the migration method is executed by the client or the server.
  • a process migration method is provided, and its technical solution mainly includes the following steps S10-S60:
  • the process set includes multiple processes, each of which includes process document information, Process nodes and process node relationships.
  • the target system is a constructed system platform for migration
  • the target system is The platform system that has been developed using the language to be unified
  • the source system contains the process set
  • the source system is the system that needs to be migrated
  • the source system has original processes developed using different development languages
  • the process set includes all processes operating in the source system, and each process includes process document information, process nodes, and process node relationships.
  • the process document information is information related to the process, and the process node is the job node on the process of the process, the process node relationship is the association relationship between process nodes, the domain migration list is a collection of domains that need to be migrated, and the domain migration list lists the to-be-migrated domains that need to be migrated .
  • the process set of the target system, the source system and the domain migration list in the system process migration request are obtained.
  • the original process is an existing process in the source system.
  • S102 Perform language category recognition on each of the original processes through a flowchart construction model, and identify a language type corresponding to each of the original processes.
  • the flow chart construction model is a model capable of identifying the language category for developing the original flow.
  • the language category includes C++ language, ASP language, Delphi language, Java language, etc., the process of identifying the language category.
  • the process of identifying the language category In order to obtain the input interface of the original process, by inputting request instructions of different language categories to the input interface, the language category corresponding to the request instruction for which the response is detected is marked as the language type corresponding to the original process
  • the process of identifying the language category also includes the process of acquiring the main program code in each of the original processes, and identifying by extracting the language format feature of the main program code, and the language format feature is a variety of development
  • the features of the writing format of the language are introduced into the neural network model to extract and identify the features of the language format.
  • the main program code is the program code that starts to execute the original process.
  • the flowchart construction model includes a plurality of the language types and the interpretation scripts corresponding to each of the language types one-to-one, and the interpretation scripts are generated according to the source program code developed according to the corresponding language type.
  • the script tool of the flowchart corresponding to the source code, the interpretation process is to analyze all the source code in the original process, analyze all the process nodes in the original process, identify each of the The association relationship between process nodes (that is, the reference relationship between variables), the process of associating and collapsing each of the process nodes in the original process, and summarizing all the process nodes associated and collapsed to generate a flowchart
  • the source program code includes the main program code.
  • the present application realizes the acquisition of all the original processes of the source system; the language category identification is performed on each of the original processes through a flowchart construction model, and the language type corresponding to each of the original processes is identified; the flow chart construction model is used to obtain Interpretation script corresponding to the language type corresponding to the original process, and interpret the original process through the acquired interpretation script corresponding to the original process, and construct a corresponding to the original process according to the flowchart generation method.
  • the process in this way, realizes the language category identification through the flow chart construction model, identifies the development language described in each original process, obtains the interpretation script corresponding to the language type corresponding to each original process, and interprets each original process.
  • S20 Input all the processes into a domain design model, perform domain semantic feature extraction on the process document information and the process nodes in each of the processes by using the domain design model, and identify the corresponding process to each of the processes. domain object.
  • the domain semantic feature extraction is performed on the process document information and the process nodes in each of the processes by using the domain design model, where the domain semantic features are features related to domain objects, and domain semantics.
  • the process of identifying the contextual semantic features and scene features between objects is to extract the contextual semantic features and scene features between domain objects on the process document information and the process nodes through the domain design model.
  • Context semantic recognition is performed on the contextual semantic features between objects
  • contextual scene recognition is performed on the extracted contextual scene features
  • a first recognition result for contextual semantic recognition and a second recognition result for contextual scene recognition are identified.
  • the domain boundary location is performed to determine the identification process of the domain object corresponding to the process.
  • the process document information includes process requirement document and process scenario information
  • the process requirement document is a document of the requirement class input for developing the process, that is, the process requirement document includes the interface design of the process node, the execution of the process node
  • the process scene information includes the scene of the process application, the scene object that the process is oriented to, and the purpose or goal of the process, and the process scene information is the process application.
  • the process node includes a process interface and a node object;
  • the domain design model is a model trained based on a domain-driven design (DDD) algorithm and used to identify the domain of the process and perform process migration,
  • DDD domain-driven design
  • the concept of domain-driven design is to use a unified common language as a tool for mutual training between domains, discover and dig out the main domain feature objects in the process of continuous training, and then design these domain feature objects into a domain model. , build the domain model with code, the domain object represents the domain of the corresponding process, and the type of the domain object is a predefined domain name.
  • step S20 that is, the domain semantic feature extraction is performed on the process document information and the process nodes in each of the processes by using the domain design model, Domain objects corresponding to each of the described processes are identified, including:
  • the splicing method can be set according to requirements.
  • the process interface is spliced before and after the process requirement document, and the spliced process requirement document is subjected to embedding conversion to obtain the The first text object, wherein, the embedding is converted to divide the input text content into a single word or word, and each word or word is mapped and converted into a word vector corresponding to it, and the mapping relationship is obtained by training the embedding model, Combine all the converted word vectors with feature vectors to obtain a string of feature vector values, and determine it as the first text object, and insert the node object after each sentence in the process scene information, and combine to generate all the describe the second text object.
  • the process document information includes process requirement document and process scene information
  • the process node includes process interface and node object
  • the process interface is the interface used in the process node, for example: the process interface is to obtain the relevant information of the approver
  • the interface of the information database the node object is a process node-oriented object, that is, the process node contains a collection of input objects, for example, the node object includes the input amount, time, and text information of each input field in the process node.
  • the process requirement document in the process is spliced with the process interface to obtain a first text object corresponding to the process, and the process is The process scene information in and the node object are combined to obtain a second text object corresponding to the process, including:
  • the process of splicing before and after is to insert the content of the process interface before the process requirement document, insert the content of the process interface after the process requirement document, and insert the inserted process requirement.
  • the document is determined to be the process of splicing, and the embedding is converted to dividing the input text content into a single word or word, and mapping each word or word into a word vector corresponding to it.
  • the mapping relationship is obtained by training the embedding model. , and combining all the converted word vectors with feature vectors to obtain a series of feature vector values, which are determined as the first text object.
  • the merging process is to insert the node object after each punctuation mark in the process scene information, and then insert a separator after the node object, so as to realize that after each sentence.
  • the operation process of inserting the node object is combined to generate the second text object.
  • the present application realizes that by splicing the process interface before and after the process requirement document, and performing embedding conversion, the first text object corresponding to the process is obtained; each sentence in the process scene information The node object is then inserted and merged into the second text object corresponding to the process.
  • the method of combining the process interface and the process requirement document is realized, providing a data basis for subsequent context semantic recognition, and combining
  • the method of process scene information and node objects provides a data basis for subsequent contextual scene recognition, improves the accuracy and quality of recognition, and finally improves the accuracy of subsequent field object recognition.
  • S202 perform context semantic recognition on the first text object by using the first text recognition model in the domain design model, and recognize the first recognition result, and at the same time use the second text recognition model in the domain design model to identify all the text objects.
  • the second text object is used for long context scene recognition, and the second recognition result is recognized.
  • the first text recognition model realizes that by identifying the contextual semantic features in the first text object, and according to the extracted contextual semantic features, the node category results are identified, and the identified nodes are identified.
  • the classification result is used to identify the business field, and the model of the first recognition result is obtained.
  • the first text recognition model is the language model that has been trained.
  • the context semantic feature is the semantic aspect of the context association between two sentences and is related to the node Category-related features
  • the node category result is the weight distribution of the identified business categories
  • the node category is the business category
  • the process node is added to the first text object to allow the The process node is combined in the context of the process requirement document, so that the recognition accuracy of the first recognition result of the first text recognition model is improved, the node category can be more easily recognized, and the recognition efficiency of the first recognition result is improved.
  • the second text recognition model realizes that by identifying the contextual scene features in the second text object, and according to the extracted contextual scene features, the scene category results are identified, and the identified scene The classification result is used for scene domain recognition, and a model of the second recognition result is obtained.
  • the second text recognition model is a language model that has been trained, and the contextual scene feature is a contextual association between two sentences.
  • the scene category is the category of the preset applied scene
  • the scene category result includes at least one scene category and its corresponding prediction probability, that is, the probability distribution of the identified categories of each scene, in the second
  • the node object is added to the text object, the purpose is to combine the node object in the context of the process scene information, so that the recognition accuracy of the second recognition result of the second text recognition model is improved, the scene category can be more easily recognized, and the The recognition efficiency of the second recognition result is improved.
  • the domain design model includes the first text recognition model and the second text recognition model
  • the first recognition result is the result of determining the probability distribution of each domain from the business domain dimension according to the node category result.
  • the second identification result is the result of determining the probability distribution of each field from the scene field dimension according to the scene category result.
  • step S202 that is, performing context semantic recognition on the first text object through the first text recognition model in the domain design model, and recognizing the first recognition result, including:
  • the first text recognition model is a Bi-LSTM-based language model.
  • the first text recognition model is a model constructed on the basis of the Bi-LSTM network structure
  • the Bi-LSTM network structure model includes a forward LSTM model and a backward LSTM model.
  • the LSTM model captures bidirectional contextual semantics, and performs bidirectional contextual semantic convolution on the first text object through the first text recognition model, and extracts the semantic features of the context, which are two sentences. Semantic aspects of the contextual association between and node category-related features.
  • the extracted contextual semantic features are arranged by the first text recognition model, output into a service feature vector, and a one-dimensional service is output through the fully connected layer in the first text recognition model.
  • the fully-connected feature vector is weighted and multiplied by the corresponding vector in the service fully-connected feature vector by the weight corresponding to each of the node categories, so as to identify the node category result.
  • the node category result includes at least one node category and its corresponding weight, that is, the node category result is the weight distribution of the identified categories of each service, and the node category result indicates that predictions in the field of each service are predicted.
  • the weights range from 0 to 1.
  • S2023 Perform business domain recognition on the node category result by using the first text recognition model to obtain the first recognition result corresponding to the first text object.
  • the identification of the business domain is to identify the domain of the business dimension to which the process node belongs, that is, to predict the probability distribution of the domain of each business dimension through the weight distribution in the node category result, which can also be understood as: The probability distribution obtained by aggregating the node categories in the field of the same business dimension in the node category result.
  • the present application realizes that the context semantic feature in the first text object is extracted by the first text recognition model; the first text recognition model is a language model based on Bi-LSTM;
  • the extracted context semantic feature identifies the node category result corresponding to the first text object; performs business domain identification on the node category result through the first text recognition model, and obtains a result corresponding to the first text object.
  • the corresponding first recognition result in this way, realizes that the context semantic feature in the first text object is extracted by the first text recognition model based on the Bi-LSTM language model, and the business domain recognition method is used to automatically recognize the first text.
  • the first recognition result of the object makes it easier to recognize the node category, which improves the recognition efficiency of the first recognition result, reduces manual recognition, and improves the recognition accuracy and efficiency for the follow-up.
  • step S202 that is, the long context scene recognition is performed on the second text object through the second text recognition model in the domain design model, and the second recognition result is recognized, including:
  • the second text recognition model is a Bert-based recognition model.
  • the second text recognition model is a model constructed on the basis of Bert's network structure, and the model of the Bert network structure is trained by jointly adjusting the bidirectional Transformer algorithm in all layers, and the third A model in which the text in the second text object is converted into a word sense vector and labeled, and the second text object is extracted by the second text recognition model.
  • the context scene feature, the context scene feature is the difference between two sentences.
  • Context-sensitive features related to scene categories. S2025 perform context relevance recognition according to the extracted scene feature by the second text recognition model, and recognize a scene category result corresponding to the second text object.
  • the extracted context scene features are identified by the second text recognition model, and the correlation analysis is performed on the text related to the scene category that is identified as the context, and a scene feature vector is analyzed,
  • the analysis process is a process of analyzing the association relationship between the context spans or scene dimensions, and outputs a one-dimensional scene fully-connected feature vector through the fully-connected layer in the second text recognition model. output the node category result.
  • the scene category result includes at least one scene category and its corresponding prediction probability, that is, the scene category result is the probability distribution of the identified categories of each scene, and the scene category result indicates the prediction in the field of each scene.
  • the probability value is in the range of 0% to 100%.
  • S2026 Perform scene domain recognition on the scene category result by using the second text recognition model to obtain the second recognition result corresponding to the second text object.
  • the identification of the scene domain is to identify the domain of the scene dimension to which the process node belongs, that is, to predict the probability distribution of the domain of each scene dimension through the probability distribution in the scene category result, which can also be understood as: The probability distribution obtained by aggregating the scene categories in the same scene dimension field in the scene category result.
  • the present application implements scene feature extraction for the second text object through the second text recognition model;
  • the second text recognition model is a Bert-based recognition model;
  • the scene feature is used to identify the contextual relevance, and the scene category result corresponding to the second text object is identified; the scene category result is identified by the second text recognition model.
  • the second recognition result corresponding to the object so that the context scene feature in the second text object is extracted by the second text recognition model based on the Bert recognition model, and the second text object is automatically recognized by using the scene field recognition method.
  • the second identification result of reduces manual identification, and identifies process nodes through the scene dimension, which improves the accuracy and efficiency of subsequent identification.
  • S203 Perform domain boundary positioning on the first identification result and the second identification result by using the domain design model, and determine the domain object corresponding to the process.
  • the domain boundary is positioned to convert the first identification result and the second identification result into a numerical range of one dimension, and then locate the business center corresponding to the first identification result in the domain polygon. point, and the scene center point corresponding to the second recognition result, comprehensively locate the field object according to the distance between the business center point and the scene center point from the boundary of each field object, for example: the value range is 1 In the range from 10 to 10, all domain objects form a polygon with the same points as the total number of all the domain objects, each point represents a 10-point value of the domain object, and the center of the polygon is the 0-point value of each domain object, That is, the value of the first identification result is the field of the business corresponding to 1, the median value of the second identification result is the field of the scene corresponding to 100%, and one of the field objects corresponds to the field of a business and the field of a scene.
  • the distribution of the recognition results Locate the corresponding points in the fields of each service in the formed polygon, then connect them into a service polygon, and take the center of the service polygon as the service center point of the first recognition result, According to the distribution of the second recognition result, the corresponding points in the field of each scene are located in the formed polygons, and then they are connected to form a scene-like polygon, and the center of the scene-like polygon is taken as the center of the second recognition result.
  • the scene center point according to the distance between the business center point and the scene center point and the respective boundaries of the domain objects, that is, the business center point and the scene center point fall within the range of the boundaries of the respective domain objects, and then
  • the final domain object is determined according to the overlapping ratio of the scopes, which is the domain object corresponding to the process.
  • the present application realizes that by splicing the process requirement document and the process interface in the process, a first text object corresponding to the process is obtained, and at the same time, the process scene information in the process is combined with the process interface.
  • the node objects are merged to obtain a second text object corresponding to the process; the context semantic recognition is performed on the first text object by the first text recognition model in the domain design model, and the first recognition result is recognized, and the The second text recognition model in the domain design model performs long context scene recognition on the second text object, and recognizes a second recognition result; the first recognition result and the second recognition result are identified by the domain design model.
  • the recognition result is used to locate the domain boundary, and the domain object corresponding to the process is determined.
  • the domain boundary positioning of the system determines the domain of the process, and realizes the automatic identification of the domain of the process for subsequent migration of the process in the domain that needs to be migrated.
  • S30 Screen out the flow corresponding to the domain object matching the domain to be migrated in the domain migration list from the flow set, and determine the screened flow as the flow to be migrated.
  • the domain object matching the to-be-relocated domain is filtered out from the process set, and then the process corresponding to the matched domain object is found, and the found process is determined. is the process to be migrated, and the process to be migrated is a process that needs to be migrated.
  • the domain migration list includes a plurality of preset domains to be migrated.
  • S40 Aggregate the process node and the process node relationship in the process to be migrated with the same domain object by using the domain design model to obtain an aggregation result corresponding to the domain object; the aggregation result includes a The entity corresponding to the process node, the value object associated with the entity, and the aggregate root.
  • the process node and the process node relationship in the process to be migrated with the same domain object are aggregated through the domain design model, and a process node map is constructed.
  • the same described process nodes are collected into one process node, and this process node includes the process of the process node relationship associated with the original process node, and the construction process is to perform all the collected process nodes according to all the process node relationships.
  • the Conway's Law algorithm is to combine similar process nodes and process node relationships to generate a new process node, the new process node can include the function of the original process node and the relationship of the original process node, the new process The node is determined as an entity, the aggregated process node relationship is determined as the aggregation root, and the process node contained in the entity is determined as an algorithm of value objects.
  • step S40 that is, the relationship between the process node and the process node in the process to be migrated with the same domain object is determined by the domain design model. Perform aggregation to get the aggregation results corresponding to the objects in this domain, including:
  • the aggregation is to separate the process nodes in the to-be-migrated process of the same domain object from all the process nodes, and perform the relationship between the separated two process nodes. Convergence, in this way, constructs the process node graph of the domain object, thereby constructing the process node graph of all the domain objects.
  • the Conway's Law algorithm is to combine similar process nodes and process node relationships under the same process node graph to generate a new process node, and the new process node can include the functions of the original process node. and the original process node relationship, the new process node is determined as an entity, the aggregated process node relationship is determined as the aggregation root, the process node contained in the entity is determined as the value object algorithm, and the aggregation result includes and the process node.
  • the present application realizes the aggregation of the process nodes and the process node relationships in the process to be migrated with the same domain object through the domain design model, and constructs a process node map; using the Conway's law algorithm, through The domain design model aggregates the process node graph to obtain the aggregation results corresponding to the domain objects.
  • the domain design model aggregates the processes of the same domain objects, constructs the process node graph, and uses the The Law of Power calculates, aggregates the process node graph to output the aggregation results, reduces the cost of manual aggregation, and automatically aggregates each process node and process node relationship into entities, value objects and aggregation roots, saving the capacity of the target process and greatly compressing It reduces the footprint of the process and unifies the build platform and development language.
  • the entity includes an entity name, an entity storage data format and an entity interface
  • the entity data table conforming to the entity is created in the target system, and the table name of the data table is associated with the entity name , the fields in the data table correspond to the value objects associated with the entity, the database structure stored in the entity data table is the same as the data format stored in the entity, and the entity data table can be accessed through the entity interface to communicate.
  • the target system reconstructs each target process corresponding to each of the to-be-migrated processes, reconstructs all the target processes, and determines that the system process migration request is completed.
  • each of the process nodes in the process to be migrated is replaced by the entity corresponding to it, and the entity data table corresponding to the entity is associated, and the process node relationship is replaced by the value object.
  • Perform scene substitution with the aggregate root that is, different value objects will correspond to different relationships of the aggregate root.
  • the target process corresponding to the process to be migrated is reconstructed, and the target process is realized
  • the process function of the target process is the same as that of the process to be migrated, and all the target processes are unified in one development language, and the processes in different languages are unified, which is convenient for subsequent operation and maintenance.
  • the present application realizes that by receiving a system process migration request, the process set of the target system, the source system and the domain migration list in the system process migration request are obtained; all the processes are input into the domain design model, and the domain design
  • the model performs domain semantic feature extraction on the process document information and the process nodes in each of the processes, and identifies the domain objects corresponding to each of the processes; filters out the domain migration list from the process set For the process corresponding to the domain object that matches the domain to be migrated in, the screened out process is determined as the process to be migrated; the domain design model compares all the processes to be migrated with the same domain object in the process to be migrated.
  • the entity data table, the value object, and the aggregate root reconstruct each target process corresponding to each process to be migrated in the target system, and reconstruct all the target processes to determine the system process migration request Finish.
  • the process collection and domain migration list of the target system, the source system and the domain migration list in the system process migration request are obtained; the process document information and the process nodes in each of the processes are analyzed by the domain design model.
  • the domain of each process to be migrated is identified through domain semantic feature extraction and recognition, and then the domain design model is used to aggregate the processes in the same domain.
  • the domain design model that drives the design algorithm migrates the process of the domain that needs to be migrated to the target system, without manual screening and development and construction, directly aggregates to generate entities, value objects and aggregate roots, unifies all processes into one development language, and creates a unified
  • the entity data table and the reconstruction of the target process greatly save the storage space of the process, achieve the effect of accurately and reliably migrating the process to the target system, and improve the efficiency and quality of the process migration.
  • a process migration apparatus is provided, and the process migration apparatus corresponds one-to-one with the process migration method in the above-mentioned embodiment.
  • the process migration apparatus includes a receiving module 11 , an extraction module 12 , a matching module 13 , an aggregation module 14 , a creation module 15 and a reconstruction module 16 .
  • the detailed description of each functional module is as follows:
  • the receiving module 11 is configured to respond to the system process migration request, and obtain the process set of the target system, the source system and the domain migration list in the system process migration request; the process set includes multiple processes, and each of the processes includes Process document information, process nodes and process node relationships;
  • the extraction module 12 is configured to input all the processes into the domain design model, perform domain semantic feature extraction on the process document information and the process nodes in each of the processes through the domain design model, and identify and the domain object corresponding to the process;
  • a matching module 13 configured to filter out the process corresponding to the domain object matching the domain to be migrated in the domain migration list from the process set, and determine the filtered process as the process to be migrated;
  • the aggregation module 14 is configured to aggregate the process node and the process node relationship in the process to be migrated with the same domain object through the domain design model to obtain an aggregation result corresponding to the domain object; the The aggregation result includes an entity corresponding to the process node, a value object associated with the entity, and an aggregation root;
  • a creation module 15 configured to create an entity data table associated with each entity in the target system according to the entity in each of the aggregation results and the value object associated with the entity;
  • the reconstruction module 16 is configured to, according to the process to be migrated, the entity corresponding to the process node in the process to be migrated, and the entity data table, the value object, and the entity associated with the entity Aggregate root, reconstruct each target process corresponding to each process to be migrated in the target system, reconstruct all the target processes, and determine that the system process migration request is completed.
  • Each module in the above-mentioned process migration apparatus may be implemented in whole or in part by software, hardware and combinations thereof.
  • the above-mentioned modules can be embedded in or independent of the processor in the computer equipment in the form of hardware, and can also be stored in the memory in the computer equipment in the form of software, so that the processor calls and executes the corresponding operations of the above-mentioned various modules.
  • a computer device in one embodiment, can be a server, and its internal structure diagram can be as shown in FIG. 7 .
  • the computer device includes a processor, memory, a network interface, and a database connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities.
  • the memory of the computer device includes a readable storage medium, an internal memory.
  • the readable storage medium stores an operating system, computer readable instructions and a database.
  • the internal memory provides an environment for the execution of the operating system and computer-readable instructions in the readable storage medium.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer readable instructions when executed by a processor, implement a process migration method.
  • the readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.
  • one or more readable storage media storing computer-readable instructions are provided, and the readable storage media provided in this embodiment include non-volatile readable storage media and volatile readable storage media medium; computer-readable instructions are stored on the readable storage medium, and when the computer-readable instructions are executed by one or more processors, cause one or more processors to implement the process migration method in the foregoing embodiment.
  • a computer-readable storage medium on which computer-readable instructions are stored, and when the computer-readable instructions are executed by a processor, the process migration method in the foregoing embodiment is implemented.
  • Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Machine Translation (AREA)

Abstract

本申请涉及数据同步技术领域，本申请公开了一种流程迁移方法、装置、设备及介质，所述方法包括：通过获取系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；通过领域设计模型对各流程中的流程文档信息和流程节点进行领域语义特征提取，识别出与各个流程对应的领域对象；从流程集合中筛选出与领域迁移清单中的待迁领域匹配的领域对象对应的待迁移流程；将相同领域对象的待迁移流程进行聚合，得到聚合结果；在目标系统创建实体数据表；根据待迁移流程、所有实体以及实体数据表、值对象和聚合根，重构各目标流程。本申请实现了准确和可靠的将流程迁移至目标系统的效果，提高了流程迁移的效率和质量。

Description

流程迁移方法、装置、设备及介质
本申请要求于2020年12月30日提交中国专利局、申请号为202011620085.5，发明名称为“流程迁移方法、装置、设备及介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及大数据的数据同步技术领域，尤其涉及一种流程迁移方法、装置、设备及介质。
背景技术
发明人发现随着互联网、云计算和物联网等信息技术的迅猛发展，相应带来的流程数据量也是呈现爆发式的增长，由于系统的硬件局限或者平台局限，必然会存在流程迁移的情况，在迁移过程中，首先，需要从众多的流程中筛选出需要迁移的范畴或者领域的流程，现有技术中，大多数是人工筛选，需要投入大量的工作人员的工作量；其次，对筛选后的流程进行复制迁移，由于不同的流程会存在不同开发语言，就需要人工转换成一种语言，开发时间长；最后，复制迁移后的流程仍然很庞大，流程之间错综复杂，对后续的维护仍然存在较大的复杂度，不便于维护。因此，现有的流程迁移的方案存在人工投入工作量大，开发时间长，以及不便维护的不足，严重影响流程迁移的效率、准确性和可维护性。
发明内容
本申请提供一种流程迁移方法、装置、计算机设备及存储介质，，提高了流程迁移的效率和质量。
一种流程迁移方法，包括：
响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
一种流程迁移装置，包括：
接收模块，用于响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程 文档信息、流程节点和流程节点关系；
提取模块，用于将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
匹配模块，用于从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
聚合模块，用于通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
创建模块，用于根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
重构模块，用于根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
一种计算机设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机可读指令，所述处理器执行所述计算机可读指令时实现如下步骤：
响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
一个或多个存储有计算机可读指令的可读存储介质，所述计算机可读指令被一个或多个处理器执行时，使得所述一个或多个处理器执行如下步骤：
响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述 流程节点对应的实体、与所述实体关联的值对象和聚合根；
根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
本申请提供的基于领域设计模型的流程迁移方法、装置、计算机设备及存储介质，节省了流程的存储空间，提高了流程迁移的效率和质量。
附图说明
为了更清楚地说明本申请实施例的技术方案，下面将对本申请实施例的描述中所需要使用的附图作简单地介绍。
图1是本申请一实施例中流程迁移方法的应用环境示意图；
图2是本申请一实施例中流程迁移方法的流程图；
图3是本申请一实施例中流程迁移方法的步骤S20的流程图；
图4是本申请一实施例中流程迁移方法的步骤S40的流程图；
图5是本申请一实施例中流程迁移方法的步骤S50的流程图；
图6是本申请一实施例中流程迁移装置的转换模块的原理框图；
图7是本申请一实施例中计算机设备的示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请提供的流程迁移方法，可应用在如图1的应用环境中，其中，客户端(计算机设备)通过网络与服务器进行通信。其中，客户端(计算机设备)包括但不限于为各种个人计算机、笔记本电脑、智能手机、平板电脑、摄像头和便携式可穿戴设备。服务器可以用独立的服务器或者是多个服务器组成的服务器集群来实现。
所述迁移方法由客户端或者服务端执行。
在一实施例中，如图2所示，提供一种流程迁移方法，其技术方案主要包括以下步骤S10-S60：
S10，响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系。
可理解地，通过所述系统流程迁移请求，获取到所述目标系统、所述源系统和所述领域迁移清单，所述目标系统为构建好的用于迁移后的系统平台，所述目标系统已经运用即将统一的语言开发的平台系统，所述源系统包含有所述流程集合，所述源系统为需要被迁移流程的系统，所述源系统中存在使用了不同开发语言开发的原始流程，所述流程集合包括了所有在所述源系统运作的流程，每个所述流程包括流程文档信息、流程节点和流程节点关系，所述流程文档信息为与该流程相关的信息，所述流程节点为该流程的流程上的作业节点，所述流程节点关系为流程节点之间的关联关系，所述领域迁移清单为需要迁移的领域的集合，所述领域迁移清单罗列了需要迁移的待迁领域。
在一实施例中，如图3所示，所述步骤S10之前，即所述接收到系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单之前，包括：
S101，获取所述源系统的所有原始流程。
可理解地，所述原始流程为所述源系统中已有的流程。
S102，通过流程图构建模型对各所述原始流程进行语言类别识别，识别出与各所述原始流程对应的语言类型。
可理解地，所述流程图构建模型为能够识别出开发所述原始流程的语言类别的模型，比如语言类别包括C++语言、ASP语言、Delphi语言、Java语言等等，所述语言类别识别的过程为通过获取所述原始流程的输入接口，通过向所述输入接口输入不同的语言类别的请求指令，将与检测到响应的请求指令对应的语言类别标记为与该原始流程对应的所述语言类型的识别过程，所述语言类别识别的过程还包括获取各所述原始流程中的主程序代码，通过提取所述主程序代码的语言格式特征进行识别的过程，所述语言格式特征为各种开发语言的编写格式的特征，引入神经网络模型进行所述语言格式特征提取并识别。
其中，所述主程序代码为开始执行该原始流程的程序代码。
S103，通过流程图构建模型获取与所述原始流程对应的语言类型对应的解读脚本，并通过获取的与所述原始流程对应的所述解读脚本对所述原始流程进行解读，并按照流程图生成方式构建与该原始流程对应的所述流程。
可理解地，所述流程图构建模型包括多种所述语言类型和与各个所述语言类型一一对应的所述解读脚本，所述解读脚本为根据与其对应的语言类型开发的源程序代码生成与该源程序代码对应的流程图的脚本工具，所述解读过程为对所述原始流程中的所有源程序代码进行分析，分析出所述原始流程中的所有所述流程节点，识别各个所述流程节点之间的关联关系(即变量之间的引用关系)，将所述原始流程中的各个所述流程节点进行关联及合拢，汇总关联及合拢的所有所述流程节点生成流程图的过程，所述源程序代码包括所述主程序代码。
本申请实现了通过获取所述源系统的所有原始流程；通过流程图构建模型对各所述原始流程进行语言类别识别，识别出与各所述原始流程对应的语言类型；通过流程图构建模型获取与所述原始流程对应的语言类型对应的解读脚本，并通过获取的与所述原始流程对应的所述解读脚本对所述原始流程进行解读，并按照流程图生成方式构建与该原始流程对应的所述流程，如此，实现了通过流程图构建模型进行语言类别识别，识别出各个原始流程所述的开发语言，获取与各原始流程对应的语言类型对应的解读脚本，以及对各原始流程进行解读，构建出与原始流程一一对应的流程，达到了针对不同开发语言的流程进行解读并构建成具有流程图格式的流程，减少了人工转换的工作量，为后续流程迁移提供高质量的具有流程图格式的流程，提高了效率和减少了人工成本。
S20，将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象。
可理解地，通过所述领域设计模型对各个所述流程中的所述流程文档信息和所述流程节点进行所述领域语义特征提取，所述领域语义特征为与领域对象相关的特征，以及领域对象之间的上下文语义特征和场景特征，识别的过程为通过所述领域设计模型对所述流程文档信息和所述流程节点进行领域对象之间的上下文语义特征和场景特征提取，对提取的领域对象之间的上下文语义特征进行上下文语义识别，和对提取的上下文场景特征进行上下文场景识别，识别出针对上下文语义识别的第一识别结果和针对上下文场景识别的第二识别结果，根据所述第一识别结果和所述第二识别结果，进行领域界限定位，确定出与所述流程对应的所述领域对象的识别过程。
其中，所述流程文档信息包括流程需求文档和流程场景信息；所述流程需求文档为开发所述流程输入的需求类的文档，即所述流程需求文档包括流程节点的接口设计、流程节点的执行顺序和上下流程节点之间的关系，所述流程场景信息包括所述流程应用的场景、 所述流程面向的场景对象以及该流程实现的目的或者目标，所述流程场景信息为体现与该流程应用的场景相关的信息，所述流程节点包括流程接口和节点对象；所述领域设计模型为基于领域驱动设计(DDD)算法进行训练完成的且用于识别流程所述领域及进行流程迁移的模型，所述领域驱动设计概念为以一种统一的通用语言作为领域之间相互训练的工具，在不断训练的过程中发现和挖出主要的领域特征对象，然后将这些领域特征对象设计成一个领域模型，用代码搭建该领域模型，所述领域对象表征了与其对应的流程的所述领域，所述领域对象的种类为预先定义出的领域名称。
在一实施例中，如图4所示，所述步骤S20中，即所述通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象，包括：
S201，将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象；所述流程文档信息包括流程需求文档和流程场景信息；所述流程节点包括流程接口和节点对象。
可理解地，所述拼接的方式可以根据需求设定，优选地，将所述流程接口在所述流程需求文档的前后进行拼接，将拼接后的所述流程需求文档进行embedding转换，得到所述第一文本对象，其中，所述embedding转换为对输入的文本内容划分为单个字或词，对每个字或词进行映射转换成与其对应的词向量，映射关系为通过训练完embedding模型获得，将转换后的所有词向量进行特征向量组合得到一串特征向量值，将其确定为所述第一文本对象，同时在所述流程场景信息中的每句后插入所述节点对象，合并生成所述第二文本对象。
其中，所述流程文档信息包括流程需求文档和流程场景信息；所述流程节点包括流程接口和节点对象，所述流程接口为流程节点中使用到的接口，例如：流程接口为获取审批人员的相关信息的数据库的接口，所述节点对象为流程节点面向的对象，即流程节点中包含输入的对象的集合，例如：节点对象包括流程节点中输入的金额、时间和各输入栏的文本信息等。
在一实施例中，所述步骤S201中，即所述将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象，包括：
S2011，将所述流程接口在所述流程需求文档的前后进行拼接，并进行embedding转换，得到与所述流程对应的所述第一文本对象。
可理解地，所述前后进行拼接的过程为在所述流程需求文档之前插入所述流程接口的内容，在所述流程需求文档之后插入所述流程接口的内容，将插入后的所述流程需求文档确定为拼接完成的过程，所述embedding转换为对输入的文本内容划分为单个字或词，对每个字或词进行映射转换成与其对应的词向量，映射关系为通过训练完embedding模型获得，将转换后的所有词向量进行特征向量组合得到一串特征向量值，将其确定为所述第一文本对象。
S2012，在所述流程场景信息中的每句后插入所述节点对象，并合并成与所述流程对应的所述第二文本对象。
可理解地，所述合并的过程为在所述流程场景信息中的每个标点符号的后面插入所述节点对象，再在所述节点对象后插入一个分隔符后，如此实现了在每句后插入所述节点对象的操作过程，合并生成所述第二文本对象。
本申请实现了通过将所述流程接口在所述流程需求文档的前后进行拼接，并进行embedding转换，得到与所述流程对应的所述第一文本对象；在所述流程场景信息中的每句后插入所述节点对象，并合并成与所述流程对应的所述第二文本对象，如此，实现了结 合流程接口与流程需求文档的方法，为后续的上下文语义识别提供了数据基础，以及结合流程场景信息和节点对象的方法，为后续的上下文场景识别提供了数据基础，提高了识别的准确性和质量，最终提高了后续领域对象识别的准确性。
S202，通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，同时通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果。
可理解地，所述第一文本识别模型实现了通过识别所述第一文本对象中的上下文语义特征，并根据提取出的所述上下文语义特征，识别出节点类别结果，以及对识别出的节点类别结果进行业务领域识别，得到第一识别结果的模型，所述第一文本识别模型为训练完成的语言模型，所述上下文语义特征为两个句子之间的上下文关联的语义方面的且与节点类别相关的特征，所述节点类别结果为识别出的各个业务的类别的权重分布，所述节点类别为业务的类别，在所述第一文本对象中增加所述流程节点，目的是让所述流程需求文档的上下文中结合流程节点，让第一文本识别模型的第一识别结果的识别准确率提高，更容易能够识别出节点类别，提高了第一识别结果的识别效率。
可理解地，所述第二文本识别模型实现了通过识别所述第二文本对象中的上下文场景特征，并根据提取出的所述上下文场景特征，识别出场景类别结果，以及对识别出的场景类别结果进行场景领域识别，得到第二识别结果的模型，所述第二文本识别模型为训练完成的语言模型，所述上下文场景特征为两个句子之间的上下文关联的与场景类别相关的特征，所述场景类别为预设的应用到的场景的分类，所述场景类别结果包括至少一个场景类别及与其对应的预测概率，即识别出的各个场景的类别的概率分布，在所述第二文本对象中增加所述节点对象，目的是让所述流程场景信息的上下文中结合节点对象，让第二文本识别模型的第二识别结果的识别准确率提高，更容易能够识别出场景类别，提高了第二识别结果的识别效率。
其中，所述领域设计模型包括所述第一文本识别模型和所述第二文本识别模型，所述第一识别结果为根据节点类别结果，从业务领域维度确定出各个领域的概率分布结果，所述第二识别结果为根据场景类别结果，从场景领域维度确定出各个领域的概率分布结果。
在一实施例中，所述步骤S202中，即所述通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，包括：
S2021，通过所述第一文本识别模型提取所述第一文本对象中的上下文语义特征；所述第一文本识别模型为基于Bi-LSTM的语言模型。
可理解地，所述第一文本识别模型为在Bi-LSTM的网络结构基础上构建的模型，所述Bi-LSTM的网络结构的模型包括前向LSTM模型和后向LSTM模型，通过双方向的LSTM模型捕捉双方向的上下文语义，通过所述第一文本识别模型对所述第一文本对象进行双方向的上下文语义卷积，提取所述上下文的语义特征，所述上下文语义特征为两个句子之间的上下文关联的语义方面的且与节点类别相关的特征。
S2022，通过所述第一文本识别模型根据提取的所述上下文语义特征，识别出与所述第一文本对象对应的节点类别结果。
可理解地，通过所述第一文本识别模型对提取的所述上下文语义特征进行编排，输出成一个业务特征向量，并通过所述第一文本识别模型中的全连接层输出一个一维的业务全连接特征向量，通过各所述节点类别对应的权重，与相应的所述业务全连接特征向量中的向量进行加权相乘，从而识别出所述节点类别结果。
其中，所述节点类别结果包括至少一个节点类别及与其对应的权重，即所述节点类别结果为识别出的各个业务的类别的权重分布，所述节点类别结果表明了在各个业务的领域预测出的权重，权重的范围为0到1的范围。
S2023，通过所述第一文本识别模型对所述节点类别结果进行业务领域识别，得到与 所述第一文本对象对应的所述第一识别结果。
可理解地，所述业务领域识别为针对流程节点所属的业务维度的领域进行识别，也即通过所述节点类别结果中的权重分布预测出各个业务维度的领域的概率分布情况，也可理解为将所述节点类别结果中相同业务维度的领域下的节点类别进行聚合得到的概率分布。
本申请实现了通过所述第一文本识别模型提取所述第一文本对象中的上下文语义特征；所述第一文本识别模型为基于Bi-LSTM的语言模型；通过所述第一文本识别模型根据提取的所述上下文语义特征，识别出与所述第一文本对象对应的节点类别结果；通过所述第一文本识别模型对所述节点类别结果进行业务领域识别，得到与所述第一文本对象对应的所述第一识别结果，如此，实现了通过基于Bi-LSTM的语言模型的第一文本识别模型提取第一文本对象中的上下文语义特征，运用业务领域识别方法，自动识别出第一文本对象的第一识别结果，更容易能够识别出节点类别，提高了第一识别结果的识别效率，减少了人工识别，也为后续提高了识别准确性和效率。
在一实施例中，所述步骤S202中，即所述通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果，包括：
S2024，通过所述第二文本识别模型对所述第二文本对象进行场景特征提取；所述第二文本识别模型为基于Bert的识别模型。
可理解地，所述第二文本识别模型为在Bert的网络结构基础上构建的模型，所述Bert的网络结构的模型为通过联合调节所有层中的双向Transformer算法进行训练，并且对所述第二文本对象中的文本进行词义向量转换以及标注的模型，通过所述第而文本识别模型对所述第二文本对象进行提取所述上下文场景特征，所述上下文场景特征为两个句子之间的上下文关联的与场景类别相关的特征。S2025，通过所述第二文本识别模型根据提取的所述场景特征进行上下文关联度识别，识别出与所述第二文本对象对应的场景类别结果。
可理解地，通过所述第二文本识别模型对提取的所述上下文场景特征进行关联度识别，所述关联度识别为上下文的涉及场景类别的文字进行关联性分析，分析出一个场景特征向量，所述分析过程为上下文之间的场景跨度或者场景维度进行关联关系分析的过程，并通过所述第二文本识别模型中的全连接层输出一个一维的场景全连接特征向量，通过全连接识别出所述节点类别结果。
其中，所述场景类别结果包括至少一个场景类别及与其对应的预测概率，即所述场景类别结果为识别出的各个场景的类别的概率分布，所述场景类别结果表明了在各个场景的领域预测出的概率值，概率值的范围为0％到100％的范围。
S2026，通过所述第二文本识别模型对所述场景类别结果进行场景领域识别，得到与所述第二文本对象对应的所述第二识别结果。
可理解地，所述场景领域识别为针对流程节点所属的场景维度的领域进行识别，也即通过所述场景类别结果中的概率分布预测出各个场景维度的领域的概率分布情况，也可理解为将所述场景类别结果中相同场景维度的领域下的场景类别进行聚合得到的概率分布。
本申请实现了通过所述第二文本识别模型对所述第二文本对象进行场景特征提取；所述第二文本识别模型为基于Bert的识别模型；通过所述第二文本识别模型根据提取的所述场景特征进行上下文关联度识别，识别出与所述第二文本对象对应的场景类别结果；通过所述第二文本识别模型对所述场景类别结果进行场景领域识别，得到与所述第二文本对象对应的所述第二识别结果，如此，实现了通过基于Bert的识别模型的第二文本识别模型提取第二文本对象中的上下文场景特征，运用场景领域识别方法，自动识别出第二文本对象的第二识别结果，减少了人工识别，以及通过场景维度对流程节点进行识别，为后续提高了识别的准确性和效率。
S203，通过所述领域设计模型对所述第一识别结果和所述第二识别结果进行领域界限定位，确定出与所述流程对应的所述领域对象。
可理解地，所述领域界限定位为将所述第一识别结果和所述第二识别结果转换成一个维度的数值范围，然后在领域多边形中定位出与所述第一识别结果对应的业务中心点，和与所述第二识别结果对应的场景中心点，在根据所述业务中心点和所述场景中心点距离各个领域对象的界限的距离，综合定位出领域对象，例如：数值范围为1至10的范围，将所有领域对象构成一个与所有所述领域对象总数相同点的多边形，每个点代表一个所述领域对象的10分值，该多边形的中心为各个领域对象的0分值，即第一识别结果中值为1对应的业务的领域，第二识别结果中值为100％对应的场景的领域，一个所述领域对象对应一个业务的领域和一个场景的领域，根据所述第一识别结果的分布情况在构成的多边形中定位出各个业务的领域相应的点，然后将其连接成一个业务类多边形，取该业务类多边形的中心作为所述第一识别结果的业务中心点，根据所述第二识别结果的分布情况在构成的多边形中定位出各个场景的领域相应的点，然后将其连接成一个场景类多边形，取该场景类多边形的中心作为所述第二识别结果的场景中心点，根据所述业务中心点和所述场景中心点距离所述领域对象的各个界限的距离，即所述业务中心点和所述场景中心点落入各个领域对象的界限的范围，然后根据范围的重叠占比情况确定出最终的领域对象，则其为与所述流程对应的所述领域对象。
本申请实现了通过将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象；通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，同时通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果；通过所述领域设计模型对所述第一识别结果和所述第二识别结果进行领域界限定位，确定出与所述流程对应的所述领域对象，如此，通过第一文本识别模型和第二文本识别模型，分别从上下文语义和上下文文场景进行识别，通过两个维度的领域界限定位，确定流程的领域，实现了自动识别流程的所属领域，以供后续将需要迁移的领域下的流程进行迁移。
S30，从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程。
可理解地，从所述流程集合中筛选出与所述待迁领域匹配的所述领域对象，再查找到与匹配的所述领域对象对应的所述流程，并将查找到的所述流程确定为所述待迁移流程，所述待迁移流程为需要进行迁移的流程。
其中，所述领域迁移清单包括多个预设的所述待迁领域。
S40，通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根。
可理解地，通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行汇聚，并构建出流程节点图谱，所述汇聚的过程为将相同的所述流程节点汇集到一个流程节点，该流程节点包含了与原流程节点关联的所述流程节点关系的过程，构建的过程为根据所有所述流程节点关系将所有汇集后的流程节点进行关联，形成一个网状的所述流程节点图谱；再运用康威定律算法，对所述流程节点图谱进行聚合，得到与所述领域对象对应的所述聚合结果，所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根。
其中，所述康威定律算法为将类似的流程节点和流程节点关系进行合并，生成一个新流程节点，该新的流程节点能够包括原来的流程节点的功能和原来流程节点关系，将该新流程节点确定确定为一个实体，聚合后的流程节点关系确定为聚合根，将该实体包含的流程节点确定为值对象的算法。
在一实施例中，如图5所示，所述步骤S40中，即所述通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果，包括：
S401，通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行汇聚，并构建出流程节点图谱。
可理解地，所述汇聚为从所有所述流程节点中将相同领域对象的所述待迁移流程中的所述流程节点分离出来，并将分离出来的两个所述流程节点之间的关系进行汇合，如此，构建出该领域对象的所述流程节点图谱，从而构建出所有所述领域对象的所述流程节点图谱。
S402，运用康威定律算法，通过所述领域设计模型对所述流程节点图谱进行聚合，得到与所述领域对象对应的所述聚合结果。
可理解地，所述康威定律算法为将相同所述流程节点图谱下的类似的流程节点和流程节点关系进行合并，生成一个新流程节点，该新的流程节点能够包括原来的流程节点的功能和原来流程节点关系，将该新流程节点确定为一个实体，聚合后的流程节点关系确定为聚合根，将该实体包含的流程节点确定为值对象的算法，所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根。
本申请实现了通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行汇聚，并构建出流程节点图谱；运用康威定律算法，通过所述领域设计模型对所述流程节点图谱进行聚合，得到与所述领域对象对应的所述聚合结果，如此，通过领域设计模型将相同领域对象的流程进行汇聚，构建流程节点图谱，并运用康威定律算，对流程节点图谱进行聚合输出聚合结果，减少了人工聚合的成本，自动将各流程节点和流程节点关系进行聚合成实体、值对象和聚合根，节省了目标流程的容量，大大压缩了流程的占用空间，并统一了构建平台和开发语言。
S50，根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表。
可理解地，所述实体包括实体名、实体存储数据格式和实体接口，在所述目标系统中创建符合所述实体的所述实体数据表，所述数据表的表名与所述实体名关联，所述数据表中的字段与所述实体关联的所述值对象相对应，所述实体数据表存储的数据库结构与所述实体存储数据格式相同，所述实体数据表可以通过所述实体接口进行通信。
S60，根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
可理解地，将所述待迁移流程中的各个所述流程节点通过用与其对应的所述实体进行替代，并且关联该实体对应的所述实体数据表，以及将流程节点关系用所述值对象和所述聚合根进行场景替代，即不同的所述值对象会对应不同的所述聚合根的关系，如此，重构出与所述待迁移流程对应的所述目标流程，所述目标流程实现的流程功能与所述待迁移流程的流程功能相同，并且将所有所述目标流程统一用一种开发语言，对不同语言的流程进行了统一，便于后续的运维。
本申请实现了通过接收到系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所 述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；根据各与所述领域对象对应的所述聚合结果中的所有所述实体和与其关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；根据所述待迁移流程、所有与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构完所有所述目标流程以确定所述系统流程迁移请求完成。
如此，实现了通过获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将其确定为待迁移流程；通过所述领域设计模型将相同领域对象的所述待迁移流程进行聚合，得到聚合结果；根据各所述聚合结果中的所有所述实体和与其关联的所述值对象，在所述目标系统创建实体数据表；根据所述待迁移流程、所有与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，重构各目标流程，因此，实现了通过领域语义特征提取及识别，识别出各个需迁移的流程的领域，再通过领域设计模型将相同领域的流程进行聚合，聚合得到含有实体、值对象和聚合根的聚合结果，根据该聚合结果，创建实体数据表以及重构与各流程一一对应的目标流程，完成流程迁移，自动识别各流程的所属领域，运用基于领域驱动设计算法的领域设计模型将需要迁移的领域的流程迁移至目标系统，无需人工筛选和开发构建，直接聚合生成实体、值对象和聚合根，将所有流程统一成一种开发语言，和创建统一的实体数据表以及重构目标流程，大大节省了流程的存储空间，达到准确和可靠的将流程迁移至目标系统的效果，提高了流程迁移的效率和质量。
在一实施例中，提供一种流程迁移装置，该流程迁移装置与上述实施例中流程迁移方法一一对应。如图6所示，该流程迁移装置包括接收模块11、提取模块12、匹配模块13、聚合模块14、创建模块15和重构模块16。各功能模块详细说明如下：
接收模块11，用于响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
提取模块12，用于将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
匹配模块13，用于从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
聚合模块14，用于通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
创建模块15，用于根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
重构模块16，用于根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
关于流程迁移装置的具体限定可以参见上文中对于流程迁移方法的限定，在此不再赘述。上述流程迁移装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于 计算机设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中，提供了一种计算机设备，该计算机设备可以是服务器，其内部结构图可以如图7所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括可读存储介质、内存储器。该可读存储介质存储有操作系统、计算机可读指令和数据库。该内存储器为可读存储介质中的操作系统和计算机可读指令的运行提供环境。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机可读指令被处理器执行时以实现一种流程迁移方法。本实施例所提供的可读存储介质包括非易失性可读存储介质和易失性可读存储介质。
在一个实施例中，提供了一个或多个存储有计算机可读指令的可读存储介质，本实施例所提供的可读存储介质包括非易失性可读存储介质和易失性可读存储介质；该可读存储介质上存储有计算机可读指令，该计算机可读指令被一个或多个处理器执行时，使得一个或多个处理器实现上述实施例中流程迁移方法。
在一个实施例中，提供了一种计算机可读存储介质，其上存储有计算机可读指令，计算机可读指令被处理器执行时实现上述实施例中流程迁移方法。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机可读指令来指令相关的硬件来完成，所述的计算机可读指令可存储于一非易失性计算机可读取存储介质或易失性可读存储介质中，该计算机可读指令在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上所述实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围，均应包含在本申请的保护范围之内。

Claims (20)

  1. 一种流程迁移方法，其中，包括：
    响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
    将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
    从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
    通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
    根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
    根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
  2. 如权利要求1所述的流程迁移方法，其中，所述接收到系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单之前，包括：
    获取所述源系统的所有原始流程；
    通过流程图构建模型对各所述原始流程进行语言类别识别，识别出与各所述原始流程对应的语言类型；
    通过流程图构建模型获取与所述原始流程对应的语言类型对应的解读脚本，并通过获取的与所述原始流程对应的所述解读脚本对所述原始流程进行解读，并按照流程图生成方式构建与该原始流程对应的所述流程。
  3. 如权利要求1所述的流程迁移方法，其中，所述通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象，包括：
    将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象；所述流程文档信息包括流程需求文档和流程场景信息；所述流程节点包括流程接口和节点对象；
    通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，同时通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果；
    通过所述领域设计模型对所述第一识别结果和所述第二识别结果进行领域界限定位，确定出与所述流程对应的所述领域对象。
  4. 如权利要求3所述的流程迁移方法，其中，所述将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象，包括：
    将所述流程接口在所述流程需求文档的前后进行拼接，并进行embedding转换，得到与所述流程对应的所述第一文本对象；
    在所述流程场景信息中的每句后插入所述节点对象，并合并成与所述流程对应的所述 第二文本对象。
  5. 如权利要求3所述的流程迁移方法，其中，所述通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，包括：
    通过所述第一文本识别模型提取所述第一文本对象中的上下文语义特征；所述第一文本识别模型为基于Bi-LSTM的语言模型；
    通过所述第一文本识别模型根据提取的所述上下文语义特征，识别出与所述第一文本对象对应的节点类别结果；
    通过所述第一文本识别模型对所述节点类别结果进行业务领域识别，得到与所述第一文本对象对应的所述第一识别结果。
  6. 如权利要求5所述的流程迁移方法，其中，所述通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果，包括：
    通过所述第二文本识别模型对所述第二文本对象进行场景特征提取；所述第二文本识别模型为基于Bert的识别模型；
    通过所述第二文本识别模型根据提取的所述场景特征进行上下文关联度识别，识别出与所述第二文本对象对应的场景类别结果；
    所述场景类别结果包括至少一个场景类别及与其对应的预测概率；
    通过所述第二文本识别模型对所述场景类别结果进行场景领域识别，得到与所述第二文本对象对应的所述第二识别结果。
  7. 如权利要求1所述的流程迁移方法，其中，所述通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果，包括：
    通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行汇聚，并构建出流程节点图谱；
    运用康威定律算法，通过所述领域设计模型对所述流程节点图谱进行聚合，得到与所述领域对象对应的所述聚合结果。
  8. 一种流程迁移装置，其中，包括：
    接收模块，用于响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
    提取模块，用于将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
    匹配模块，用于从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
    聚合模块，用于通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
    创建模块，用于根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
    重构模块，用于根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
  9. 一种计算机设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机可读指令，其中，所述处理器执行所述计算机可读指令时实现如下步骤： 响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
    将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
    从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
    通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
    根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
    根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
  10. 如权利要求9所述的计算机设备，其中，所述接收到系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单之前，所述处理器执行所述计算机可读指令时还实现如下步骤：
    获取所述源系统的所有原始流程；
    通过流程图构建模型对各所述原始流程进行语言类别识别，识别出与各所述原始流程对应的语言类型；
    通过流程图构建模型获取与所述原始流程对应的语言类型对应的解读脚本，并通过获取的与所述原始流程对应的所述解读脚本对所述原始流程进行解读，并按照流程图生成方式构建与该原始流程对应的所述流程。
  11. 如权利要求9所述的计算机设备，其中，所述通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象，包括：
    将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象；所述流程文档信息包括流程需求文档和流程场景信息；所述流程节点包括流程接口和节点对象；
    通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，同时通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果；
    通过所述领域设计模型对所述第一识别结果和所述第二识别结果进行领域界限定位，确定出与所述流程对应的所述领域对象。
  12. 如权利要求11所述的计算机设备，其中，所述将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象，包括：
    将所述流程接口在所述流程需求文档的前后进行拼接，并进行embedding转换，得到与所述流程对应的所述第一文本对象；
    在所述流程场景信息中的每句后插入所述节点对象，并合并成与所述流程对应的所述第二文本对象。
  13. 如权利要求11所述的计算机设备，其中，所述通过所述领域设计模型中的第一文 本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，包括：
    通过所述第一文本识别模型提取所述第一文本对象中的上下文语义特征；所述第一文本识别模型为基于Bi-LSTM的语言模型；
    通过所述第一文本识别模型根据提取的所述上下文语义特征，识别出与所述第一文本对象对应的节点类别结果；
    通过所述第一文本识别模型对所述节点类别结果进行业务领域识别，得到与所述第一文本对象对应的所述第一识别结果。
  14. 如权利要求13所述的计算机设备，其中，所述通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果，包括：
    通过所述第二文本识别模型对所述第二文本对象进行场景特征提取；所述第二文本识别模型为基于Bert的识别模型；
    通过所述第二文本识别模型根据提取的所述场景特征进行上下文关联度识别，识别出与所述第二文本对象对应的场景类别结果；
    所述场景类别结果包括至少一个场景类别及与其对应的预测概率；
    通过所述第二文本识别模型对所述场景类别结果进行场景领域识别，得到与所述第二文本对象对应的所述第二识别结果。
  15. 一个或多个存储有计算机可读指令的可读存储介质，其中，所述计算机可读指令被一个或多个处理器执行时，使得所述一个或多个处理器执行如下步骤：
    响应与系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单；所述流程集合包括多个流程，每个所述流程包括流程文档信息、流程节点和流程节点关系；
    将所有所述流程输入领域设计模型中，通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流程对应的领域对象；
    从所述流程集合中筛选出与所述领域迁移清单中的待迁领域匹配的所述领域对象对应的所述流程，将筛选出的所述流程确定为待迁移流程；
    通过所述领域设计模型对与相同领域对象的所述待迁移流程中的所述流程节点和所述流程节点关系进行聚合，得到与该领域对象对应的聚合结果；所述聚合结果包括与所述流程节点对应的实体、与所述实体关联的值对象和聚合根；
    根据各所述聚合结果中的所述实体和与所述实体关联的所述值对象，在所述目标系统创建与各所述实体关联的实体数据表；
    根据所述待迁移流程、与所述待迁移流程中的所述流程节点对应的所述实体以及与该实体关联的所述实体数据表、所述值对象和所述聚合根，在所述目标系统重构与各所述待迁移流程对应的各目标流程，重构所有所述目标流程并确定所述系统流程迁移请求完成。
  16. 如权利要求15所述的可读存储介质，其中，所述接收到系统流程迁移请求，获取所述系统流程迁移请求中的目标系统、源系统的流程集合和领域迁移清单之前，所述计算机可读指令被一个或多个处理器执行时，使得所述一个或多个处理器还执行如下步骤：
    获取所述源系统的所有原始流程；
    通过流程图构建模型对各所述原始流程进行语言类别识别，识别出与各所述原始流程对应的语言类型；
    通过流程图构建模型获取与所述原始流程对应的语言类型对应的解读脚本，并通过获取的与所述原始流程对应的所述解读脚本对所述原始流程进行解读，并按照流程图生成方式构建与该原始流程对应的所述流程。
  17. 如权利要求16所述的可读存储介质，其中，所述通过所述领域设计模型对各所述流程中的所述流程文档信息和所述流程节点进行领域语义特征提取，识别出与各个所述流 程对应的领域对象，包括：
    将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象；所述流程文档信息包括流程需求文档和流程场景信息；所述流程节点包括流程接口和节点对象；
    通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，同时通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果；
    通过所述领域设计模型对所述第一识别结果和所述第二识别结果进行领域界限定位，确定出与所述流程对应的所述领域对象。
  18. 如权利要求17所述的可读存储介质，其中，所述将所述流程中的所述流程需求文档和所述流程接口进行拼接，得到与该流程对应的第一文本对象，同时将所述流程中的所述流程场景信息和所述节点对象进行合并，得到与该流程对应的第二文本对象，包括：
    将所述流程接口在所述流程需求文档的前后进行拼接，并进行embedding转换，得到与所述流程对应的所述第一文本对象；
    在所述流程场景信息中的每句后插入所述节点对象，并合并成与所述流程对应的所述第二文本对象。
  19. 如权利要求17所述的可读存储介质，其中，所述通过所述领域设计模型中的第一文本识别模型对所述第一文本对象进行上下文语义识别，识别出第一识别结果，包括：
    通过所述第一文本识别模型提取所述第一文本对象中的上下文语义特征；所述第一文本识别模型为基于Bi-LSTM的语言模型；
    通过所述第一文本识别模型根据提取的所述上下文语义特征，识别出与所述第一文本对象对应的节点类别结果；
    通过所述第一文本识别模型对所述节点类别结果进行业务领域识别，得到与所述第一文本对象对应的所述第一识别结果。
  20. 如权利要求19所述的可读存储介质，其中，所述通过所述领域设计模型中的第二文本识别模型对所述第二文本对象进行长下文场景识别，识别出第二识别结果，包括：
    通过所述第二文本识别模型对所述第二文本对象进行场景特征提取；所述第二文本识别模型为基于Bert的识别模型；
    通过所述第二文本识别模型根据提取的所述场景特征进行上下文关联度识别，识别出与所述第二文本对象对应的场景类别结果；
    所述场景类别结果包括至少一个场景类别及与其对应的预测概率；
    通过所述第二文本识别模型对所述场景类别结果进行场景领域识别，得到与所述第二文本对象对应的所述第二识别结果。
PCT/CN2021/090527 2020-12-30 2021-04-28 流程迁移方法、装置、设备及介质 WO2022142015A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011620085.5 2020-12-30
CN202011620085.5A CN112732423B (zh) 2020-12-30 2020-12-30 流程迁移方法、装置、设备及介质

Publications (1)

Publication Number Publication Date
WO2022142015A1 true WO2022142015A1 (zh) 2022-07-07

Family

ID=75609625

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/090527 WO2022142015A1 (zh) 2020-12-30 2021-04-28 流程迁移方法、装置、设备及介质

Country Status (2)

Country Link
CN (1) CN112732423B (zh)
WO (1) WO2022142015A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115147096A (zh) * 2022-09-06 2022-10-04 京华信息科技股份有限公司 一种基于ocr的流程快速建模方法及系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110071876A1 (en) * 2009-09-18 2011-03-24 International Business Machines Corporation Planning and orchestrating workflow instance migration
CN105741087A (zh) * 2016-03-15 2016-07-06 中国农业大学 业务流程导入的方法及系统
CN107885758A (zh) * 2016-09-30 2018-04-06 华为技术有限公司 一种虚拟节点的数据迁移方法和虚拟节点
CN109558089A (zh) * 2018-12-03 2019-04-02 湖南御家科技有限公司 一种数据迁移方法、装置、设备及可读存储介质
CN110209652A (zh) * 2019-05-20 2019-09-06 平安科技（深圳）有限公司 数据表迁移方法、装置、计算机设备和存储介质

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111522796B (zh) * 2020-04-24 2022-10-28 苏州达家迎信息技术有限公司 系统间的数据迁移方法、装置、计算机设备及介质
CN111813739A (zh) * 2020-09-11 2020-10-23 平安国际智慧城市科技股份有限公司 数据迁移方法、装置、计算机设备及存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110071876A1 (en) * 2009-09-18 2011-03-24 International Business Machines Corporation Planning and orchestrating workflow instance migration
CN105741087A (zh) * 2016-03-15 2016-07-06 中国农业大学 业务流程导入的方法及系统
CN107885758A (zh) * 2016-09-30 2018-04-06 华为技术有限公司 一种虚拟节点的数据迁移方法和虚拟节点
CN109558089A (zh) * 2018-12-03 2019-04-02 湖南御家科技有限公司 一种数据迁移方法、装置、设备及可读存储介质
CN110209652A (zh) * 2019-05-20 2019-09-06 平安科技（深圳）有限公司 数据表迁移方法、装置、计算机设备和存储介质

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115147096A (zh) * 2022-09-06 2022-10-04 京华信息科技股份有限公司 一种基于ocr的流程快速建模方法及系统

Also Published As

Publication number Publication date
CN112732423B (zh) 2023-01-20
CN112732423A (zh) 2021-04-30

Similar Documents

Publication Publication Date Title
WO2021042503A1 (zh) 信息分类抽取方法、装置、计算机设备和存储介质
JP6549332B2 (ja) 機械学習に基づくネットワークモデル構築方法及び装置
CN111026842B (zh) 自然语言处理方法、自然语言处理装置及智能问答系统
WO2021151328A1 (zh) 症状数据处理方法、装置、计算机设备及存储介质
CN109670191A (zh) 机器翻译的校准优化方法、装置与电子设备
WO2024022354A1 (zh) 结合rpa及ai实现ia的对象推荐方法、装置及存储介质
US11886850B2 (en) Transformation templates to automate aspects of computer programming
CN113536795B (zh) 实体关系抽取的方法、系统、电子装置和存储介质
CN112183030A (zh) 基于预设神经网络的事件抽取方法、装置、计算机设备及存储介质
CN112652386A (zh) 分诊数据处理方法、装置、计算机设备及存储介质
CN114064852A (zh) 自然语言的关系抽取方法、装置、电子设备和存储介质
WO2022142015A1 (zh) 流程迁移方法、装置、设备及介质
CN110968664A (zh) 一种文书检索方法、装置、设备及介质
CN113343692B (zh) 搜索意图的识别方法、模型训练方法、装置、介质及设备
CN116721713B (zh) 一种面向化学结构式识别的数据集构建方法和装置
WO2020057023A1 (zh) 自然语言的语义解析方法、装置、计算机设备和存储介质
EP4064038B1 (en) Automated generation and integration of an optimized regular expression
CN117296064A (zh) 计算环境中的可解释人工智能
CN114090722A (zh) 查询内容自动补全的方法及装置
CN114238715A (zh) 基于社会救助的问答系统、构建方法、计算机设备及介质
CN114021541A (zh) 演示文稿生成方法、装置、设备及存储介质
CN110928896A (zh) 一种数据查询方法及装置
US20240143334A1 (en) Text block classification by dependency tree generation
CN112580309B (zh) 文档数据处理方法、装置、计算机设备及存储介质
EP4300366A1 (en) Method, apparatus, and system for multi-modal multi-task processing

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21912777

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21912777

Country of ref document: EP

Kind code of ref document: A1