Workflow description file generation method and related products

Abstract

This disclosure provides a method for generating workflow description files and related products. The specific implementation of this method includes: obtaining workflow definition information and generating a workflow description file based on the workflow definition information. This method, by configuring connection edges and introducing constraint rule matching verification during workflow definition and execution, achieves an intelligent transformation and upgrade of connection edges from simple connection relationships to those with independent data adaptation, semantic alignment, and real-time verification capabilities. This completely solves the problems of incompatible data formats and semantic inconsistencies between upstream and downstream task nodes, significantly improving the flexibility of workflow configuration, debugging efficiency, and execution stability.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of artificial intelligence technology, specifically to a workflow description file generation method and related products. Background Technology

[0002] With the deep application of artificial intelligence (AI) technology in biopharmaceutical R&D, core processes such as molecular design, antigen epitope analysis, and antibody development are gradually realizing the collaborative application of AI models and professional tools. Building automated AI workflows has become a key path to improve the efficiency of biopharmaceutical R&D and unleash the value of AI technology. The biopharmaceutical R&D process generates massive amounts of heterogeneous data, covering various types including molecular structure formats such as CIF / PDB, protein sequence data, and antigen epitope semantic data. Furthermore, different AI models and R&D tools have strong constraints on the format, semantics, and structure of the data, which places extremely high demands on the data flow and adaptation capabilities between workflow nodes.

[0003] However, in practical applications of biomedical R&D, existing workflow platforms still have many shortcomings and struggle to meet the core needs of heterogeneous data processing and AI collaboration within the field. On one hand, traditional workflow platforms treat nodes as the sole computing unit, with each node merely handling simple node connection and data transfer without any data processing capabilities. They employ a weak management model for data transfer between nodes, neglecting both data format and structure compatibility and the ability to perform heterogeneous data conversion at the syntactic and semantic levels. This completely fails to meet the core requirement of strong data type consistency in the biomedical field. On the other hand, existing workflow definition languages ​​are of a general design, lacking standardized node and data interface definitions specific to biomedical scenarios. They cannot provide refined and standardized constraints on the input and output data of AI models and R&D tools, and they lack a dedicated heterogeneous data conversion and adaptation system, making it difficult to solve unique problems in areas such as incompatible molecular structure formats and inconsistent data semantic representations.

[0004] Therefore, it is necessary to propose a workflow description file generation method to solve at least one of the above-mentioned technical problems. Summary of the Invention

[0005] The embodiments of this disclosure provide a method for generating workflow description files and related products.

[0006] In a first aspect, this disclosure provides a method for generating a workflow description file, the method being executed by an electronic device, the method comprising: Obtain workflow definition information, which includes: at least two task nodes and at least one node connection component, wherein: The task node is used to execute specific tasks and has corresponding input constraint rules and output constraint rules; The node connection component is used to connect two adjacent task nodes and to perform the following data adaptation and transmission operations: acquiring upstream output data and its output constraint rules from the upstream task node; acquiring input constraint rules from the downstream task node; performing matching verification based on the output constraint rules and the input constraint rules; when the matching does not satisfy the input constraint rules, performing data adaptation calculation on the upstream output data based on predefined data transformation rules to generate downstream input data that satisfies the input constraint rules, and inputting the downstream input data to the downstream task node; when the matching satisfies the input constraint rules, directly inputting the upstream output data to the downstream task node. A workflow description file is generated based on the workflow definition information. The workflow description file is used by the execution engine to parse and execute the workflow.

[0007] In some optional implementations, the input constraint rules and the output constraint rules are used to limit the structural and content characteristics of the data, and include at least one of the following: data type constraint rules, data structure constraint rules, file format constraint rules, field integrity constraint rules, and enumeration value constraint rules; preferably, the predefined data transformation rules are used to transform the upstream output data into data that satisfies the input constraint rules, and the data transformation rules include at least one of the following: data type transformation rules, data structure transformation rules, file format transformation rules, field integrity supplementation rules, and enumeration value mapping rules; preferably, the data transformation rules also include semantic mapping rules, used to transform the first semantic representation in the upstream output data into a second semantic representation that satisfies the input constraint rules; preferably, the task node types include: artificial intelligence model nodes, tool invocation nodes, and sub-workflow nodes, wherein the artificial intelligence model nodes are used to invoke artificial intelligence models, the tool invocation nodes are used to invoke preset tools, and the sub-workflow nodes are used to define sub-workflows and execute specific tasks corresponding to the sub-workflows.

[0008] In some optional implementations, the first semantic representation and the second semantic representation differ in business meaning and correspond to different expressions of the same object; preferably, the semantic mapping rule is used to convert between at least one of the following: variable attributes and immutable attributes, a first data label set and a second data label set, a first constraint set and a second constraint set; preferably, the semantic mapping rule is determined based on at least one of the following: a predefined semantic mapping table, a rule engine, a machine learning model, and a manually configured mapping relationship; preferably, the semantic mapping rule is triggered when the output constraint rule and the input constraint rule do not match at the semantic level; preferably, the semantic mapping rule is used in biomedical data processing scenarios to perform semantic alignment conversion on data related to molecular structure, antigen epitopes, or sequence constraints.

[0009] In some optional implementations, the workflow definition information further includes a process control structure, which includes at least one of the following: a parallel structure, a conditional branching structure, and a loop structure; preferably, the loop structure is used to re-execute at least some task nodes until the preset evaluation index conditions are met when the target evaluation index does not meet the preset evaluation index conditions; preferably, the parallel structure is used to execute at least two task nodes in parallel; preferably, the conditional branching structure is used to execute the task node corresponding to the corresponding branch condition when the preset branch condition is met; preferably, the workflow definition information further includes a node definition for each task node, the node definition including a node identifier for uniquely identifying the task node, a node type identifier for limiting the type of the task node, and functional description information for describing the specific task function performed by the task node; preferably... Preferably, the node connection component includes a sequence of connection edges consisting of at least one connection edge. The output data of the upstream task node is transmitted as input data to the first connection edge in the sequence of connection edges, and the output data of the last connection edge in the sequence of connection edges is input data to the downstream task node. The output data of the upstream connection edge in two adjacent connection edges is input data to the downstream connection edge. Each connection edge is used to perform data transformation processing on the input data. Preferably, the workflow definition information also includes the connection edge definition of each connection edge. The connection edge definition includes an edge identifier for uniquely identifying the connection edge, an adapter type identifier for limiting the connection edge as a data transformation adapter, functional description information for describing the function of the connection edge, input constraint rules for limiting the format of the data to be transformed, and output constraint rules for limiting the format of the transformed data.

[0010] In some optional implementations, obtaining workflow definition information includes: Obtain workflow definition information generated by the user through a visual interactive interface, wherein the visual interactive interface includes at least one of the following display objects: a node icon for generating the task node definition in response to user operation, an edge icon for generating the connection edge sequence in response to user operation, a task node definition input area for generating the task node definition in response to user configuration operation, and an edge definition input area for generating the connection edge definition in response to user configuration operation. The method further includes: The execution engine is invoked to execute the executable code corresponding to the workflow description file to execute the workflow. When executing the workflow, the data input to the workflow is used as the input data of the first task node in the workflow. The task is executed according to the flow logic of the workflow, and the data output by the last task node in the workflow is used as the output data of the workflow.

[0011] In some optional implementations, the method further includes: In the visual interactive interface, in response to the detection of a new connection edge, a matching verification is performed based on the output constraint rules of the upstream task node and the input constraint rules of the downstream task node of the new connection edge; when the matching verification passes, the new connection edge is displayed in the first visual style; when the matching verification fails, the new connection edge is displayed in the second visual style.

[0012] Secondly, this disclosure provides a workflow description document generation apparatus, which is executed by an electronic device, and the apparatus includes: The definition information acquisition unit is used to acquire workflow definition information, which includes: at least two task nodes and at least one node connection component, wherein: The task node is used to execute specific tasks and has corresponding input constraint rules and output constraint rules; The node connection component is used to connect two adjacent task nodes and to perform the following data adaptation and transmission operations: acquiring upstream output data and its output constraint rules from the upstream task node; acquiring input constraint rules from the downstream task node; performing matching verification based on the output constraint rules and the input constraint rules; when the matching does not satisfy the input constraint rules, performing data adaptation calculation on the upstream output data based on predefined data transformation rules to generate downstream input data that satisfies the input constraint rules, and inputting the downstream input data to the downstream task node; when the matching satisfies the input constraint rules, directly inputting the upstream output data to the downstream task node. The generation unit is used to generate a workflow description file based on the workflow definition information. The workflow description file is used to be parsed and executed by the execution engine.

[0013] In some optional implementations, the input constraint rules and the output constraint rules are used to limit the structural and content characteristics of the data, and include at least one of the following: data type constraint rules, data structure constraint rules, file format constraint rules, field integrity constraint rules, and enumeration value constraint rules; preferably, the predefined data transformation rules are used to transform the upstream output data into data that satisfies the input constraint rules, and the data transformation rules include at least one of the following: data type transformation rules, data structure transformation rules, file format transformation rules, field integrity supplementation rules, and enumeration value mapping rules; preferably, the data transformation rules also include semantic mapping rules, used to transform the first semantic representation in the upstream output data into a second semantic representation that satisfies the input constraint rules; preferably, the task node types include: artificial intelligence model nodes, tool invocation nodes, and sub-workflow nodes, wherein the artificial intelligence model nodes are used to invoke artificial intelligence models, the tool invocation nodes are used to invoke preset tools, and the sub-workflow nodes are used to define sub-workflows and execute specific tasks corresponding to the sub-workflows.

[0014] In some optional implementations, the first semantic representation and the second semantic representation differ in business meaning and correspond to different expressions of the same object; preferably, the semantic mapping rule is used to convert between at least one of the following: variable attributes and immutable attributes, a first data label set and a second data label set, a first constraint set and a second constraint set; preferably, the semantic mapping rule is determined based on at least one of the following: a predefined semantic mapping table, a rule engine, a machine learning model, and a manually configured mapping relationship; preferably, the semantic mapping rule is triggered when the output constraint rule and the input constraint rule do not match at the semantic level; preferably, the semantic mapping rule is used in biomedical data processing scenarios to perform semantic alignment conversion on data related to molecular structure, antigen epitopes, or sequence constraints.

[0015] In some optional implementations, the workflow definition information further includes a process control structure, which includes at least one of the following: a parallel structure, a conditional branching structure, and a loop structure; preferably, the loop structure is used to re-execute at least some task nodes until the preset evaluation index conditions are met when the target evaluation index does not meet the preset evaluation index conditions; preferably, the parallel structure is used to execute at least two task nodes in parallel; preferably, the conditional branching structure is used to execute the task node corresponding to the corresponding branch condition when the preset branch condition is met; preferably, the workflow definition information further includes a node definition for each task node, the node definition including a node identifier for uniquely identifying the task node, a node type identifier for limiting the type of the task node, and functional description information for describing the specific task function performed by the task node; preferably... Preferably, the node connection component includes a sequence of connection edges consisting of at least one connection edge. The output data of the upstream task node is transmitted as input data to the first connection edge in the sequence of connection edges, and the output data of the last connection edge in the sequence of connection edges is input data to the downstream task node. The output data of the upstream connection edge in two adjacent connection edges is input data to the downstream connection edge. Each connection edge is used to perform data transformation processing on the input data. Preferably, the workflow definition information also includes the connection edge definition of each connection edge. The connection edge definition includes an edge identifier for uniquely identifying the connection edge, an adapter type identifier for limiting the connection edge as a data transformation adapter, functional description information for describing the function of the connection edge, input constraint rules for limiting the format of the data to be transformed, and output constraint rules for limiting the format of the transformed data.

[0016] In some optional implementations, the definition information acquisition unit is further configured to: Obtain workflow definition information generated by the user through a visual interactive interface, wherein the visual interactive interface includes at least one of the following display objects: a node icon for generating the task node definition in response to user operation, an edge icon for generating the connection edge sequence in response to user operation, a task node definition input area for generating the task node definition in response to user configuration operation, and an edge definition input area for generating the connection edge definition in response to user configuration operation. The device further includes: The execution engine invocation unit is used to invoke the execution engine to execute the executable code corresponding to the workflow description file in order to execute the workflow. When executing the workflow, the data input to the workflow is used as the input data of the first task node in the workflow. The task is executed according to the flow logic of the workflow, and the data output by the last task node in the workflow is used as the output data of the workflow.

[0017] In some alternative embodiments, the apparatus further includes: The matching and verification unit is used in the visual interactive interface to perform matching and verification based on the output constraint rules of the upstream task node and the input constraint rules of the downstream task node of the newly added connection edge in response to the detection of the newly added connection edge operation; when the matching and verification pass, the newly added connection edge is displayed in a first visual style; when the matching and verification fail, the newly added connection edge is displayed in a second visual style.

[0018] Thirdly, this disclosure provides an electronic device, including: One or more processors; Storage device, on which one or more programs are stored, When the above-described one or more programs are executed by the above-described one or more processors, the above-described one or more processors implement the method as described in any embodiment of the first aspect of this disclosure.

[0019] Fourthly, this disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by one or more processors, implements the method described in any embodiment of the first aspect of this disclosure.

[0020] Fifthly, this disclosure provides a computer program product including a computer program / instructions that, when executed by a processor, implement the method described in any embodiment of the first aspect of this disclosure.

[0021] The workflow description file generation method, apparatus, electronic device, and storage medium provided in the embodiments of this disclosure, by acquiring workflow definition information and generating workflow description files based on the workflow definition information, can achieve technical effects including but not limited to the following: First, the edges in the workflow are upgraded from simple node connection carriers to intelligent conversion units with independent data adaptation and computing capabilities, allowing edges and nodes to jointly undertake computing tasks, thereby achieving strong constraint management of heterogeneous biomedical data from the bottom layer.

[0022] This disclosed embodiment breaks away from the traditional workflow architecture of node computation and edge-only flow, endowing edges with core computational capabilities for data format verification, syntax conversion, and semantic mapping. This effectively solves the problem of weak management and lack of attention to data format in workflow platforms regarding data transfer between nodes. By performing full-dimensional matching and verification of input and output constraint rules for upstream and downstream task nodes at the edge level and executing targeted adaptation calculations, strong type consistency constraints on heterogeneous biomedical data are achieved at key stages of data flow. This overcomes the shortcomings of traditional workflows in adapting to the large-scale syntactic and semantic heterogeneous data conversion needs in the pharmaceutical field.

[0023] Second, by integrating special process control structures such as parallelism, conditional branching, and loops with the intelligent data adaptation capabilities of edges, it enables the engineering, deterministic definition and execution of biomedical R&D processes, adapting to the core needs of the field for standardization and iterative optimization of experimental processes.

[0024] This disclosed embodiment introduces a loop structure that supports iterative optimization, a parallel structure that improves efficiency, and a dynamically selected conditional branch structure into the workflow. Combined with the heterogeneous data adaptation capability of the edges, the workflow can accurately express the standardized process in biomedical research and development. The loop structure, combined with evaluation indicators in fields such as affinity prediction, enables automated iterative optimization of the research and development process. The parallel structure enables the synchronous invocation of multiple AI models / tools, allowing the workflow to fully adapt to the actual research and development logic of biomedical molecular design. This overcomes the shortcomings of traditional workflow process control capabilities being out of touch with domain business needs.

[0025] Third, we have built a complete process system from visual configuration and standardized description to compilation and execution, which supports the compilation of customized workflows into code that can be recognized by existing execution engines. While realizing domain-specific workflow capabilities, we have ensured compatibility with existing AI workflow orchestration frameworks.

[0026] This disclosure embodiment designs a visual interactive interface that allows users to configure nodes and edges by dragging and dropping, and provides real-time visual feedback on the constraint matching status of edges. This lowers the configuration threshold for domain workflows, effectively solves the problem of incompatibility between customized workflows and existing execution platforms, and overcomes the shortcomings of high implementation costs and poor compatibility of customized solutions.

[0027] Fourth, by implementing real-time verification of edge constraint matching and visual style differentiation in the visual interactive interface, combined with standardized node and adapter libraries, the configuration efficiency and debugging accuracy of the workflow are greatly improved, and the probability of errors in heterogeneous data processing processes is reduced.

[0028] This disclosure embodiment automatically performs matching and verification based on the constraint rules of upstream and downstream task nodes when a user adds a new connection edge, and distinguishes the verification results through different visual styles. At the same time, it provides reusable node and adapter resources, enabling users to quickly locate data incompatibility issues and accurately select adaptation and conversion rules. This effectively solves the problems of difficulty in discovering data format and semantic incompatibility issues and low debugging efficiency during workflow configuration. Attached Figure Description

[0029] Other features, objects, and advantages of this disclosure will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings. The drawings are for illustrative purposes only and are not intended to limit the invention. In the drawings: Figure 1 This is an exemplary system architecture diagram to which one embodiment of this disclosure can be applied; Figure 2A This is a flowchart of an embodiment of the workflow description file generation method disclosed herein; Figure 2B This is an exploded flowchart of one embodiment of step 201 of the present disclosure; Figure 3 This is a schematic diagram of a structure of an embodiment of the workflow description document generation apparatus according to the present disclosure; Figure 4 This is a schematic diagram of the structure of a computer system suitable for implementing embodiments of the present disclosure. Detailed Implementation

[0030] The present disclosure will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0031] It should be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other. This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0032] Figure 1 An exemplary system architecture 100 is shown, in which embodiments of the workflow description document generation method, apparatus, electronic device, and storage medium of this disclosure can be applied.

[0033] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0034] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as workflow description document generation applications, voice interaction applications, video conferencing applications, short video social applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0035] Terminal devices 101, 102, and 103 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with microphones and speakers, including but not limited to smartphones, tablets, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 players (Moving Picture Experts Group Audio Layer IV), portable computers, and desktop computers, etc. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices. They can be implemented as multiple software programs or software modules, or as a single software program or software module. No specific limitations are made here.

[0036] Server 105 can be a server that provides various services, such as a backend server that processes workflow definition information obtained from terminal devices 101, 102, and 103. The backend server can perform corresponding processing based on the workflow definition information.

[0037] In some cases, the workflow description file generation method provided in this disclosure can be jointly executed by terminal devices 101, 102, and 103 and server 105. For example, the step of "obtaining workflow definition information" can be executed by terminal devices 101, 102, and 103, and the step of "generating a workflow description file based on the workflow definition information" can be executed by server 105. This disclosure does not limit this. Correspondingly, the workflow description file generation apparatus can also be respectively set in terminal devices 101, 102, and 103 and server 105.

[0038] In some cases, the workflow description file generation method provided in this disclosure can be executed by server 105. Accordingly, the workflow description file generation device can also be set in server 105. In this case, system architecture 100 may not include terminal devices 101, 102, and 103.

[0039] In some cases, the workflow description file generation method provided in this disclosure can be executed by terminal devices 101, 102, and 103. Correspondingly, the workflow description file generation device can also be set in terminal devices 101, 102, and 103. In this case, the system architecture 100 may not include server 105.

[0040] It should be noted that server 105 can be either hardware or software. When server 105 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When server 105 is software, it can be implemented as multiple software programs or software modules (for example, used to provide distributed services), or as a single software program or software module. No specific limitations are made here.

[0041] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0042] Continue to refer to Figure 2A , Figure 2A A flow 200 is shown as an embodiment of the workflow description file generation method according to this disclosure. Figure 2A The workflow description file generation method shown can be applied to Figure 1 The terminal device or server shown. The process 200 includes the following steps: Step 201: Obtain workflow definition information.

[0043] In this embodiment, the execution body of the workflow description file generation method (e.g.) Figure 1 Server 105 can first obtain workflow definition information. Workflow definition information can be obtained by the executing entity from the user through a visual interactive interface. The visual interactive interface can include at least one of the following display objects: node icons for generating task node definitions in response to user operations, edge icons for generating connection edge sequences in response to user operations, a task node definition input area for generating task node definitions in response to user configuration operations, and an edge definition input area for generating connection edge definitions in response to user configuration operations. Users can manipulate node icons and edge icons in this visual interactive interface to generate task node definitions and connection edge sequences, and can also complete corresponding configuration operations in the task node definition input area and edge definition input area. The executing entity can respond to user interactions and configuration operations such as clicking, dragging, sliding, and inputting in the visual interactive interface to collect and integrate workflow definition information containing at least two task nodes, at least one node connection component, and the flow control structure, related definitions of each node and connection edge, etc.

[0044] Here, the workflow definition information includes at least two task nodes and at least one node connection component. The task nodes are used to execute specific tasks and have corresponding input constraint rules and output constraint rules.

[0045] A task node is a core unit in a workflow that undertakes specific calculations, processing, or functional execution. It can be used in data processing scenarios such as biomedicine to perform specific business tasks such as epitope analysis, AI sequence generation, and affinity prediction. Each task node is configured with corresponding input and output constraint rules to limit the structure and content characteristics of the received and output data. Task node types can include: AI model nodes, tool invocation nodes, and sub-workflow nodes. AI model nodes are used to invoke AI models, tool invocation nodes are used to invoke preset tools, and sub-workflow nodes are used to define sub-workflows and execute the specific tasks corresponding to those sub-workflows.

[0046] A node connection component is a dedicated adaptation unit in a workflow used to connect two adjacent task nodes. It is used to perform data adaptation and transmission operations between upstream and downstream task nodes, realize the format verification, syntax conversion and semantic alignment of data between upstream and downstream task nodes, so that the output data of the upstream task node can adapt to the input requirements of the downstream task node, and ensure the smooth flow and effective interaction of data between task nodes.

[0047] Input constraint rules and output constraint rules are used to limit the structural and content characteristics of data, and may include at least one of the following: data type constraint rules, data structure constraint rules, file format constraint rules, field integrity constraint rules, and enumeration value constraint rules.

[0048] As an example, data type constraint rules are rules that explicitly define the basic data types of input or output data of task nodes. They are used to standardize the basic data types that task nodes can receive and produce. For example, in the biomedical scenario, data is limited to integer molecular site numbers, string protein sequence identifiers, array-type epitope position sets, etc. This avoids node task execution failures caused by type mismatch at the basic type level and ensures the consistency of basic data types when data flows between nodes.

[0049] Data structure constraint rules are rules that limit the organization and hierarchical relationship of input or output data of task nodes. They are used to clarify the structural composition requirements of data. For example, in the biomedical scenario, antigen structure data is limited to an object-type structure that includes sequence resources and sequence type subfields, or table position data is limited to an integer array structure. This ensures that task nodes can parse and process data according to the preset structure and adapt to the data organization requirements of the internal computing logic of the task node.

[0050] File format constraint rules are rules that limit the storage format of file-type data input or output by task nodes. They are one of the core constraint rules in heterogeneous data processing in biomedicine. They are used to standardize the format of file data, such as limiting protein structure data to CIF and PDB formats, and protein sequence data to FASTA format. This solves the different requirements of different task nodes for file formats and avoids data unreadable problems caused by file format incompatibility.

[0051] Field integrity constraints are rules that limit the existence and validity of required and core fields in the structured data input or output of task nodes. They are used to ensure the integrity of the core information of structured data. For example, in the biomedical scenario, the input data of the epitope analysis node must include sequence resources and sequence type fields, and the output data must include antigen structure and epitope location fields to prevent the node's calculation logic from being interrupted or invalid data from being produced due to the absence of core fields.

[0052] Enumerated value constraint rules are rules that define the range of values ​​for specific fields in the input or output data of task nodes. They are used to standardize the legal values ​​of fields. For example, in the biomedical scenario, the value of the sequence type field is limited to "protein", and the value of the data status field is limited to "valid" or "invalid". This ensures that the field values ​​meet the expected calculation logic of the node, avoids calculation errors caused by illegal values, and achieves precise limitation of business semantics.

[0053] By setting input and output constraint rules for each task node, the structure and content characteristics of data flowing between nodes in the workflow are precisely defined in all dimensions and with strong standards. This standardizes and constrains heterogeneous data from the source of data flow, effectively avoiding situations where task node execution fails, computation logic is interrupted, or invalid data is produced due to issues such as data type mismatch, inconsistent structure, incompatible format, missing core fields, or illegal field values. This ensures the consistency and effectiveness of data flow between nodes in heterogeneous data-intensive scenarios such as biomedicine.

[0054] In some optional implementations, the workflow definition information may also include node definitions for each task node. The node definition may include a node identifier for uniquely identifying the task node, a node type identifier for limiting the type of the task node, and functional description information for describing the specific task functions performed by the task node.

[0055] Here, node type identifiers are used to distinguish the category and execution function type of different task nodes. They clarify the role and processing type of the current task node in the workflow, facilitating the execution engine's identification and invocation of the corresponding processing logic. Node type identifiers can include data reading nodes, data cleaning nodes, format conversion nodes, sequence alignment nodes, and antigen epitope prediction nodes. Specifically, data reading nodes are task nodes used to retrieve raw biomedical data from local files, databases, or interfaces. Data cleaning nodes are task nodes used to preprocess raw data, such as denoising, deduplication, and missing value imputation, to ensure data quality. Format conversion nodes are task nodes used to convert data between different storage and representation formats to adapt to subsequent task node processing. Sequence alignment nodes are task nodes used to compare the biological sequence to be tested with a reference sequence to identify homologous regions and differentially expressed sites. Antigen epitope prediction nodes are task nodes used to analyze and predict epitope regions on antigen molecules based on sequence or structural features.

[0056] Functional description information provides a detailed description of the specific execution content, processing objects, and achievement goals of a task node, clearly defining the business logic and processing scope of that task node. Functional description information may include the processing method for input data, the algorithm or model used, the data output format, and the analysis objectives for biomedical data.

[0057] By configuring complete node definitions, including node identifiers, node type identifiers, and functional descriptions, for each task node in the workflow definition information, unique identification, standardized classification, and clear functional definition of task nodes are achieved. This facilitates the workflow engine to quickly identify, schedule, and execute the corresponding node logic, while also improving the readability, maintainability, and scalability of the workflow, thereby ensuring the stable and orderly operation of complex biomedical data processing workflows.

[0058] The node connection component is used to connect two adjacent task nodes and to perform data adaptation and transmission operations, which may include, for example... Figure 2B The following steps 2011 to 2015 are shown: Step 2011: Obtain the upstream output data and its output constraint rules from the upstream task node.

[0059] Here, the executing entity retrieves the upstream output data after the upstream task node is completed by parsing the pre-configured upstream task node identifier and data association relationship in the workflow definition information. At the same time, it extracts the preset output constraint rules from the standardized definition information of the upstream task node. These output constraint rules include various limiting requirements such as the data type, data structure, and file format of the upstream output data, so as to achieve accurate and synchronous acquisition of the upstream output data and the corresponding output constraint rules.

[0060] By acquiring the upstream output data and its output constraint rules from the upstream task nodes, the retrieval of upstream output data and rules is automated and standardized, eliminating the need for manual intervention in the screening and transmission of data and rules. This ensures that the acquired upstream output data and rules are highly consistent with the actual execution results and preset requirements of the upstream task nodes.

[0061] Step 2012: Obtain the input constraint rules of the downstream task nodes.

[0062] Here, the executing entity locates the downstream task node associated with the current node connection component based on the node connection relationship established in the workflow definition information, and directly extracts its pre-configured input constraint rules from the standardized definition information of the downstream task node. These input constraint rules contain all the restrictions imposed by the downstream task node on the input data in terms of data type, data structure, file format, field integrity, enumeration values, etc., so as to achieve accurate and complete acquisition of the downstream input constraint rules.

[0063] By extracting input constraint rules from the preset definitions of downstream task nodes based on node connection relationships, the data reception requirements of downstream task nodes are automatically and standardizedly obtained without the need for manual configuration or transmission of rule parameters. This ensures that the obtained input constraint rules are highly matched with the actual execution logic of downstream task nodes, avoiding errors in adaptation and verification results caused by incomplete or biased rule acquisition.

[0064] Step 2013: Perform matching verification based on output constraint rules and input constraint rules.

[0065] In practice, the output constraint rules of the upstream task node and the input constraint rules of the downstream task node are checked for consistency in a dimension-by-dimensional manner. The data type, data structure, file format, field integrity, enumeration value and semantic constraint requirements are compared in turn to determine whether the constraint attributes of the upstream output data fully meet the input constraint requirements of the downstream task node. If all dimensions match, the check is considered to have passed. If any dimension does not match, the check is considered to have failed. The specific mismatched dimensions and contents can be recorded.

[0066] By performing multi-dimensional and sequential systematic matching and verification of the constraint rules of upstream and downstream task nodes, the system achieves accurate and comprehensive judgment of data flow adaptability. It can quickly locate data mismatch issues in terms of format, structure, and semantics, thus providing a clear basis for targeted data adaptation and conversion operations. This avoids the failure of downstream task nodes due to the direct transmission of mismatched data, ensuring the targeting and effectiveness of data adaptation and transmission operations.

[0067] Step 2014: When the matching does not meet the input constraint rules, perform data adaptation calculation on the upstream output data based on the predefined data transformation rules to generate downstream input data that meets the input constraint rules, and input the downstream input data to the downstream task node.

[0068] Here, predefined data transformation rules refer to a set of standardized rules set in advance to achieve multi-dimensional adaptation and transformation of data format, structure, semantics, etc., in response to the differences in data constraint rules between different task nodes in the workflow. Predefined data transformation rules are used to transform upstream output data into data that meets input constraint rules. Data transformation rules may include at least one of the following: data type transformation rules, data structure transformation rules, file format transformation rules, field integrity supplementation rules, and enumeration value mapping rules.

[0069] As an example, data type conversion rules refer to standardized conversion rules set for the problem of mismatch between the basic data types of upstream and downstream task nodes. They are used to convert the basic data types of upstream output data (such as integers, strings, arrays, booleans, etc.) into the target data types required by the input constraint rules of downstream task nodes. For example, converting the string number of the epitope position in the biomedical scenario into an integer number solves the problem that downstream task nodes cannot parse data due to data type mismatch.

[0070] Data structure transformation rules refer to standardized transformation rules set to address the mismatch between the data organization forms and hierarchical relationships of upstream and downstream task nodes. These rules are used to reorganize, split, and nest the structured components of upstream output data to make them conform to the structural form required by the input constraint rules of downstream task nodes. For example, the flat antigen data structure in the biomedical scenario can be transformed into an object-type structure containing subfields to ensure that downstream task nodes can parse the various components of the data according to the preset logic.

[0071] File format conversion rules are the core conversion rules in heterogeneous data processing in biomedicine. They are standardized conversion rules designed to address the problem of mismatched file data storage formats between upstream and downstream task nodes. These rules are used to convert the file data output by upstream task nodes from its original format to the target file format required by the input constraints of downstream task nodes. For example, they can convert CIF format files of protein structures in biomedicine scenarios to PDB format files, and FASTA format protein sequence files to TXT format files, thus solving the problem of downstream task nodes being unable to read data due to file format incompatibility.

[0072] Field integrity supplementation rules refer to standardized supplementation rules set for the problem of missing core fields required by the input constraint rules of downstream task nodes in upstream output data. They are used to supplement missing core fields in upstream output data and improve the validity of field information according to preset logic. For example, in the biomedical scenario, the core field "antigen source" required by the downstream skeleton design node is supplemented to the epitope analysis output data to prevent the interruption of the calculation logic of downstream task nodes due to the absence of core fields.

[0073] Enumeration value mapping rules refer to standardized mapping rules set up to address the mismatch in the enumeration value range of the same business field between upstream and downstream task nodes. They are used to convert the original enumeration values ​​of specific fields in the upstream output data into valid enumeration values ​​required by the input constraint rules of the downstream task nodes according to the preset mapping relationship. For example, in the biomedical scenario, the "Y / N" value of the data status field is mapped to the "valid / invalid" value, and the "0 / 1" value is mapped to the "mutable / non-mutable" value, ensuring that the field values ​​meet the expected calculation logic of the downstream task nodes.

[0074] The data transformation rules also include semantic mapping rules. Semantic mapping rules are standardized transformation rules set for the problem that the data between upstream and downstream task nodes match in syntax but do not match in business semantic expression. They are used to convert the first semantic representation in the upstream output data into a second semantic representation that meets the input constraint rules. Here, the first semantic representation and the second semantic representation differ in business meaning and correspond to different ways of expressing the same object.

[0075] When the data constraint rules of upstream and downstream task nodes do not match, targeted adaptation calculations are performed on the upstream output data based on predefined data transformation rules to generate downstream input data that conforms to the downstream input constraint rules. This achieves automated, standardized format conversion and semantic alignment of heterogeneous data without the need for manual intervention in data processing and adaptation operations. This effectively solves the core problem of data type, structure, format, and semantic incompatibility between nodes in scenarios such as biomedicine. It allows heterogeneous data that could not be directly transferred to accurately adapt to the input requirements of downstream task nodes, ensuring the normal execution of downstream task nodes and the integrity of their computational logic. At the same time, relying on predefined transformation rules, the efficiency and consistency of data adaptation are achieved, which greatly improves the smoothness of data flow in the workflow and the reliability of automated workflow execution.

[0076] Semantic mapping rules are used to transform between at least one of the following: mutable attributes and immutable attributes, a first set of data labels and a second set of data labels, and a first set of constraints and a second set of constraints. Here, the transformation between mutable and immutable attributes refers to mapping the semantic differences in attribute constraints on data objects in scenarios such as biomedicine to mutable attributes in upstream output data that represent the data object's modifiability / adjustability, into immutable attributes that downstream task nodes require the data object to be prohibited from modification / adjustment. Alternatively, the reverse transformation can be performed, such as converting the mutable amino acid site attribute output by the upstream skeleton design node into the non-mutable amino acid site attribute required by the downstream AI sequence generation node, thus achieving semantic alignment between the data object's operation permissions and constraint requirements.

[0077] The conversion between the first data tag set and the second data tag set refers to addressing the semantic differences in the annotation tags and naming identifiers of the same type of data by upstream and downstream task nodes. The first data tag set (containing one or more data tags) to which the upstream output data belongs is converted into the second data tag set required by the input constraint rules of the downstream task nodes according to a preset mapping relationship. For example, in the biomedical scenario, the "antigen epitope region" and "protein binding site" tag sets annotated by the upstream task nodes are converted into the English tag sets "Epitope" and "Binding Site" uniformly used by the downstream task nodes, or the "molecular ID" and "site number" tags are converted into standardized tags "Mol_ID" and "Pos_No", so as to achieve semantic uniformity of the data annotation system.

[0078] The conversion between the first set of constraints and the second set of constraints refers to the semantic expression or rule dimension difference between the business constraints of upstream and downstream task nodes in data screening and calculation. The first set of constraints (including one or more business constraint rules) attached to the upstream output data is mapped to the second set of constraints that can be recognized by the downstream task nodes when performing calculations. For example, in the biomedical scenario, the constraint set of "affinity > 500" and "molecular weight < 800 Da" of the upstream task nodes is converted into the constraint set of "binding strength score ≥ 500" and "molecular mass ≤ 800 Daltons" required by the downstream task nodes.

[0079] Semantic mapping rules are determined based on at least one of the following: a predefined semantic mapping table, a rule engine, a machine learning model, and manually configured mapping relationships.

[0080] Here, the predefined semantic mapping table is a pre-configured standardized semantic correspondence table that directly stores semantic mapping key-value pairs of upstream and downstream data, enabling rapid matching and conversion in simple semantic scenarios. The rule engine can be configured with multi-dimensional semantic conversion logic and triggering conditions to adapt to the combined semantic mapping needs in complex business scenarios. The machine learning model is trained based on heterogeneous data samples in the biomedical field, enabling it to autonomously learn and identify semantic relationships in data, achieving intelligent mapping in non-standardized semantic scenarios. Manually configured mapping relationships are flexibly defined by business personnel according to actual R&D needs, adapting to personalized and customized semantic conversion scenarios. Multiple determination methods can be combined as needed to meet the semantic alignment needs of different complexities and types in biomedical data processing.

[0081] In practice, when the output constraint rules and the input constraint rules do not match at the semantic level, the semantic mapping rule is triggered. The semantic mapping rule is used in biomedical data processing scenarios to perform semantic alignment transformation on data related to molecular structure, antigen epitopes or sequence constraints.

[0082] In the context of biomedical data processing, after triggering a semantic mapping rule, the system first identifies the specific data types where upstream and downstream semantics do not match, determining whether they are related to molecular structure, antigenic epitopes, or sequence constraints. Then, the corresponding semantic mapping rule is invoked. For example, for molecular structure-related data, the semantic representation of molecular spatial structure in upstream formats such as CIF / PDB is converted into the pre-defined structural semantic definitions of downstream task nodes. For antigenic epitope-related data, the semantic information such as epitope site numbers and region definitions annotated upstream is adapted into an epitope semantic representation form recognizable by downstream task nodes. For sequence constraint-related data, the semantic constraints such as mutagenic / non-mutagenic sites and sequence modification requirements output upstream are converted into the corresponding semantic constraint rules for downstream task nodes when performing sequence generation and optimization.

[0083] By setting semantic mapping rules for workflow data adaptation and transmission operations, precise semantic alignment is achieved for data between upstream and downstream task nodes that only match in syntax but have inconsistent business semantic expressions. This breaks through the limitations of traditional data conversion that only focuses on format and structure adjustment, thus effectively solving the semantic incompatibility problem of heterogeneous data in professional fields such as biomedicine. It allows different semantic representations of the same business object to flow smoothly and be accurately parsed and calculated by downstream task nodes, avoiding errors in node calculation results or execution failures due to semantic deviations, and ensuring the accuracy and effectiveness of automated execution of professional workflows.

[0084] Step 2015: When the input constraint rules are met, the upstream output data is directly input to the downstream task node.

[0085] In practice, after the node connection component completes the full-dimensional matching and verification of the output constraint rules of the upstream task node and the input constraint rules of the downstream task node, if it determines that the two match in all dimensions such as data type, data structure, file format, field integrity, enumeration value and business semantics and satisfy the downstream input constraint rules, it will trigger the direct transmission execution logic. Through the data transmission channel built into the workflow, it will directly retrieve the upstream output data output after the upstream task node has been executed, perform data format adaptation encapsulation according to the input data reception specification of the downstream task node, and push the encapsulated upstream output data directly to the input interface of the downstream task node.

[0086] By directly inputting upstream output data into downstream task nodes when matching constraint rules across all dimensions, the workflow achieves data flow without redundancy and with high timeliness. This significantly improves the data transmission efficiency between nodes and the overall execution speed of the workflow, reduces the risk of data loss and errors caused by additional data operations, ensures the accuracy and integrity of data transmission, and further enhances the overall performance and stability of automated workflow execution in biomedical data processing scenarios.

[0087] In some optional implementations, the workflow definition information may also include a flow control structure. The flow control structure may include at least one of the following: a parallel structure, a conditional branching structure, and a loop structure. Wherein: A loop structure refers to a workflow structure with repetitive execution logic. It's used to re-execute at least some task nodes when the target evaluation metric fails to meet preset evaluation metric conditions, until the preset conditions are met. In practice, when the loop structure is working, after completing some task nodes, the executing entity obtains the corresponding target evaluation metric and compares it with the preset evaluation metric conditions. If the comparison result is unsatisfactory, the loop execution mechanism is triggered, jumping back to the starting position of the specified task node, re-executing that part of the task node, and recalculating the target evaluation metric. This execution and comparison process is repeated until the target evaluation metric meets the preset conditions, at which point the loop exits.

[0088] Parallel structure refers to a structure in a workflow that supports the synchronous execution of multiple tasks, used to execute at least two task nodes in parallel.

[0089] When operating in a parallel architecture, after entering a parallel branch node, the execution entity simultaneously creates multiple independent execution threads or task instances, assigning at least two task nodes to different execution units to start running synchronously. Each task node independently executes its corresponding processing logic without interfering with others or waiting for other task nodes to complete. After all parallel task nodes have completed execution, the execution entity summarizes and integrates the output results of each task node before entering a unified execution flow.

[0090] Conditional branching structure refers to a structure in a workflow that selects the execution path based on the judgment result. It is used to execute the task node corresponding to the preset branch condition when the preset branch condition is met.

[0091] When the conditional branch structure is working, the executing entity first obtains the judgment data on which the preset branch conditions depend, performs condition judgment based on the judgment data, selects the matching target branch from multiple preset branch paths according to the judgment result, automatically jumps to and executes the task node corresponding to the target branch, and skips other branches that do not meet the conditions directly without executing their corresponding task logic. After the target branch is completed, the subsequent workflow continues to advance.

[0092] By introducing process control structures such as parallel structures, conditional branching structures, and loop structures into the workflow definition information, flexible scheduling and fine-grained control of workflow execution logic are achieved. It can simultaneously execute multiple tasks according to actual business needs, dynamically select execution paths according to conditions, or cyclically optimize some processing processes based on evaluation indicators. This significantly improves the adaptability of the workflow to complex biomedical data processing scenarios and enhances the automation level and processing efficiency of workflow execution.

[0093] In some optional implementations, the node connection component may include a sequence of connection edges consisting of at least one connection edge. The output data of the upstream task node is transmitted as input data to the first connection edge in the sequence of connection edges, and the output data of the last connection edge in the sequence of connection edges is transmitted as input data to the downstream task node. The output data of the upstream connection edge in two adjacent connection edges is transmitted as input data to the downstream connection edge. Each connection edge is used to perform data transformation processing on the input data.

[0094] Here, the connection edge sequence refers to an ordered combination consisting of at least one connection edge connected sequentially in the order of data flow. It is used to form a multi-level data transmission and processing channel between upstream and downstream task nodes, and to perform step-by-step transformation and processing on the data transmitted from upstream, and finally output input data that meets the requirements of downstream task nodes.

[0095] By using a sequence of connecting edges composed of multiple connecting edges to realize data transmission between nodes, it is possible to perform hierarchical and step-by-step transformation processing of data, adapt to the data adaptation needs of multiple stages and multiple types in complex scenarios, avoid the logic of a single transformation being too complex, improve the flexibility and stability of data processing, and make the data flow logic of the workflow clearer.

[0096] In some optional implementations, the workflow definition information may also include the connection edge definition of each connection edge. The connection edge definition package may include an edge identifier for uniquely identifying the connection edge, an adapter type identifier for limiting the connection edge to a data conversion adapter, functional description information for describing the function of the connection edge, input constraint rules for limiting the data format to be converted, and output constraint rules for limiting the converted data format, etc.

[0097] Here, the connection edge definition is used to standardize and structure the description of each connection edge in the workflow, providing the workflow execution engine with a complete configuration basis for data flow and data transformation, and ensuring that the identifier, type, function and constraint rules of each connection edge can be accurately parsed and executed.

[0098] Edge identifiers are identification information used to uniquely distinguish different connected edges, enabling independent location, retrieval, and management of each connected edge.

[0099] A data conversion adapter is a data processing unit integrated into a connection edge. It is used to perform adaptation operations such as format conversion, semantic mapping, and constraint verification on the data passing through the connection edge, so that the data meets the transmission requirements of upstream and downstream task nodes.

[0100] The functional description information of the connecting edge describes the specific data processing behavior performed by the connecting edge, including data format conversion, data cleaning, data filtering, semantic mapping, constraint verification, and numerical normalization.

[0101] By configuring a complete connection edge definition for each connection edge, a standardized and refined description of the data transformation logic between task nodes is achieved. This enables the workflow engine to accurately identify and execute the corresponding data adaptation processing, which not only improves the configurability and maintainability of the workflow, but also provides a reliable guarantee for the stable execution of complex biomedical data processing procedures.

[0102] Step 202: Generate a workflow description file based on the workflow definition information.

[0103] Here, a workflow description file refers to an executable file formed by structurally encapsulating workflow definition information according to a preset format, which is used by the execution engine to parse and execute the workflow.

[0104] When generating the workflow description file, the workflow definition information is read first, and then structured encoding and splicing are performed according to the preset file format specifications to form a complete workflow description file that can be directly parsed by the engine.

[0105] By generating standardized workflow description files, workflow configuration information can be expressed in a standardized and structured manner, which facilitates efficient parsing and loading by the execution engine and ensures the stable and reliable automated execution of complex data processing flows.

[0106] In some alternative implementations, after step 202 is performed, step 203 may also be performed: Step 203: Invoke the execution engine to execute the executable code corresponding to the workflow description file in order to execute the workflow.

[0107] Here, when executing a workflow, the data input to the workflow is used as the input data for the first task node in the workflow. The task is executed according to the workflow's process logic, and the data output by the last task node in the workflow is used as the output data of the workflow.

[0108] Executable code refers to the instruction code that can be directly recognized and run, generated by the execution engine based on the parsing, compilation, or transformation of the workflow description file. It is used to drive each task node in the workflow to execute sequentially according to the preset process logic, realizing the complete workflow execution logic such as data flow, data processing, process control, and task scheduling.

[0109] As an example, when executing a workflow, the execution engine takes the business data from the external input workflow as the input data for the first task node in the workflow. Following the node execution order, connection relationships, and flow control logic defined in the workflow description file, it sequentially schedules and executes the data processing operations of each task node and its connections. After all task nodes have been executed, the processing result data output by the last task node in the workflow is used as the final output data for the entire workflow.

[0110] By calling the execution engine to run executable code to automate workflows, the entire process from workflow configuration to actual business processing is automated, eliminating the need for manual intervention in node scheduling and data transmission. This significantly improves the execution efficiency and intelligence level of complex data processing scenarios such as biomedicine.

[0111] In some alternative implementations, after step 203 is performed, step 204 may also be performed: Step 204: In the visual interactive interface, in response to the detection of a new connection edge operation, a matching and verification is performed based on the output constraint rules of the upstream task node and the input constraint rules of the downstream task node of the new connection edge.

[0112] Here, when the matching verification passes, the newly added connecting edge will be displayed in the first visual style; when the matching verification fails, the newly added connecting edge will be displayed in the second visual style.

[0113] As an example, the first visual style and the second visual style are different display formats used in the visual interactive interface to distinguish the results of the connection edge verification. When the matching verification passes, the first visual style can be displayed as a normal solid line, a green connecting line, and standard brightness. When the matching verification fails, the second visual style can be displayed as a red dashed line, a gray semi-transparent line, a bright warning color, or a flashing prompt, which are significantly different from the normal state.

[0114] By using different visual styles to distinguish between newly added connection edges that pass and fail validation in the visual interactive interface, users can intuitively and quickly identify the constraint matching status of connection edges, promptly discover and locate incompatibility issues between data rules between nodes, effectively improve the intuitiveness and efficiency of workflow editing and debugging, reduce the probability of errors in complex workflow configuration, and enhance the user experience.

[0115] The workflow description file generation method provided in the above embodiments of this disclosure first obtains workflow definition information, and finally generates a workflow description file based on the workflow definition information. In this way, by configuring connection edges and introducing constraint rule matching and verification during workflow definition and execution, the connection edges are upgraded from simple connection relationships to intelligent transformation with independent data adaptation, semantic alignment and real-time verification capabilities. This completely solves the problems of incompatible data formats and semantic inconsistencies between upstream and downstream task nodes, and greatly improves the flexibility of workflow configuration, debugging efficiency and execution stability.

[0116] Further reference Figure 3 As an implementation of the methods shown in the above figures, this disclosure provides an embodiment of a workflow description document generation apparatus, which is similar to... Figure 2A Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0117] like Figure 3As shown, the workflow description file generation device 300 of this embodiment includes a definition information acquisition unit 301 and a generation unit 302. The definition information acquisition unit 301 is used to acquire workflow definition information, which includes at least two task nodes and at least one node connection component. The task nodes are used to execute specific tasks and have corresponding input constraint rules and output constraint rules. The node connection component is used to connect two adjacent task nodes and to perform the following data adaptation and transmission operations: acquiring upstream output data and its output constraint rules from the upstream task node; acquiring the input constraint rules of the downstream task node; performing matching verification based on the output constraint rules and input constraint rules; when the matching does not meet the input constraint rules, performing data adaptation calculation on the upstream output data based on predefined data transformation rules to generate downstream input data that meets the input constraint rules, and inputting the downstream input data to the downstream task node; when the matching meets the input constraint rules, directly inputting the upstream output data to the downstream task node. The generation unit 302 is used to generate a workflow description file based on the workflow definition information. The workflow description file is used by the execution engine to parse and execute the workflow.

[0118] In this embodiment, the specific processing of the definition information acquisition unit 301 and the generation unit 302 of the workflow description document generation device 300 and the resulting technical effects can be referred to respectively. Figure 2A The relevant descriptions of steps 201 and 202 in the corresponding embodiments will not be repeated here.

[0119] In some optional implementations, input constraint rules and output constraint rules are used to limit the structural and content characteristics of the data, and include at least one of the following: data type constraint rules, data structure constraint rules, file format constraint rules, field integrity constraint rules, and enumeration value constraint rules; preferably, predefined data transformation rules are used to transform upstream output data into data that satisfies the input constraint rules, and the data transformation rules include at least one of the following: data type transformation rules, data structure transformation rules, file format transformation rules, field integrity supplementation rules, and enumeration value mapping rules; preferably, the data transformation rules also include semantic mapping rules, used to transform a first semantic representation in the upstream output data into a second semantic representation that satisfies the input constraint rules; preferably, the task node types include: artificial intelligence model nodes, tool call nodes, and sub-workflow nodes, wherein the artificial intelligence model nodes are used to call artificial intelligence models, the tool call nodes are used to call preset tools, and the sub-workflow nodes are used to define sub-workflows and execute specific tasks corresponding to the sub-workflows.

[0120] In some optional implementations, the first semantic representation and the second semantic representation differ in business meaning and correspond to different ways of expressing the same object; preferably, the semantic mapping rule is used to convert between at least one of the following: variable attributes and immutable attributes, a first data label set and a second data label set, a first constraint set and a second constraint set; preferably, the semantic mapping rule is determined based on at least one of the following: a predefined semantic mapping table, a rule engine, a machine learning model, and a manually configured mapping relationship; preferably, the semantic mapping rule is triggered when the output constraint rule and the input constraint rule do not match at the semantic level; preferably, the semantic mapping rule is used in biomedical data processing scenarios to perform semantic alignment conversion on data related to molecular structure, antigen epitopes, or sequence constraints.

[0121] In some optional implementations, the workflow definition information further includes a process control structure, which includes at least one of the following: a parallel structure, a conditional branching structure, and a loop structure; preferably, the loop structure is used to re-execute at least some task nodes until the preset evaluation index conditions are met when the target evaluation index does not meet the preset evaluation index conditions; preferably, the parallel structure is used to execute at least two task nodes in parallel; preferably, the conditional branching structure is used to execute the task node corresponding to the corresponding branch condition when the preset branch condition is met; preferably, the workflow definition information also includes a node definition for each task node, which includes a node identifier for uniquely identifying the task node, a node type identifier for limiting the type of the task node, and functional description information for describing the specific task function performed by the task node; preferably... The node connection component includes a sequence of connection edges consisting of at least one connection edge. The output data of the upstream task node is transmitted as input data to the first connection edge in the connection edge sequence, and the output data of the last connection edge in the connection edge sequence is input data to the downstream task node. The output data of the upstream connection edge in two adjacent connection edges is input data to the downstream connection edge. Each connection edge is used to perform data transformation processing on the input data. Preferably, the workflow definition information also includes the connection edge definition of each connection edge. The connection edge definition includes an edge identifier for uniquely identifying the connection edge, an adapter type identifier for limiting the connection edge as a data transformation adapter, functional description information for describing the function of the connection edge, input constraint rules for limiting the format of the data to be transformed, and output constraint rules for limiting the format of the transformed data.

[0122] In some optional implementations, the definition information acquisition unit 301 may be further configured as follows: Obtain workflow definition information generated by the user through a visual interactive interface. The visual interactive interface includes at least one of the following display objects: a node icon for generating task node definitions in response to user operations, an edge icon for generating connection edge sequences in response to user operations, a task node definition input area for generating task node definitions in response to user configuration operations, and an edge definition input area for generating connection edge definitions in response to user configuration operations. The device also includes: The execution engine invocation unit (not shown in the figure) is used to invoke the execution engine to execute the executable code corresponding to the workflow description file in order to execute the workflow. When executing the workflow, the data of the input workflow is used as the input data of the first task node in the workflow. The task is executed according to the workflow flow logic, and the data output by the last task node in the workflow is used as the output data of the workflow.

[0123] In some alternative embodiments, the apparatus further includes: The matching and verification unit (not shown in the figure) is used in the visual interactive interface to perform matching and verification based on the output constraint rules of the upstream task node and the input constraint rules of the downstream task node of the newly added connection edge in response to the detection of the newly added connection edge operation. When the matching and verification pass, the newly added connection edge is displayed in the first visual style; when the matching and verification fail, the newly added connection edge is displayed in the second visual style.

[0124] It should be noted that the implementation details and technical effects of each module and unit in the workflow description document generation device provided in the embodiments of this disclosure can be referred to the descriptions of other embodiments in this disclosure, and will not be repeated here.

[0125] The following is for reference. Figure 4 It shows a schematic diagram of the structure of a computer system 400 suitable for implementing the terminal device of this disclosure. Figure 4 The computer system 400 shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0126] like Figure 4 As shown, the computer system 400 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage device 408 into a random access memory (RAM) 403. The RAM 403 also stores various programs and data required for the operation of the computer system 400. The processing device 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0127] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 408 including, for example, magnetic tapes, hard disks, etc.; and communication devices 409. Communication device 409 allows computer system 400 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 A computer system 400 with various electronic devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0128] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 409, or installed from storage device 408, or installed from ROM 402. When the computer program is executed by processing device 401, it performs the functions defined in the methods of embodiments of this disclosure.

[0129] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0130] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0131] The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the following functions: Figure 2A The embodiments shown and their alternative implementations illustrate a method for generating workflow description files.

[0132] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0133] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0134] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The names of the units are not necessarily limiting in certain circumstances; for example, a definition information acquisition unit can also be described as a "unit for acquiring workflow definition information".

[0135] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A method for generating a workflow description file, characterized in that, The method is performed by an electronic device, and the method includes: Obtain workflow definition information, which includes: at least two task nodes and at least one node connection component, wherein: The task node is used to execute specific tasks and has corresponding input constraint rules and output constraint rules; The node connection component is used to connect two adjacent task nodes and to perform the following data adaptation and transmission operations: acquiring upstream output data and its output constraint rules from the upstream task node; acquiring input constraint rules from the downstream task node; performing matching verification based on the output constraint rules and the input constraint rules; when the matching does not satisfy the input constraint rules, performing data adaptation calculation on the upstream output data based on predefined data transformation rules to generate downstream input data that satisfies the input constraint rules, and inputting the downstream input data to the downstream task node; when the matching satisfies the input constraint rules, directly inputting the upstream output data to the downstream task node. A workflow description file is generated based on the workflow definition information. The workflow description file is used by the execution engine to parse and execute the workflow.

2. The method according to claim 1, characterized in that, The input constraint rules and output constraint rules are used to limit the structural and content characteristics of the data, and include at least one of the following: data type constraint rules, data structure constraint rules, file format constraint rules, field integrity constraint rules, and enumeration value constraint rules; preferably, the predefined data transformation rules are used to transform the upstream output data into data that satisfies the input constraint rules, and the data transformation rules include at least one of the following: data type transformation rules, data structure transformation rules, file format transformation rules, field integrity supplementation rules, and enumeration value mapping rules; preferably, the data transformation rules also include semantic mapping rules, used to transform the first semantic representation in the upstream output data into a second semantic representation that satisfies the input constraint rules; preferably, the task node types include: artificial intelligence model nodes, tool call nodes, and sub-workflow nodes, wherein the artificial intelligence model nodes are used to call artificial intelligence models, the tool call nodes are used to call preset tools, and the sub-workflow nodes are used to define sub-workflows and execute specific tasks corresponding to the sub-workflows.

3. The method according to claim 2, characterized in that, The first semantic representation and the second semantic representation differ in business meaning and correspond to different ways of expressing the same object; preferably, the semantic mapping rule is used to convert between at least one of the following: variable attributes and immutable attributes, a first data label set and a second data label set, a first constraint set and a second constraint set; preferably, the semantic mapping rule is determined based on at least one of the following: a predefined semantic mapping table, a rule engine, a machine learning model, and a manually configured mapping relationship; preferably, the semantic mapping rule is triggered when the output constraint rule and the input constraint rule do not match at the semantic level; preferably, the semantic mapping rule is used in biomedical data processing scenarios to perform semantic alignment conversion on data related to molecular structure, antigen epitopes, or sequence constraints.

4. The method according to claim 1, characterized in that, The workflow definition information also includes a process control structure, which includes at least one of the following: a parallel structure, a conditional branching structure, and a loop structure; preferably, the loop structure is used to re-execute at least some task nodes when the target evaluation indicator does not meet the preset evaluation indicator conditions, until the preset evaluation indicator conditions are met; preferably, the parallel structure is used to execute at least two task nodes in parallel; preferably, the conditional branching structure is used to execute the task node corresponding to the corresponding branch condition when the preset branch condition is met; preferably, the workflow definition information also includes a node definition for each task node, the node definition including a node identifier for uniquely identifying the task node, a node type identifier for limiting the type of the task node, and functional description information for describing the specific task function performed by the task node; preferably, the node... The point connection component includes a sequence of connection edges consisting of at least one connection edge. The output data of the upstream task node is transmitted as input data to the first connection edge in the sequence, and the output data of the last connection edge in the sequence is input data to the downstream task node. The output data of the upstream connection edge in two adjacent connection edges is input data to the downstream connection edge. Each connection edge is used to perform data transformation processing on the input data. Preferably, the workflow definition information also includes the connection edge definition of each connection edge. The connection edge definition includes an edge identifier for uniquely identifying the connection edge, an adapter type identifier for limiting the connection edge as a data transformation adapter, functional description information for describing the function of the connection edge, input constraint rules for limiting the format of the data to be transformed, and output constraint rules for limiting the format of the transformed data.

5. The method according to claim 1, characterized in that, The acquisition of workflow definition information includes: Obtain workflow definition information generated by the user through a visual interactive interface, wherein the visual interactive interface includes at least one of the following display objects: a node icon for generating the task node definition in response to user operation, an edge icon for generating the connection edge sequence in response to user operation, a task node definition input area for generating the task node definition in response to user configuration operation, and an edge definition input area for generating the connection edge definition in response to user configuration operation. The method further includes: The execution engine is invoked to execute the executable code corresponding to the workflow description file to execute the workflow. When executing the workflow, the data input to the workflow is used as the input data of the first task node in the workflow. The task is executed according to the flow logic of the workflow, and the data output by the last task node in the workflow is used as the output data of the workflow.

6. The method according to claim 5, characterized in that, The method further includes: In the visual interactive interface, in response to the detection of a new connection edge, a matching verification is performed based on the output constraint rules of the upstream task node and the input constraint rules of the downstream task node of the new connection edge; when the matching verification passes, the new connection edge is displayed in the first visual style; when the matching verification fails, the new connection edge is displayed in the second visual style.

7. A workflow description document generation apparatus, the apparatus being executed by an electronic device, the apparatus comprising: The definition information acquisition unit is used to acquire workflow definition information, which includes: at least two task nodes and at least one node connection component, wherein: The task node is used to execute specific tasks and has corresponding input constraint rules and output constraint rules; The node connection component is used to connect two adjacent task nodes and to perform the following data adaptation and transmission operations: acquiring upstream output data and its output constraint rules from the upstream task node; acquiring input constraint rules from the downstream task node; performing matching verification based on the output constraint rules and the input constraint rules; when the matching does not satisfy the input constraint rules, performing data adaptation calculation on the upstream output data based on predefined data transformation rules to generate downstream input data that satisfies the input constraint rules, and inputting the downstream input data to the downstream task node; when the matching satisfies the input constraint rules, directly inputting the upstream output data to the downstream task node. The generation unit is used to generate a workflow description file based on the workflow definition information. The workflow description file is used to be parsed and executed by the execution engine.

8. An electronic device, comprising: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method as described in any one of claims 1-6.

9. A computer-readable medium having a computer program stored thereon, wherein, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

10. A computer program product comprising a computer program / instructions that, when executed by a processor, implement the method as described in any one of claims 1-6.

Images