Workflow operator recommendation method and device, equipment, storage medium and product

Abstract

The invention discloses a workflow operator recommendation method and device, equipment, a storage medium and a product, and relates to the technical field of data processing, and the workflow operator recommendation method comprises the steps: obtaining a precursor operator selected by a user based on a current workflow task chain; acquiring a subsequent operator demand input by the user based on the precursor operator; retrieving subsequent operator requirements based on the operator knowledge base to obtain an operator candidate set; and sorting the operator candidate set through a resorting model, and recommending the sorted target operator set to a user. The operators are gradually determined in the workflow in the form of interaction with the user to generate the workflow, and the system recommends the appropriate operators in each step according to the requirements of the user and the context information of the current path, so that the flexibility and the controllability can be greatly improved, and the use threshold of the user is reduced.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a workflow operator recommendation method, device, equipment, storage medium and product. Background Art

[0002] As business needs diversify, existing business systems have higher and higher requirements for work efficiency and data-driven decision-making. Enterprises usually use workflow platforms to optimize business workflows and improve competitiveness.

[0003] One of the representative platforms that uses dialogue to assist in generating workflows among existing workflow platforms is LutraAI. This workflow platform is an intelligent data science and machine learning platform designed to help users automatically generate workflows. However, its technical implementation relies on pre-defined rules and models to generate a completed workflow in a one-time black box manner. This generation method lacks controllability, and users cannot control or intervene in the generation process. The generated workflow results may not meet the specific needs of users and need to be adjusted and modified repeatedly, which greatly increases the cost and time of use. Summary of the invention

[0004] The main purpose of this application is to provide a workflow operator recommendation method, device, equipment, storage medium and product, aiming to solve the technical problem in the prior art that the workflow generation process needs to be repeatedly adjusted and modified, which is time-consuming and costly.

[0005] To achieve the above purpose, the present application proposes a workflow operator recommendation method, which includes:

[0006] Get the predecessor operator selected by the user based on the current workflow task chain;

[0007] Obtaining a subsequent operator requirement input by the user based on the predecessor operator;

[0008] Retrieving the subsequent operator requirements based on the operator database to obtain an operator candidate set;

[0009] The operator candidate set is sorted by a re-ranking model, and the sorted target operator set is recommended to the user.

[0010] In one embodiment, the step of searching the subsequent operator requirements based on the operator database to obtain the operator candidate set includes:

[0011] Performing preprocessing based on the subsequent operator requirements to obtain a search query term;

[0012] Obtain the relationship information between operators through the knowledge graph in the operator database;

[0013] Determining an initial candidate set from an operator database according to the relationship information and the search query;

[0014] Based on the search query, keyword search and semantic search are performed in the initial candidate set to obtain an operator candidate set.

[0015] In one embodiment, the step of sorting the operator candidate set by using a re-ranking model and recommending the sorted target operator set to the user includes:

[0016] De-duplicating and merging operators in the operator candidate set to obtain a de-duplicated operator candidate set;

[0017] Comprehensively sort the candidate set of deduplication operators through the re-ranking model to obtain the target operator set;

[0018] The target operator set is recommended to the user.

[0019] In one embodiment, after the step of sorting the operator candidate set by the re-ranking model and recommending the sorted target operator set to the user, the method further includes:

[0020] If the workflow mode is automatically generated, the pre-stored operator transition probability information is obtained based on the knowledge graph;

[0021] Mapping the workflow state of the current workflow task chain to the knowledge graph to obtain a current workflow subgraph;

[0022] Determining a transition probability between the predecessor operator in the current workflow subgraph and a target operator in the target operator set based on the operator transition probability information;

[0023] A successor operator of the predecessor operator is determined in the target operator set based on the transition probability.

[0024] In one embodiment, before the step of obtaining the predecessor operator selected by the user based on the current workflow task chain, the method further includes:

[0025] Extract keywords from the operator description text using a preset model to obtain text features and semantic features of the operator;

[0026] Determine the relationship information between the operators based on the text features and the semantic features;

[0027] A knowledge graph between the operators is constructed based on the relationship information.

[0028] In one embodiment, the method further comprises:

[0029] Get historical workflow data;

[0030] Determine, based on the historical workflow data, historical operators used in each historical workflow and connection relationships between the historical operators;

[0031] Determining operator transition probability information based on the record frequency of the connection relationship;

[0032] The operator transition probability information is stored in the knowledge graph.

[0033] In addition, to achieve the above purpose, the present application also proposes a workflow operator recommendation device, the workflow operator recommendation device comprising:

[0034] A predecessor operator determination module is used to obtain the predecessor operator selected by the user based on the current workflow task chain;

[0035] A successor operator selection module, used to obtain the successor operator requirements input by the user based on the predecessor operator;

[0036] An operator requirement retrieval module is used to retrieve the subsequent operator requirements based on the operator database to obtain an operator candidate set;

[0037] The target operator recommendation module is used to sort the operator candidate set through a re-ranking model and recommend the sorted target operator set to the user.

[0038] In addition, to achieve the above-mentioned purpose, the present application also proposes a workflow operator recommendation device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the workflow operator recommendation method described above.

[0039] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the workflow operator recommendation method described above are implemented.

[0040] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the workflow operator recommendation method described above are implemented.

[0041] One or more technical solutions proposed in this application have at least the following technical effects:

[0042] This application obtains the predecessor operator selected by the user based on the current workflow task chain; obtains the successor operator requirements input by the user based on the predecessor operator; retrieves the successor operator requirements based on the operator knowledge base to obtain the operator candidate set; sorts the operator candidate set through the reordering model, and recommends the sorted target operator set to the user. Since the operators are gradually determined in the workflow through interaction with the user to generate the workflow, the system recommends appropriate operators at each step based on the user's needs and the current path context information, which can greatly improve flexibility and controllability and reduce the user's usage threshold. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0045] Figure 1 A flowchart of the first embodiment of the workflow operator recommendation method of the present application is provided;

[0046] Figure 2 A flowchart diagram of the second embodiment of the workflow operator recommendation method of this application is provided;

[0047] Figure 3 A flowchart diagram of the workflow operator recommendation method embodiment 3 of this application is provided;

[0048] Figure 4 This is a schematic diagram of the overall process in one implementation of the workflow operator recommendation method of this application;

[0049] Figure 5 This is a schematic diagram of the module structure of the workflow operator recommendation device according to an embodiment of the present application;

[0050] Figure 6 A schematic diagram of the device structure of the hardware operating environment involved in the workflow operator recommendation method in the embodiment of the present application.

[0051] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0052] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0053] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0054] The main solution of the embodiment of the present application is: obtain the predecessor operator selected by the user based on the current workflow task chain; obtain the successor operator requirements input by the user based on the predecessor operator; search the successor operator requirements based on the operator knowledge base to obtain a set of operator candidates; sort the operator candidate set through a reordering model, and recommend the sorted target operator set to the user.

[0055] In the current workflow platforms of the existing technology, it is usually based on predefined rules and models to generate a complete workflow in a one-time black box manner. This generation method lacks controllability, and users cannot control or intervene in the generation process. The generated workflow results may not meet the specific needs of users and need to be adjusted and modified repeatedly, which greatly increases the cost and time of use. Moreover, the workflows generated by the existing technology are often simple linear step splicing, which cannot effectively handle tree-structured multi-branch and multi-path tasks with complex input-output relationships. At the same time, adding new operators or tools to the existing solutions usually requires retraining the entire model, which is a complex and costly process. This limits the system's scalability, especially under the ever-changing business needs, it is difficult to quickly respond and adapt to the introduction of new tools and operators. In addition, the existing workflow generation technology has high requirements for users, especially relying on high-quality user input. Ordinary users find it difficult to generate workflows that meet their needs without sufficient technical background or domain knowledge. Moreover, due to the lack of professional domain knowledge, it performs poorly in vertical fields (such as finance, medical care, etc.) and often cannot meet industry-specific needs. In practical applications, the generated workflow lacks transparency and explainability, and it is difficult for users to understand the internal logic of the workflow generation process, which may lead to a decrease in users' trust in the workflow generation results.

[0056] This application provides a solution that assists in generating workflows step by step through the user's dialogue-style whole process. The system recommends appropriate operators at each step based on the user's needs and current path context information. This innovation breaks down the generation process into controllable individual steps, avoiding the black-box one-time generation process, allowing users to make adjustments at each step to generate a workflow that better meets their needs, which can greatly improve flexibility and controllability and lower the user threshold.

[0057] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a computer, a server, etc., or an electronic device, a virtual device, etc. that can realize the above functions. The following takes a workflow operator recommendation device (hereinafter referred to as the recommendation device) as an example to illustrate this embodiment and the following embodiments.

[0058] Based on this, the present application embodiment provides a workflow operator recommendation method, referring to Figure 1 , Figure 1 This is a flowchart of the first embodiment of the workflow operator recommendation method of the present application.

[0059] In this embodiment, the workflow operator recommendation method includes steps S10 to S40:

[0060] Step S10, obtaining a predecessor operator selected by the user based on the current workflow task chain;

[0061] It can be understood that operators are the basic units for performing specific operations or tasks, and can be regarded as "building blocks" in workflows. By combining multiple operators, a chain-like structure representation can be formed, that is, a workflow task chain, thereby realizing the processing, analysis and decision-making of complex workflows. The above-mentioned specific operations or tasks can be data cleaning, conversion, feature extraction, model training, etc.

[0062] It should be noted that in the embodiment of the present application, a comprehensive operator vector knowledge base is constructed by abstracting the workflow and operators. This knowledge base combines the keyword retrieval of BM25 and the Embedding representation of the RoBERTa encoder to comprehensively reflect the functional characteristics and context dependencies of the operators.

[0063] It can be understood that based on the operators in the operator vector knowledge base (ie, the operator knowledge base), the construction and updating of the operator knowledge graph can be realized, thereby improving the accuracy and context relevance of the operator recommendation.

[0064] It should be understood that the above-mentioned knowledge graph may include all operators and their associations. Through the knowledge graph, the compatibility, dependency and common combinations between operators can be considered in the recommendation process, thereby improving the accuracy and context relevance of the recommendation results.

[0065] In the embodiment of the present application, the recommendation device can determine the initial operator of the workflow task chain based on the user's work requirements, and the initial operator can be regarded as the start of the workflow execution. The above work requirements can be input by the user through language or in text form, and the embodiment of the present application does not limit this.

[0066] It is understandable that at least one operator may be included in the current workflow task chain, and the user may select one of the operators in the current workflow task chain as a predecessor operator. By determining the successor operator of the predecessor operator, the workflow task chain may be expanded so that when the workflow is executed, if the current operator has been executed, the successor operator can be executed. In order to determine the successor operator, in the embodiment of the present application, step S20 obtains the successor operator requirement input by the user based on the predecessor operator;

[0067] In the embodiment of the present application, the embodiment of the present application uses voice input by the user in order to describe the workflow operator recommendation method of the present application, but this does not constitute a limitation on the user command input method in the embodiment of the present application. In actual applications, it can be input by voice or in other ways.

[0068] It should be noted that, based on the analysis of the voice instructions input by the user, the requirements of the successor operator corresponding to the predecessor operator can be determined. Since the workflow operator recommendation is based on the user's voice in a conversational manner, the gradual expansion of the workflow task chain is achieved. The recommendation device can make appropriate operator recommendations at each step (that is, the determination of each operator) based on the user's needs and the context information of the current path. By decomposing the generation process of the workflow task chain into several controllable steps, the black box one-time generation is avoided, allowing users to adjust the workflow in each step to generate a workflow task chain that better meets their needs, greatly improving the flexibility and controllability of workflow generation and lowering the user's usage threshold.

[0069] It can be understood that the successor operator requirement is the basis on which the user determines the successor operator of the predecessor operator, which may include the operations required by the successor operator (such as reading, deleting, adding, modifying, etc.), target operation objects (such as tables, databases, files, etc.), scope, etc., and the embodiments of the present application do not limit this.

[0070] Step S30, searching the subsequent operator requirements based on the operator knowledge base to obtain an operator candidate set;

[0071] Step S40: sort the operator candidate set by using a re-ranking model, and recommend the sorted target operator set to the user.

[0072] It should be noted that the operator knowledge base in the embodiment of the present application is a database generated based on operators and workflows. Operators and corresponding relationships between operators can be stored in the operator knowledge base. By searching the operator knowledge base based on the user's subsequent operator requirements, candidate operators that meet the subsequent operator requirements can be determined. Based on the determined candidate operators, an operator candidate set can be determined.

[0073] It should be understood that in order to improve the accuracy of operator recommendations, the embodiments of the present application can perform multi-dimensional searches based on subsequent operator requirements, such as semantic searches, keyword searches, historical demand operator recommendations, etc., and the embodiments of the present application do not limit this.

[0074] It is understandable that different sets of operator candidates can be obtained based on searches of different dimensions. The same operators may be included in these operator candidate sets. Therefore, when determining the operator candidate sets, the operators in each operator candidate set can be de-duplicated and merged, and each operator can be sorted based on the correlation between the operator and the subsequent operator requirements, so that when recommending the target operator (the target operator is also the operator in the target operator set), the user can give priority to the subsequent operator based on the order of the operators.

[0075] It should be understood that when recommending a target operator to a user, the recommendation device can provide the recommended operator list in a visual form for the user to view the detailed functions of the successor operator, so that the user can select the target successor operator from it. When the user determines the target successor operator, the recommendation device can add the target successor operator to the workflow task chain to update the workflow task chain. For the updated workflow task chain, the user can select the predecessor operator again, and by repeating the operations of selecting the predecessor operator and determining the successor operator, the final target workflow task chain can be generated.

[0076] It should be noted that in order to improve the accuracy of recommendations, the embodiments of the present application can also obtain the path context information in the current workflow task chain during retrieval, and perform operator compatibility and dependency retrieval in the knowledge graph based on the path context information.

[0077] This application visualizes the operator relationships based on the knowledge graph, and displays the operators and task chains, the input-output logical relationships between operators, common operator combinations, etc. to the user through the knowledge graph. The knowledge graph is used to assist in the process of recommending operators, so that users can intuitively understand the relationship between the currently recommended operators and the overall structure of the workflow, thereby improving the transparency of the workflow generation process and user trust.

[0078] At the same time, the embodiment of the present application provides a simple and intuitive user interface. The user only needs to select an operator in the workflow interface and enter the requirements for the next operator in the input box (or through voice input, etc.). The recommendation device can generate several recommended candidate operators under the selected operator. The user only needs to click to select one of them, and the recommendation device can automatically connect it to the workflow, which simplifies the operation steps and lowers the user's learning threshold.

[0079] The embodiment of the present application obtains the predecessor operator selected by the user based on the current workflow task chain; obtains the successor operator requirements input by the user based on the predecessor operator; retrieves the successor operator requirements based on the operator knowledge base to obtain the operator candidate set; sorts the operator candidate set through the reordering model, and recommends the sorted target operator set to the user. Since the operator is gradually determined in the workflow through interaction with the user to generate the workflow, the system recommends appropriate operators at each step based on the user's needs and the current path context information, which can greatly improve flexibility and controllability and reduce the user's usage threshold.

[0080] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 The step of retrieving the subsequent operator requirements based on the operator knowledge base to obtain the operator candidate set includes:

[0081] Step S31, performing preprocessing based on the subsequent operator requirements to obtain a search query term;

[0082] In the embodiment of the present application, the recommendation device can combine keyword retrieval, semantic retrieval and knowledge graph-assisted search to perform multi-way recall to determine the operator candidate set. Finally, the operator candidate set is de-duplicated, merged and sorted through a sophisticated re-ranking model, thereby providing users with high-quality operator recommendations.

[0083] Specifically, the recommendation device can use natural language processing technology to pre-process the subsequent operator requirements input by the user to identify key requirements and generate search query terms. The pre-processing process may include grammatical parsing, keyword extraction, context understanding, etc., which are not limited in the embodiments of the present application. By pre-processing the subsequent operator requirements, it can be ensured that the user requirements are accurately captured.

[0084] Step S32, obtaining the relationship information between operators through the knowledge graph in the operator knowledge base;

[0085] Step S33, determining an initial candidate set from an operator knowledge base according to the relationship information and the search query;

[0086] It should be noted that the recommendation device can narrow down the range of candidate operators through the knowledge graph. The knowledge graph can provide relationship information between operators, such as functional similarity, contextual dependency, and data flow relationship. Based on this relationship information, the recommendation device can filter out operators that are compatible with the nodes in the current workflow task chain and related to user needs to form an initial candidate set.

[0087] Step S34, performing keyword search and semantic search in the initial candidate set based on the search query to obtain an operator candidate set.

[0088] It should be noted that when the initial candidate set is determined based on the knowledge graph, the recommendation device can use a keyword retrieval algorithm to further filter out the operators most relevant to the user query from the initial candidate set to optimize the recommendation scope.

[0089] It is understandable that the above-mentioned keyword retrieval algorithm can be a BM25 retrieval algorithm based on word frequency and inverse document frequency, or other algorithms, such as TF-IDF algorithm, LDA algorithm, etc., and the embodiments of the present application are not limited to this.

[0090] It should be understood that the recommendation device can also perform cosine similarity matching on the Embedding representation of the initial candidate operator in the initial candidate set and the semantics of the subsequent operator requirements, and screen out the operator that is closest to the description input by the user, so that the selected candidate operator also meets the user's needs at the semantic level.

[0091] It is understandable that by performing semantic retrieval and keyword retrieval on the initial candidate set respectively, different operator candidate sets can be obtained, and by removing duplicates, merging, and sorting the operator candidate set, a recommended list of operators that need to be recommended can be determined, so as to output a number of operators that best meet the current workflow task chain and subsequent operator requirements for user selection. That is, the step of sorting the operator candidate set through the reordering model and recommending the sorted target operator set to the user includes: removing duplicates and merging the operators in the operator candidate set to obtain a deduplicated operator candidate set; comprehensively sorting the deduplicated operator candidate set through the reordering model to obtain a target operator set; and recommending the target operator set to the user.

[0092] In one implementation of the embodiment of the present application, the user can start asking questions about the candidate operators in the operator recommendation list. The recommendation device can provide the LLM with relevant information such as the operator function, the current workflow task chain, and the subsequent operator requirements input by the user as context, and the LLM can conduct multiple rounds of dialogue to help the user obtain operator information or usage samples.

[0093] In one implementation of an embodiment of the present application, when determining the target operator set, operators can also be automatically selected from the target operator set based on the workflow auto-completion function (that is, when the user chooses to automatically generate a workflow mode), to help the user automatically select the target successor operator of the predecessor operator based on the current workflow task chain.

[0094] Specifically, the recommendation device of the embodiment of the present application can pre-collect and store a large amount of historical workflow data, which may include the operators used in each workflow task chain and the connection relationship between the operators. By recording the frequency of these connection relationships, the transition probability information between the operators can be obtained, and the transition probability information can be stored in the knowledge graph in the operator knowledge base.

[0095] Furthermore, the recommendation device can map the current workflow task chain to the knowledge graph to become a current workflow subgraph. On the predecessor operator of the current workflow subgraph, the next operator with the highest probability can be selected according to the transition probability of the Markov chain. If there are multiple next successor operators with higher probability. If there are multiple operators with relatively high probability, they can be sorted according to the transition probability, or randomly selected to increase the diversity of operators. In addition, higher-order Markov chains can also be used to capture more accurate operator connection relationships.

[0096] Specifically, after the step of sorting the operator candidate set by the re-ranking model and recommending the sorted target operator set to the user, the method further includes:

[0097] If the workflow mode is automatically generated, the pre-stored operator transition probability information is obtained based on the knowledge graph;

[0098] Mapping the workflow state of the current workflow task chain to the knowledge graph to obtain a current workflow subgraph;

[0099] Determining a transition probability between the predecessor operator in the current workflow subgraph and a target operator in the target operator set based on the operator transition probability information;

[0100] A successor operator of the predecessor operator is determined in the target operator set based on the transition probability.

[0101] In one implementation of the embodiment of the present application, when performing operator recommendations, the knowledge graph may also be queried to ensure that the selected operator not only meets the functional requirements but also complies with the contextual compatibility and dependencies of the current workflow.

[0102] The embodiment of the present application uses Markov chain prediction and knowledge graph compatibility check to automatically identify unfinished parts in the workflow and complete them. This function allows the workflow to be completed in an automated way even if it is incomplete during the generation process, improving the generation efficiency and user convenience, solving the risk of users missing steps or logic when building complex workflows, and significantly reducing the tediousness of manual operations.

[0103] The embodiment of the present application obtains search query words by performing preprocessing based on the subsequent operator requirements; obtains the relationship information between operators through the knowledge graph in the operator knowledge base; determines the initial candidate set from the operator knowledge base based on the relationship information and the search query words; and performs keyword search and semantic search in the initial candidate set based on the search query words to obtain the operator candidate set. Since the initial candidate set is screened out based on the relationship information between operators, and keyword search and semantic search are performed based on the initial candidate set, the accuracy of operator recommendation is improved.

[0104] Based on the first embodiment and / or the second embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those of the above-mentioned first embodiment and / or the second embodiment can refer to the above introduction, and will not be repeated later. Figure 3 In the embodiment of the present application, before the step of obtaining the predecessor operator selected by the user based on the current workflow task chain, it also includes:

[0105] Step S100, extracting keywords from the description text of the operator through a preset model to obtain text features and semantic features of the operator;

[0106] Step S200, determining the relationship information between the operators based on the text features and the semantic features;

[0107] Step S300, constructing a knowledge graph between the operators based on the relationship information.

[0108] In the embodiment of the present application, in order to construct an operator knowledge base, the recommendation device may first abstract the workflow task chain and the operator in the operator database. A workflow is a set of tasks or steps, where each step or task has a clear sequential relationship. An operator is a basic execution unit in a workflow task chain, which can be responsible for completing specific data processing tasks, such as data cleaning, conversion, feature extraction, or model training.

[0109] Specifically, in the embodiment of the present application, the identification ID, name, input, output, type, function description and other information of the operator are uniformly saved in the operator knowledge base through a structured file format (such as json), and the workflow task chain can be regarded as a directed acyclic graph composed of operators.

[0110] In the embodiment of the present application, in order to improve the accuracy of operator recommendation, a comprehensive operator vector knowledge base, that is, the above-mentioned operator knowledge base, can be constructed. The operator knowledge base can combine the keyword retrieval of BM25 and the Embedding representation of the RoBERTa encoder to fully reflect the functional characteristics and context dependencies of the operator. The specific implementation can be as follows:

[0111] In the embodiment of the present application, the description text of the operator can be first subjected to keyword extraction to realize the text feature of capturing the operator function. The above keyword extraction method can be based on BM25 model extraction or other extraction methods, which are not limited in the embodiment of the present application.

[0112] At the same time, pre-trained language models (such as RoBERTa) can be applied to encode the text description and structural information of the operator, from which semantic features such as functional description, input and output parameter types, and their correlation with other operators can be extracted.

[0113] In one implementation of the embodiment of the present application, in order to ensure the timeliness and accuracy of the operator knowledge base, incremental learning and online update technology can be used to regularly update the operator knowledge base and the Embedding database. This maintenance mechanism ensures that the knowledge base always reflects the latest operator information and its contextual relationship.

[0114] It should be noted that in the embodiments of the present application, operator information such as input-output dependency and functional similarity of operators can be determined, and a relationship graph (i.e., a knowledge graph) between operators can be constructed based on the operator information. In a knowledge graph, each node can represent an operator, and an edge can represent a dependency or compatibility relationship between operators.

[0115] It is understandable that, based on the operator information in the operator knowledge base, an operator knowledge base can also be constructed and obtained. The semantic search and keyword search in the above embodiment can be performed based on the operator knowledge base, and the compatibility and dependency relationship between operators can be determined based on the knowledge graph.

[0116] It should be explained that in the embodiments of the present application, the completed historical workflow template can be used to record the Markov chain transition probability of each operator in different contexts, thereby providing an important reference for the automatic completion function.

[0117] It is understandable that in the embodiments of the present application, the operator relationships and weights in the knowledge graph can be continuously updated through automated data collection and user feedback, so that its clock reflects the latest workflow trends and user preferences.

[0118] The embodiment of the present application extracts keywords from the description text of the operator through a preset model to obtain the text features and semantic features of the operator; determines the relationship information between the operators based on the text features and semantic features; and constructs a knowledge graph between the operators based on the relationship information. The embodiment of the present application abstracts the workflow and the operator and constructs the corresponding knowledge graph, which provides an important basis for the accuracy of operator recommendation.

[0119] In one implementation of the embodiment of the present application, the overall workflow diagram of the workflow operator recommendation method of the embodiment of the present application can be referred to Figure 4 .

[0120] It should be noted that Figure 4 The numerical labels in the table are used to indicate the execution order, such as label 1 is executed before label 2. For the same label, they can be executed at the same time or any one can be executed first.

[0121] Specifically, users can perform 1. predecessor operator selection based on the current workflow task chain. When determining the predecessor operator, the recommendation device can obtain 2. the successor node requirements provided by the user, and obtain 2. the path context information through the current workflow task chain. The retriever can 3. search for semantics and keywords in the operator knowledge base based on the successor node requirements, and search for compatibility and dependencies in the operator relationship knowledge graph (that is, the above-mentioned relationship graph) based on the path context. Among them, the operator knowledge base can be constructed based on feature extraction of the workflows and operators in the operator database, and the knowledge graph can be constructed based on the operator relationships in the operator database.

[0122] Furthermore, the retriever can perform 4. visualization operator recommendations based on the obtained search results, so that the user can 5. select a successor node based on the recommended operator, thereby achieving the expansion of the workflow task chain.

[0123] In an optional implementation of an embodiment of the present application, automatic expansion of the workflow task chain can be achieved based on knowledge graph and Markov chain automatic completion.

[0124] It is understandable that by repeating the above steps, a final complete workflow task chain can be generated. The user can adjust the generated complete workflow task chain.

[0125] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the workflow operator recommendation method of the present application. More simple transformations based on this technical concept are all within the protection scope of the present application.

[0126] This application also provides a workflow operator recommendation device, please refer to Figure 5 , the workflow operator recommendation device comprises:

[0127] A predecessor operator determination module 10 is used to obtain a predecessor operator selected by a user based on the current workflow task chain;

[0128] A successor operator selection module 20, used to obtain the successor operator requirement input by the user based on the predecessor operator;

[0129] An operator requirement retrieval module 30 is used to retrieve the subsequent operator requirements based on the operator database to obtain an operator candidate set;

[0130] The target operator recommendation module 40 is used to sort the operator candidate set through a re-ranking model and recommend the sorted target operator set to the user.

[0131] The workflow operator recommendation device provided by the present application adopts the workflow operator recommendation method in the above embodiment, which can solve the technical problem that the workflow generation process in the prior art needs to be repeatedly adjusted and modified, which is time-consuming and costly. Compared with the prior art, the beneficial effects of the workflow operator recommendation device provided by the present application are the same as the beneficial effects of the workflow operator recommendation method provided by the above embodiment, and the other technical features in the workflow operator recommendation device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0132] The present application provides a workflow operator recommendation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the workflow operator recommendation method in the above-mentioned embodiment one.

[0133] Reference below Figure 6 , which shows a schematic diagram of the structure of a workflow operator recommendation device suitable for implementing the embodiment of the present application. The workflow operator recommendation device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The workflow operator recommendation device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0134] like Figure 6As shown, the workflow operator recommendation device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. Various programs and data required for the operation of the workflow operator recommendation device are also stored in the RAM 1004. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems may be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 1009. The communication device 1009 may allow the workflow operator recommendation device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a workflow operator recommendation device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented or have instead.

[0135] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0136] The workflow operator recommendation device provided by the present application adopts the workflow operator recommendation method in the above embodiment, which can solve the technical problem that the workflow generation process in the prior art needs to be repeatedly adjusted and modified, which is time-consuming and costly. Compared with the prior art, the beneficial effects of the workflow operator recommendation device provided by the present application are the same as the beneficial effects of the workflow operator recommendation method provided by the above embodiment, and the other technical features in the workflow operator recommendation device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0137] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0138] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0139] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, wherein the computer-readable program instructions are used to execute the workflow operator recommendation method in the above-mentioned embodiment.

[0140] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0141] The computer-readable storage medium may be included in the workflow operator recommendation device; or may exist independently without being assembled into the workflow operator recommendation device.

[0142] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the workflow operator recommendation device, the workflow operator recommendation device:

[0143] Get the predecessor operator selected by the user based on the current workflow task chain;

[0144] Obtaining a subsequent operator requirement input by the user based on the predecessor operator;

[0145] Retrieving the subsequent operator requirements based on the operator database to obtain an operator candidate set;

[0146] The operator candidate set is sorted by a re-ranking model, and the sorted target operator set is recommended to the user.

[0147] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0148] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0149] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0150] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned workflow operator recommendation method, and can solve the technical problem that the workflow generation process in the prior art needs to be repeatedly adjusted and modified, which is time-consuming and costly. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the workflow operator recommendation method provided by the above-mentioned embodiment, and will not be repeated here.

[0151] The present application also provides a computer program product, including a computer program, which implements the steps of the workflow operator recommendation method as described above when executed by a processor.

[0152] The computer program product provided by the present application can solve the technical problem that the workflow generation process in the prior art needs to be repeatedly adjusted and modified, which is time-consuming and costly. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the workflow operator recommendation method provided by the above embodiment, which will not be repeated here.

[0153] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A workflow operator recommendation method, characterized in that: The method comprises: Get the predecessor operator selected by the user based on the current workflow task chain; Obtaining a subsequent operator requirement input by the user based on the predecessor operator; Retrieving the subsequent operator requirements based on the operator database to obtain an operator candidate set; The operator candidate set is sorted by a re-ranking model, and the sorted target operator set is recommended to the user.

2. The workflow operator recommendation method according to claim 1, characterized in that: The step of searching the subsequent operator requirements based on the operator database to obtain an operator candidate set includes: Performing preprocessing based on the subsequent operator requirements to obtain a search query term; Obtain the relationship information between operators through the knowledge graph in the operator database; Determining an initial candidate set from an operator database according to the relationship information and the search query; Based on the search query, keyword search and semantic search are performed in the initial candidate set to obtain an operator candidate set.

3. The workflow operator recommendation method according to claim 2, characterized in that: The step of sorting the operator candidate set by using the re-sorting model and recommending the sorted target operator set to the user comprises: De-duplicating and merging operators in the operator candidate set to obtain a de-duplicated operator candidate set; Comprehensively sort the candidate set of deduplication operators through the re-ranking model to obtain the target operator set; The target operator set is recommended to the user.

4. The workflow operator recommendation method according to claim 2, characterized in that: After the step of sorting the operator candidate set by the re-ranking model and recommending the sorted target operator set to the user, the method further includes: If the workflow mode is automatically generated, the pre-stored operator transition probability information is obtained based on the knowledge graph; Mapping the workflow state of the current workflow task chain to the knowledge graph to obtain a current workflow subgraph; Determining a transition probability between the predecessor operator in the current workflow subgraph and a target operator in the target operator set based on the operator transition probability information; A successor operator of the predecessor operator is determined in the target operator set based on the transition probability.

5. The workflow operator recommendation method according to claim 1, characterized in that: Before the step of obtaining the predecessor operator selected by the user based on the current workflow task chain, the method further includes: Extract keywords from the operator description text using a preset model to obtain text features and semantic features of the operator; Determine the relationship information between the operators based on the text features and the semantic features; A knowledge graph between the operators is constructed based on the relationship information.

6. The workflow operator recommendation method according to claim 5, characterized in that: The method further comprises: Get historical workflow data; Determine, based on the historical workflow data, historical operators used in each historical workflow and connection relationships between the historical operators; Determining operator transition probability information based on the record frequency of the connection relationship; The operator transition probability information is stored in the knowledge graph.

7. A workflow operator recommendation device, characterized in that: The workflow operator recommendation device comprises: A predecessor operator determination module is used to obtain the predecessor operator selected by the user based on the current workflow task chain; A successor operator selection module, used to obtain the successor operator requirements input by the user based on the predecessor operator; An operator requirement retrieval module is used to retrieve the subsequent operator requirements based on the operator database to obtain an operator candidate set; The target operator recommendation module is used to sort the operator candidate set through a re-ranking model and recommend the sorted target operator set to the user.

8. A workflow operator recommendation device, characterized in that: The device comprises: a memory, a processor, and a workflow operator recommendation program stored in the memory and executable on the processor, wherein the workflow operator recommendation program is configured to implement the steps of the workflow operator recommendation method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium stores a workflow operator recommendation program, and when the workflow operator recommendation program is executed by the processor, the steps of the workflow operator recommendation method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the workflow operator recommendation method according to any one of claims 1 to 6 are implemented.

Images