A data analysis method and related equipment

Abstract

The present application discloses a data analysis method and related equipment, the method comprising: after obtaining the business log data of the target object, extracting a business path set based on the business log data, and cutting the business path set to obtain a cutting result with business activities as cutting units, further analyzing the cutting results to obtain a business process analysis result corresponding to the business log data, and finally predicting the business process prediction result of the target object based on the business process analysis result. Since the cutting result is obtained by cutting the business path set with business activities as cutting units, data analysis can be performed in depth and meticulously based on such fine cutting results, and the business process analysis results obtained by analysis have less information deviation than the directly extracted business log data. By performing business process prediction based on such data with reduced information deviation, the accuracy of business process prediction can be improved, and the user experience can be improved.

Description

Technical Field

[0001] The present application relates to the field of Internet technology, and in particular to a data analysis method and related equipment. Background Art

[0002] At present, data analysis has gradually become a major trend in the development of information technology. Data analysis plays an important role in data prediction. With the development of automation technology, RPA (Robotic Process Automation) products have emerged. RPA products can automatically execute processes and reduce manual burdens, so RPA products have been widely used in many fields. In the current application process, users not only need RPA products to automatically execute processes, but also hope that they can predict the subsequent process development to guide corporate decision-making.

[0003] Existing process prediction solutions are generally based on historical data related to business processes. Since the analysis of historical data is often not in-depth enough, there are many information biases, which will affect the accuracy of business process prediction results. In other words, the RPA products with business process prediction functions developed in the past are not accurate enough in prediction, resulting in poor user experience. Summary of the invention

[0004] The embodiments of the present application provide a data analysis method and related equipment to improve the accuracy of RPA products in business process prediction, thereby improving the user experience.

[0005] In a first aspect, an embodiment of the present application provides a data analysis method, comprising:

[0006] Obtain the business log data of the target object;

[0007] Extracting a business path set based on the business log data; the business path set includes a business path, and the business path includes at least two business activities;

[0008] Cutting the business path set to obtain a cutting result with business activities as cutting units;

[0009] Analyzing the cutting result to obtain a business process analysis result corresponding to the business log data;

[0010] A business process prediction result of the target object is obtained based on the business process analysis result prediction.

[0011] Optionally, the segmenting the business path set to obtain a segmentation result with business activities as segmentation units includes:

[0012] Based on the business activities included in the business paths in the business path set and the execution relationship between the business activities, the probability of obtaining multiple different cutting methods; the cutting method includes the execution relationship and the cutting position of the business activity set corresponding to the business path set; the business activity set includes all business activities included in the business path set;

[0013] Determining a target cutting method from the multiple different cutting methods according to a maximum value among the acquired probabilities;

[0014] The business path set is cut according to the target cutting method to obtain a plurality of cutting results corresponding to the business path set, and each of the cutting results only includes one business activity.

[0015] Optionally, the segmenting the service path set according to the target segmentation mode to obtain a plurality of segmentation results corresponding to the service path set includes:

[0016] When the cutting result includes a business path composed of multiple business activities, the cutting result is used as a new business path set and a target cutting method is re-determined;

[0017] The new business path set is continuously cut in the re-determined target cutting manner, and the process is executed in a loop until all cutting results contain only one business activity.

[0018] Optionally, if the cutting position is set between the first business activity and the second business activity in the business activity set, the probability of obtaining a plurality of different cutting modes based on the business activities included in the business paths in the business path set and the execution relationship between the business activities includes:

[0019] determining a first execution probability between the first business activity and the second business activity under a plurality of different execution relationships;

[0020] Based on the cutting position and the business activity set, determining a first business activity set including the first business activity and determining a second business activity set including the second business activity;

[0021] Based on the first execution probabilities under the multiple different execution relationships, the first business activity set, and the second business activity set, obtaining second execution probabilities between the first business activity set and the second business activity set under the multiple different execution relationships;

[0022] The second execution probabilities under the multiple different execution relationships are used as the probabilities of the multiple different cutting modes.

[0023] Optionally, the determining a first execution probability between the first business activity and the second business activity under a plurality of different execution relationships includes:

[0024] Obtaining an average value between the number of executions of the first business activity in the business path set and the number of executions of the second business activity in the business path set;

[0025] According to the average value and based on the multiple different execution relationships, a first execution probability under the multiple different execution relationships is calculated.

[0026] Optionally, the analyzing the segmentation result to obtain a business process analysis result corresponding to the business log data includes:

[0027] All target cutting methods and all cutting results that only include one business activity are arranged in layers according to the order of cutting to obtain the business process analysis result.

[0028] Optionally, the predicting the business process prediction result of the target object based on the business process analysis result includes:

[0029] Determine the impact characteristics of the business process analysis result based on the business log data;

[0030] Extracting multiple business paths with consistent influencing characteristics from the business path set, and constructing a sample set corresponding to the consistent influencing characteristics according to the multiple business paths;

[0031] Determining the information gain of the consistent influencing feature based on the sample set and the consistent influencing feature;

[0032] Based on the information gain, a decision tree model for predicting the business process is established to predict the business process prediction result.

[0033] Optionally, establishing a decision tree model for predicting a business process based on the information gain includes:

[0034] When there are multiple consistent influencing features, sequentially constructing multiple decision nodes of the consistent influencing features in descending order of information gain;

[0035] The decision tree model is generated based on the plurality of decision nodes and the plurality of different impact results respectively corresponding to the plurality of consistent impact features.

[0036] Optionally, before obtaining the business log data of the target object, the data analysis method further includes:

[0037] In response to a desensitization request for sensitive data in the business log data, the sensitive data is desensitized to obtain desensitized business log data.

[0038] In a second aspect, an embodiment of the present application provides a data analysis device, including:

[0039] A data acquisition module is used to acquire the business log data of the target object;

[0040] A path extraction module, configured to extract a business path set based on the business log data; the business path set includes a business path, and the business path includes at least two business activities;

[0041] A path cutting module, used to cut the business path set to obtain a cutting result with business activities as cutting units;

[0042] A cutting result analysis module, used for analyzing the cutting result to obtain a business process analysis result corresponding to the business log data;

[0043] The process prediction module is used to predict the business process prediction result of the target object based on the business process analysis result.

[0044] In a third aspect, an embodiment of the present application provides a device, including: a processor, a memory, and a system bus;

[0045] The processor and the memory are connected via the system bus;

[0046] The memory is used to store one or more programs, and the one or more programs include instructions. When the instructions are executed by the processor, the processor executes any one of the implementation methods of the above-mentioned data analysis method.

[0047] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal device, the terminal device executes any one of the implementation methods of the above-mentioned data analysis method.

[0048] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0049] In an embodiment of the present application, after obtaining the business log data of the target object, a business path set can be extracted based on the business log data, and the business path set can be cut to obtain a cutting result with business activities as the cutting unit, and the cutting result is further analyzed to obtain a business process analysis result corresponding to the business log data, and finally the business process prediction result of the target object is predicted based on the business process analysis result. Since the cutting result is obtained by cutting the business path set with business activities as the cutting unit, based on this fine cutting result, in-depth and detailed data analysis can be performed to obtain the business process analysis result, and then this business process analysis result has less information deviation than the directly extracted business log data. In this way, based on this data with reduced information deviation (that is, the business process analysis result) to perform business process prediction, the accuracy of business process prediction can be improved, and the user experience can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flowchart of a data analysis method provided in an embodiment of the present application;

[0051] Figure 2 A schematic diagram of hierarchically arranging all target cutting methods and all cutting results containing only one business activity provided in an embodiment of the present application;

[0052] Figure 3 Another schematic diagram of hierarchically arranging all target cutting methods and all cutting results containing only one business activity provided in an embodiment of the present application;

[0053] Figure 4a A schematic diagram of a process tree obtained by hierarchically arranging all target cutting methods and all cutting results containing only one business activity provided in an embodiment of the present application;

[0054] Figure 4b Another schematic diagram of a process tree obtained by hierarchically arranging all target segmentation methods and all segmentation results containing only one business activity provided in an embodiment of the present application;

[0055] Figure 5 A flowchart of an implementation method for cutting a service path set provided in an embodiment of the present application;

[0056] Figure 6 A flowchart of an implementation method for obtaining a business process prediction result of a target object provided in an embodiment of the present application;

[0057] Figure 7 A schematic diagram of a decision tree model for predicting business processes based on information gain provided in an embodiment of the present application;

[0058] Figure 8 A schematic diagram of the structure of a data analysis device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] As mentioned above, the inventors found in the research on the process prediction function of RPA products that: in the related art, the business process is generally predicted directly based on the historical data related to the business process. However, since the analysis of historical data in advance is often not in-depth enough, there are many information deviations, which will affect the accuracy of the business process prediction results. For example, for business log data related to the current business process, if the business process operator mistakenly executes the business activities of other business processes, the business log data will contain data unrelated to the current business process; if the business process operator forgets to execute a business activity of the current business process, the business log data will lack data related to the business activity. This will make the business log data incomplete and / or there will be irrelevant data, resulting in more information deviations, and then the accuracy of the solution for directly predicting the business process based on the business log data is insufficient. It can be seen that the RPA products with business process prediction functions developed in the past, because they predict the business process directly based on the historical data related to the business process, are insufficient in accuracy in prediction, which will lead to poor user experience.

[0060] In order to solve the above problems, the embodiment of the present application provides a data analysis method, which includes: after obtaining the business log data of the target object, a business path set can be extracted based on the business log data, and the business path set can be cut to obtain a cutting result with business activities as cutting units, and the cutting result is further analyzed to obtain a business process analysis result corresponding to the business log data, and finally the business process prediction result of the target object is predicted based on the business process analysis result. Wherein, the above business path set includes a business path, and the business path includes at least two business activities.

[0061] It can be seen that since the above segmentation results are obtained by segmenting the business path set with business activities as the segmentation unit, based on this fine segmentation result, in-depth and detailed data analysis can be performed to obtain business process analysis results, and then this business process analysis result has less information bias than directly extracted business log data. In this way, based on this data with reduced information bias (that is, business process analysis results) to perform business process prediction, the accuracy of business process prediction can be improved, and the user experience can be improved.

[0062] It should be noted that the embodiments of the present application do not limit the execution subject of the data analysis method. For example, the data analysis method of the embodiments of the present application can be applied to data processing devices such as terminal devices or servers. Among them, the terminal device can be a smart phone, a computer, an RPA product or a tablet computer. The server can be an independent server, a cluster server or a cloud server.

[0063] In order to facilitate understanding of the technical solutions provided in the embodiments of the present application, the application scenarios of the data analysis methods provided in the embodiments of the present application are exemplarily introduced below.

[0064] In one application scenario, after the RPA product obtains the business log data of the target object, the business process prediction can be performed through the data analysis method provided in the embodiment of the present application. For example, the process of business process prediction by the RPA product can specifically include: the RPA product first extracts a business path set based on the business log data, and cuts the business path set to obtain a cutting result with business activities as the cutting unit, and further analyzes the cutting result to obtain the business process analysis result corresponding to the business log data, and finally predicts the business process prediction result of the target object based on the business process analysis result, so that the user associated with the target object, that is, the business process operator mentioned above, can use the RPA product to predict the business process.

[0065] In another application scenario, after the RPA product obtains the business log data of the target object, it can forward the business log data to the server so that the server can perform business process prediction through the data analysis method provided in the embodiment of the present application. For example, the process of the server performing business process prediction may specifically include: the server first extracts a business path set based on the business log data, and cuts the business path set to obtain a cutting result with business activities as the cutting unit, and further analyzes the cutting result to obtain the business process analysis result corresponding to the business log data, and finally predicts the business process prediction result of the target object based on the business process analysis result. Then, the server sends the business process prediction result to the RPA product so that users associated with the target object can use the RPA product to perform business process prediction.

[0066] It should be noted that the data analysis method provided in the embodiment of the present application can be applied not only to the above two application scenarios, but also to other application scenarios that require data analysis, and the embodiment of the present application does not make specific limitations on this.

[0067] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0068] Figure 1 A flowchart of a data analysis method provided in an embodiment of the present application. Figure 1 As shown, the data analysis method provided in the embodiment of the present application may include:

[0069] S101: Acquire business log data of a target object.

[0070] The target object refers to the business process that needs to be analyzed. In practical applications, not only can the business log data of the target object be obtained and operated, but also the business log data of other objects can be obtained and operated, so as to perform business process prediction based on the business log data of each object in the future, further improving the applicability of the data analysis method. In order to facilitate the understanding of the data analysis method based on the business log data of a specific object, in the embodiments of the present application, a detailed description is made by taking the target object as an example.

[0071] In addition, in the embodiments of the present application, the implementation method of obtaining the business log data of the target object may not be limited. For example, the business log data actively uploaded by the user associated with the target object may be obtained, or the business log file may be crawled from the terminal device of the user associated with the target object.

[0072] In the case of crawling business log files from the user's terminal device, since it involves the user's terminal device and other devices that may be related to the user's privacy, when the embodiments of the present application are applied to specific products or technologies, the user's permission or consent is required, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions. For example, before crawling business log files from the user's terminal device, the user's permission or consent must be sought to obtain the user's authorization to crawl the business log files.

[0073] For the situation of obtaining the business log data actively uploaded by the user, in order to eliminate the user's security concerns and improve the user's experience, the embodiment of the present application can desensitize the business log data uploaded by the user in advance to encrypt the sensitive data therein. Specifically, before obtaining the business log data of the target object, the data analysis method can also include: in response to the desensitization request for the sensitive data in the business log data, the sensitive data is desensitized to obtain the desensitized business log data. Among them, the sensitive data in the business log data can be set by the user in combination with actual needs, or it can be automatically determined in combination with the previous desensitization request, and the embodiment of the present application does not specifically limit this. Moreover, in the embodiment of the present application, the implementation method of desensitizing sensitive data is also not specifically limited. For example, the value of the sensitive data can be intercepted and hidden with special symbols to desensitize the sensitive data, or the value of the sensitive data can be replaced with a preset virtual value to desensitize the sensitive data.

[0074] S102: Extracting a service path set based on the service log data.

[0075] Before extracting the business path set based on the business log data, the data in the business log data can be mapped to entities to facilitate the extraction of the business path set. For example, the three columns of data CaseID, Activity, and StartandEndTime in the business log data can be mapped to entities respectively. CaseID can represent the identification information of the business path, and then can represent all the business activities included in the business path; Activity can represent the business activities actually executed in the business path; StartandEndTime can represent the execution order of different business activities in the business path.

[0076] A business path set may refer to a set of business paths, which may include business paths. Among them, a business path refers to a business activity actually performed for a target object. A business path may include at least two business activities. Taking the onboarding scenario of a new employee as an example, business activities may include confirming the onboarding information of a new employee, sending a collection form (hereinafter referred to as a), submitting personal information (hereinafter referred to as b), confirming the onboarding information and triggering the process (hereinafter referred to as c), verifying employee materials (hereinafter referred to as d) and refusing to process the onboarding (hereinafter referred to as e). Accordingly, the business path may be represented as<a,b,c,d> ,<a,b,c,d,e,b,c,d> ,<a,b,c,d,e,b,c,d,e> . The business path set can be expressed as [<a,b,c,d> ,<a,b,c,d,e,b,c,d> ,<a,b,c,d,e,b,c,d,e> ]. Furthermore, if there is a repeated business path, such as business path<a,b,c,d> Repeat 5 times, business path<a,b,c,d,e> Repeat 12 times, and the service path set can be expressed as [<a,b,c,d> 5 ,<a,b,c,d,e,b,c,d> 12 ,<a,b,c,d,e,b,c,d,e> ], that is, the number of times a business path is executed in a business path set can be represented in the form of a superscript, thereby simplifying the representation of the business path set. In addition to the above representation forms, business activities, business paths, and business path sets can also be represented in the form of a flowchart, which is not specifically limited in the embodiments of the present disclosure.

[0077] S103: Cut the business path set to obtain a cutting result with business activities as cutting units.

[0078] When the business path set contains only one business path, cutting the business path set can be reflected as cutting the business path. When the business path set contains multiple business paths, cutting the business path set can be reflected as cutting the multiple business paths simultaneously in the same cutting method. The cutting method may include the execution relationship between the business activities contained in the business path, and the cutting position of the business activity set corresponding to the business path set. The execution relationship may include a sequential composition relationship, a unique selection relationship, a parallel composition relationship, or a circular composition relationship. The differences between the four execution relationships are described below. In addition, combined with the example of the new employee onboarding scenario mentioned above, it can be seen that the business paths in the business path set have different execution situations for business activities. Therefore, all business activities contained in the business path set can be extracted in advance, and all business activities can be used to form the business activity set corresponding to the business path set, so as to determine the cutting position based on all business activities to avoid errors and omissions during cutting. Here, the cutting position of the business activity set can indicate the cutting position of the business path in the business path set. For example, the business path set is [<a,b,c,d> 5 ,<a,b,c,d,e,b,c,d> 12 ,<a,b,c,d,e,b,c,d,e> ], the business activity set is {a,b,c,d,e}; if the cutting position of the business activity set is between a and b, it means that the business path in the business path set needs to be cut from the position between a and b, that is, the business activity set can be expressed as {a} and {b,c,d,e}, and accordingly, the cutting result is [ 18 ]and[<b,c,d> 5 ,<b,c,d,e,b,c,d> 12 ,<b,c,d,e,b,c,d,e> ].

[0079] The result of cutting with business activities as the cutting unit means that the cutting result contains only one business activity. It can be understood that one cutting can obtain two cutting results. Therefore, when the business activity set contains two business activities, the cutting result with business activities as the cutting unit can be obtained through one cutting; and when the business activity set contains more than two business activities, it is necessary to cut multiple times to obtain the cutting result with business activities as the cutting unit. For example, if the business activity set is {a, b}, then cutting once can obtain the cutting result containing only business activity a [ ] and the split result containing only business activity b[ ]; If the business activity set is {a,b,c,d,e}, and the cutting position of the business activity set is between a and b, the cutting result after the first cutting is [ 18 ]and[<b,c,d> 5 ,<b,c,d,e,b,c,d> 12 ,<b,c,d,e,b,c,d,e> ],for[<b,c,d> 5 ,<b,c,d,e,b,c,d> 12 ,<b,c,d,e,b,c,d,e> ] This segmentation result contains a business path composed of multiple business activities, so it is necessary to continue segmenting until a segmentation result containing only one business activity is obtained (that is, [ ]、[ ]、[ <c> ]、[ <d> ]、[ <e>]).

[0080] In addition, the embodiments of the present application may not make specific limitations on the cutting of the service path set. For ease of understanding, the embodiments of the present application provide a possible implementation method. For technical details, please refer to the introduction of the embodiments below.

[0081] S104: Analyze the cutting result to obtain the business process analysis result corresponding to the business log data.

[0082] The business process analysis result corresponds to the target object and can be used to characterize the target business process of the target object, that is, to characterize the business activities and the execution relationship between the business activities. Here, the embodiment of the present application does not limit the analysis process of the segmentation result. For ease of understanding, a possible implementation method is described below.

[0083] In a possible implementation, S104 may include: arranging all target segmentation methods and all segmentation results that only include one business activity in layers according to the order of segmentation to obtain a business process analysis result.

[0084] Among them, the target cutting method refers to the method of cutting the business path set each time. For the case where the business activity set contains two business activities, only one cutting is required to obtain the cutting result containing only one business activity. The target cutting method and the corresponding cutting result can be arranged in the same layer or the two can be arranged in layers. For the case where the business activity set contains more than two business activities, it is necessary to cut multiple times to obtain the cutting result containing only one business activity. The embodiment of the present application can provide a variety of hierarchical arrangement representation methods, which are illustrated by examples below.

[0085] As an example, each target cutting method and the corresponding cutting result including only one business activity may be arranged in the same layer according to the cutting sequence, until all target cutting methods and the corresponding cutting results are arranged.< / e> < / d> < / c> Figure 2 A schematic diagram of a hierarchical arrangement of all target segmentation methods and all segmentation results containing only one business activity provided in an embodiment of the present application. Figure 2 As shown, assuming that the business path set is cut according to the target cutting method 1, the two cutting results obtained are the cutting result 11 containing only one business activity and the cutting result 12 containing a business path composed of multiple business activities, then the target cutting method 1 and the cutting result 11 can be arranged in the first layer, wherein the first layer can also include the cutting result 12. Further, taking the cutting result 12 as a new business path set, assuming that the new business path set is cut according to the target cutting method 2, the two cutting results obtained are the cutting result 21 containing only one business activity and the cutting result 22 containing a business path composed of multiple business activities, then the target cutting method 2 and the cutting result 21 can be arranged in the second layer, wherein the second layer can also include the cutting result 22. In this way, according to the order of cutting, each target cutting method and the corresponding cutting result containing only one business activity are arranged in the same layer in turn, until the target cutting method n, the corresponding cutting result n1 and the cutting result n2 are arranged in the nth layer, wherein the cutting result n1 and the cutting result n2 each contain only one business activity.

[0086] As another example, each target cutting method and the corresponding cutting result may be arranged in layers according to the order of cutting, until all target cutting methods and the corresponding cutting results are arranged. Figure 3 Another schematic diagram of arranging all target segmentation methods and all segmentation results containing only one business activity in layers is provided in the embodiment of the present application. Figure 3 As shown, assuming that the business path set is cut according to the target cutting method 1, the two cutting results obtained are the cutting result 11 containing only one business activity, and the cutting result 12 containing a business path composed of multiple business activities. Then, the target cutting method 1 and the cutting result 11 can be arranged in layers, wherein the target cutting method 1 can be arranged in the first layer, the cutting result 11 can be arranged in the second layer, and the second layer can also include the cutting result 12. Further, taking the cutting result 12 as a new business path set, assuming that the new business path set is cut according to the target cutting method 2, the two cutting results obtained are the cutting result 21 containing only one business activity, and the cutting result 22 containing a business path composed of multiple business activities. Then, the target cutting method 2 and the cutting result 21 can continue to be arranged in layers, wherein the target cutting method 2 can be arranged in the third layer, the cutting result 21 can be arranged in the fourth layer, and the fourth layer can also include the cutting result 22. In this way, according to the order of cutting, each target cutting method and the corresponding cutting result are arranged in layers in sequence, until the target cutting method n is arranged in the 2n-1 layer, and the corresponding cutting result n1 and cutting result n2 are arranged in the 2n layer, where the cutting result n1 and the cutting result n2 each contain only one business activity.

[0087] As another example, in order to provide clear and intuitive business process analysis results, the embodiment of the present application can hierarchically arrange all target cutting methods and all cutting results that only contain one business activity in the form of a process tree. Specifically, all target cutting methods can be used as internal nodes, and all cutting results that only contain one business activity can be used as leaf nodes. They can be hierarchically arranged in the order of cutting to obtain a process tree to represent the business process analysis results. Figure 4a A schematic diagram of a process tree obtained by hierarchically arranging all target segmentation methods and all segmentation results containing only one business activity provided in an embodiment of the present application. Figure 4a As shown, assuming that the business path set is cut according to the target cutting method 1, the two cutting results obtained are the cutting result 11 containing only one business activity, and the cutting result containing the business path composed of multiple business activities. Then, the target cutting method 1 and the cutting result 11 can be arranged in layers, wherein the target cutting method 1 can be arranged in the first layer as an internal node, and the cutting result 11 can be arranged in the second layer as a leaf node. Further, the cutting result containing the business path composed of multiple business activities is used as a new business path set. Assuming that the new business path set is cut according to the target cutting method 2, the two cutting results obtained are the cutting result containing the business path composed of multiple business activities and the cutting result 22 containing only one business activity. Then, the target cutting method 2 and the cutting result 22 can continue to be arranged in layers, wherein the target cutting method 2 can continue to be arranged in the second layer as an internal node, and the cutting result 22 can be arranged in the third layer as a leaf node. In this way, according to the order of cutting, all target cutting methods are used as internal nodes, and all cutting results that only contain one business activity are arranged in layers as leaf nodes until the target cutting method n and the corresponding cutting results n1 and n2 are arranged, where the cutting results n1 and n2 each contain only one business activity.

[0088] in addition, Figure 4b Another schematic diagram of a process tree obtained by hierarchically arranging all target segmentation methods and all segmentation results containing only one business activity provided in an embodiment of the present application. Figure 4b As shown, for the first cut, assume that the service path set is [<a,b,c,d> 5 ,<a,b,c,d,e,b,c,d> 12 ,<a,b,c,d,e,b,c,d,e> ], the business activity set is {a, b, c, d, e}, then the target cutting method includes the cutting position of the business activity set is between a and b, the execution relationship between a and b is a→b, and the two cutting results obtained by cutting the business path set according to the target cutting method are [ 18 ]and[<b,c,d> 5 ,<b,c,d,e,b,c,d> 12 ,<b,c,d,e,b,c,d,e> ]. Accordingly, in order to more clearly and intuitively represent the process tree, Figure 4b In the example, the target cutting method can be expressed as "0.969(→,{a},{b,c,d,e})", and arranged as an internal node in the first layer, where 0.969 is the probability of the target cutting method, which will only contain the cutting results of one business activity [ 18 ] is represented as a and arranged as a leaf node in the second layer. For the second cut, the service path set is [<b,c,d> 5 ,<b,c,d,e,b,c,d> 12 ,<b,c,d,e,b,c,d,e> ], the business activity set is {b,c,d,e}, the target cutting method includes the cutting position of the business activity set is between d and e, and the execution relationship between d and e is The two cutting results obtained by cutting the service path set according to the target cutting method are [<b,c,d> 31 ]and[ <e> 14 ]. Accordingly, in order to more clearly and intuitively represent the process tree,< / e> Figure 4b In the example, the target cutting method can be expressed as "0.981( {b,c,d},{e})", and arranged as internal nodes in the second layer, where 0.981 is the probability of the target cutting method, which will only contain the cutting results of one business activity [ <e> 14 ] is represented as e and arranged as a leaf node in the third layer. For the third cut, the service path set is [<b,c,d> 31 ], the business activity set is {b,c,d}, the target cutting method includes the cutting position of the business activity set is between b and c, the execution relationship between b and c is b→c, and the two cutting results obtained by cutting the business path set according to the target cutting method are [ 31 ]and[<c,d> 31 ]. Accordingly, in order to more clearly and intuitively represent the process tree, Figure 4b In the example, the target cutting method can be expressed as "0.989(→,{b},{c,d})" and arranged as an internal node in the third layer, where 0.989 is the probability of the target cutting method, which will only contain the cutting results of one business activity [ 31 ] is represented as b and arranged as a leaf node in the 4th layer. For the fourth cut, the service path set is [<c,d> 31 ], the business activity set is {c, d}, the target cutting method includes the cutting position of the business activity set is between c and d, the execution relationship between c and d is c→d, and the two cutting results obtained by cutting the business path set according to the target cutting method are [ <c> 31 ]and[ <d> 31 ]. Accordingly, in order to more clearly and intuitively represent the process tree, Figure 4b In the example, the target cutting method can be represented as "0.989(→,{c},{d})" and arranged as an internal node in the fourth layer, where 0.989 is the probability of the target cutting method, which will only contain the cutting results of one business activity [ <c> 31 ] is represented as c, which only contains the cutting result of one business activity [ <d> 31 ] are represented as d and are arranged as leaf nodes in the 5th layer.

[0089] In addition, most RPA products require repeated communication between the process analysts of the RPA products and the users before development to determine the business processes that satisfy the users. If the understanding between the two parties is inconsistent, the users may repeatedly modify the business processes, resulting in a long development cycle of the RPA products and high time costs. Moreover, this process is highly dependent on process analysts and requires high data analysis capabilities of process analysts, so the demand and cost of human resources are high. In response to this problem, in an embodiment of the present application, after obtaining the business process analysis results corresponding to the business log data, the business process analysis results obtained through in-depth and detailed data analysis can be directly sent to the process analysts, so that the process analysts can obtain accurate business processes, and further provide them to the developers of the RPA products for development, thereby reducing the time cost of RPA product development, reducing the burden on process analysts, and reducing labor costs.

[0090] S105: Obtaining a business process prediction result of the target object based on the business process analysis result prediction.

[0091] The business process prediction result is used to characterize the future business process of the target object. The present application embodiment does not specifically limit the method for obtaining the business process prediction result. For ease of understanding, the present application embodiment provides a possible implementation method. For technical details, please refer to the introduction of the embodiment below.

[0092] In an embodiment of the present application, after obtaining the business process prediction results of the target object, the business process prediction results can be sent to users associated with the target object, so that users can better understand the business process and make better decisions based on the business process prediction results, thereby further improving the usability and intelligence of the RPA product.

[0093] Based on the above S101-S105 related contents, it can be known that in the embodiment of the present application, after obtaining the business log data of the target object, the business path set can be extracted based on the business log data, and the business path set can be cut to obtain the cutting result with business activities as the cutting unit, and the cutting result is further analyzed to obtain the business process analysis result corresponding to the business log data, and finally the business process prediction result of the target object is obtained based on the prediction of the business process analysis result. Since the cutting result is obtained by cutting the business path set with business activities as the cutting unit, based on this fine cutting result, in-depth and detailed data analysis can be performed to obtain the business process analysis result, and then this business process analysis result has less information deviation than the directly extracted business log data. In this way, based on this data with reduced information deviation (that is, the business process analysis result) to perform business process prediction, a business process prediction result with higher accuracy can be obtained, thereby improving the accuracy of business process prediction and improving the user experience.

[0094] In order to accurately cut the business path set and obtain a fine cutting result, so as to perform in-depth and detailed data analysis on the cutting result and obtain a business process analysis result with less information deviation, the embodiment of the present application provides a possible implementation method for cutting the business path set (that is, S103), which may specifically include S1031-S1034. S1031-S1034 are described below in combination with the embodiments and drawings:

[0095] Figure 5 A flowchart of a method for implementing cutting of a service path set provided in an embodiment of the present application. Figure 5 As shown, S1031-S1034 may specifically include:

[0096] S1031: Based on the business activities included in the business paths in the business path set and the execution relationships between the business activities, obtain probabilities of multiple different cutting methods.

[0097] Among them, the execution relationship may include a sequential composition relationship, a unique selection relationship, a parallel composition relationship or a cyclic composition relationship.

[0098] The sequential composition relationship (which can be represented by the symbol "→") indicates that business activities are executed in a sequential order. For example, "a→b" means that business activity a is executed first and then business activity b; "b→a" means that business activity b is executed first and then business activity a.

[0099] The unique selection relationship (which can be represented by the symbol "×") indicates that only one of the business activities is selected for execution. For example, "a×b" means executing one of business activities a and business activity b.

[0100] The parallel composition relationship (which can be represented by the symbol "∧") indicates that there is no requirement for the execution order between business activities. For example, "a∧b" means that business activity a can be executed first and then business activity b, or business activity b can be executed first and then business activity a.

[0101] Circular composition relationship (can be used with the symbol Indicates that business activities are executed cyclically. For a cyclic composition relationship, the business activities in the cyclic composition relationship can be divided into the loop body and the redo activity. For example, Indicates that the loop body is business activity a, the redo part is business activity b, and the loop method is to execute business activity a first, then business activity b, and then business activity a again. Furthermore, the loop body can contain one or more business activities; for a loop body containing only one business activity, the symbol Indicates; for a loop body containing multiple business activities, the symbol For example, Indicates that the loop body contains only business activity a, and business activity b is a redo activity; Indicates that the loop body contains business activity a and other business activities, and business activity b is a redo activity. In addition, for the loop body, the business activities it contains can be divided into start activities, intermediate activities, and end activities. For example, if the loop body only contains business activity a, then the start activity, intermediate activity, and end activity are all a; if the loop body contains business activities b and business activities c, then the start activity may be business activity b, the end activity may be business activity c, and there is no intermediate activity; if the loop body contains business activities b, business activity c, and business activity d, then the start activity is b, the intermediate activity is b, and the end activity is c.

[0102] In addition, the present application embodiment does not limit the implementation method of obtaining the probability of multiple different cutting modes (ie, S3031). For ease of understanding, the cutting position is set between the first business activity and the second business activity in the business activity set as an example, and a possible implementation method is described.

[0103] In a possible implementation manner, S1031 may specifically include steps 11 to 14:

[0104] Step 11: Determine a first execution probability between the first business activity and the second business activity under a plurality of different execution relationships.

[0105] Optionally, step 11 may specifically include: obtaining the average value between the number of executions of the first business activity in the business path set and the number of executions of the second business activity in the business path set; and calculating the first execution probability under multiple different execution relationships based on the average value and based on multiple different execution relationships.

[0106] For example, if the service path set is [<a,b,c,d> 5 ,<a,b,c,d,e,b,c,d> 12 ,<a,b,c,d,e,b,c,d,e> ], where a is the first business activity and b is the second business activity. The first business activity a is executed 18 times in the business path set, and the second business activity b is executed 18 times in the business path set. Correspondingly, the average of the two is 18 times.

[0107] Optionally, for ease of understanding, the present application embodiment illustrates the calculation method of the first execution probability in the form of the following Table 1, which specifically defines the first execution probabilities under multiple different execution relationships corresponding to the first business activity and the second business activity under different business path relationships. Wherein, the first business activity is represented as a, and the second business activity is represented as b, then different business path relationships may include "(nothing)", There are 9 types, and the specific differences are described below. In addition, the embodiment of the present application does not specifically limit the form of the calculation method of the first execution probability.

[0108] Table 1

[0109]

[0110] Where a represents the first business activity; b represents the second business activity; z represents the number of times the first business activity is executed in the business path set and the number of times the second business activity is executed in the business path set; P×(a,b) represents the probability that a and b are in a unique selection relationship; P→(a,b) represents the probability that a and b are in a sequential composition relationship and a is executed first and then b; P→(b,a) represents the probability that b and a are in a sequential composition relationship and b is executed first and then a; It indicates the probability that a and b are in a cyclic composition relationship, and the loop body is a and the redo part is b; It indicates the probability that a and b are in a cyclic composition relationship, the loop body is a and other business activities, and the redo part is b; It indicates the probability that b and a are in a cyclic composition relationship, the loop body is b and other business activities, and the redo part is a; P∧(a,b) indicates the probability that a and b are in a parallel composition relationship;

[0111] (nothing) means that there is no direct path relationship between a and b in the business path; Indicates that a is executed first and then b in the business path, and other business activities are executed between a and b; Indicates that b is executed first and then a in the business path, and other business activities are executed between b and a; Indicates that both Execute again Indicates that a is executed first and then b in the business path; Indicates that both Execute again Indicates that b is executed first and then a in the business path; Indicates that both Execute again Indicates that both Execute again

[0112] Step 12: Based on the cut position and the business activity set, determine a first business activity set including the first business activity and determine a second business activity set including the second business activity.

[0113] For example, when the cutting position is set between the first business activity and the second business activity in the business activity set, if the business activity set is {a, b, c, d, e}, where a is the first business activity and b is the second business activity, then based on the cutting position and the business activity set, the first business activity set is determined to be {a} and the second business activity set is determined to be {b, c, d, e}.

[0114] Step 13: Based on the first execution probabilities, the first business activity set, and the second business activity set under multiple different execution relationships, obtain second execution probabilities between the first business activity set and the second business activity set under multiple different execution relationships.

[0115] For business activities under a loop composition relationship, it is necessary to distinguish between the loop body and the redo part. Therefore, for the implementation of step 13, the embodiment of the present application can be explained separately according to different execution relationship types.

[0116] When the execution relationship is a sequential synthesis relationship, a unique selection relationship or a parallel synthesis relationship, step 13 may specifically include: obtaining the number of business activities included in the first business activity set and the number of business activities included in the second business activity set; determining the second execution probability between the first business activity set and the second business activity set under the execution relationship according to the first execution probability under the execution relationship, the number of business activities included in the first business activity set and the number of business activities included in the second business activity set, and based on a pre-set probability calculation formula.

[0117] The pre-set probability calculation formula can be found in the following formula (1).

[0118]

[0119] Among them, Σ 1 is the first business activity set, Σ 2 is the second business activity set, is the second execution probability between the first business activity set and the second business activity set in a sequential composition relationship, a unique selection relationship, or a parallel composition relationship, |Σ 1 | is the number of business activities included in the first business activity set, |Σ 2 | is the number of business activities included in the second business activity set, a is the first business activity, b is the second business activity, It is the first execution probability between the first business activity and the second business activity under a sequential composition relationship, a unique selection relationship or a parallel composition relationship.

[0120] When the execution relationship is a loop synthesis relationship, step 13 may specifically include: obtaining the number of business activities included in the first business activity set and the number of business activities included in the second business activity set; obtaining the probability of the first business activity set being a loop body and the probability of the second business activity set being a loop body based on the first execution probability, the first business activity set, the second business activity set, and the number of business activities included in the first business activity set and the number of business activities included in the second business activity set under the execution relationship; determining the loop body from the first business activity set or the second business activity set according to the maximum value of the obtained probability of being a loop body; and taking the maximum value as the second execution probability between the first business activity set and the second business activity set under the execution relationship.

[0121] For example, if the business activity set is {a, b, c, d}, where a is the first business activity and b is the second business activity, and the execution relationship between the two is a loop composition relationship, then the first business activity set is {a} and the second business activity set is {b, c, d}. Based on the first execution probability under the execution relationship, the first business activity set {a}, the second business activity set {b, c, d}, and the number of business activities included in the first business activity set {a} and the number of business activities included in the second business activity set {b, c, d}, the probability of the first business activity set {a} being a loop body and the probability of the second business activity set {b, c, d} being a loop body can be obtained. Furthermore, the loop body can be determined from the first business activity set {a} or the second business activity set {b, c, d} according to the maximum value of the two probabilities (that is, if the probability of the first business activity set {a} being the loop body is high, then the first business activity set {a} is determined as the loop body; if the probability of the second business activity set {b, c, d} being the loop body is high, then the second business activity set {b, c, d} is determined as the loop body), and the maximum value is used as the second execution probability between the first business activity set {a} and the second business activity set {b, c, d} under the execution relationship.

[0122] Among them, the calculation of the probability that the first business activity set and the second business activity set are respectively used as loop bodies may specifically include: determining, based on the first execution probability under the execution relationship, the first business activity set and the second business activity set, the first probability of associating the start activity, the second probability of associating the end activity and the third probability of associating the intermediate activities in the loop body when the first business activity set is used as the loop body, and the third probability of associating the start activity, the fourth probability of associating the end activity and the fifth probability of associating the intermediate activities in the loop body when the second business activity set is used as the loop body; obtaining the probability that the first business activity set is used as the loop body based on the first probability, the second probability, the third probability, and the number of business activities included in the first business activity set and the number of business activities included in the second business activity set, and obtaining the probability that the second business activity set is used as the loop body based on the fourth probability, the fifth probability, the sixth probability, and the number of business activities included in the first business activity set and the number of business activities included in the second business activity set.

[0123] Specifically, the probability that the first business activity set includes a loop body, and the probability that the second business activity set includes a loop body can be obtained by the following formula (2).

[0124]

[0125]

[0126]

[0127]

[0128] Among them, ∑ 1 is the first business activity set, ∑ 2 is the second business activity set, S 1 is the start activity of the loop body, E 1 The end activity of the loop body, is the probability that the first business activity set or the second business activity set is a loop body, |∑ 1 | is the number of business activities included in the first business activity set, |∑ 2 | is the number of business activities included in the second business activity set, redo start The probability associated with starting the activity (ie, the first probability or the fourth probability), redo end is the probability of the end activity association (that is, the second probability or the fifth probability), indirect is the probability of the intermediate activity association (that is, the third probability or the sixth probability), a is the first business activity, b is the second business activity, L is the business path that needs to be cut at that time, and End(L) is the end part of the business path that needs to be cut at that time. is the probability that a and b are in a cyclic composition relationship, the loop body is a and other business activities, and the redo part is b. Start(L) is the starting part of the business path that needs to be cut at that time. is the probability that a and b are in a loop composition relationship, with the loop body being a and the redo part being b.

[0129] Step 14: Using the second execution probabilities under multiple different execution relationships as the probabilities of multiple different cutting methods.

[0130] Based on the relevant contents of steps 11 to 14 above, it can be known that through the cutting positions corresponding to the business activity set and the execution relationship between the business activities, multiple different cutting methods of the business path set can be accurately obtained, so as to further determine the target cutting method from the multiple different cutting methods and cut the business path set in the target cutting method.

[0131] S1032: Determine a target cutting method from a plurality of different cutting methods according to a maximum value among the acquired probabilities.

[0132] S1033: Cut the service path set according to the target cutting mode to obtain multiple cutting results corresponding to the service path set.

[0133] Here, multiple cutting results refer to cutting results that only contain one business activity. Among them, by cutting the business path set once according to the target method, two cutting results can be obtained. If the two cutting results each contain only one business activity, the cutting is completed; if the two cutting results contain a business path composed of multiple business activities, it is necessary to continue cutting until all cutting results contain only one business activity. This application does not make specific restrictions on the implementation process that needs to continue cutting. For ease of understanding, it is explained below in conjunction with a possible implementation method.

[0134] In one possible implementation, step S1033 may specifically include: when the cutting result includes a business path composed of multiple business activities, taking the cutting result as a new business path set and redetermining the target cutting method; continuing to cut the new business path set with the redetermined target cutting method, and executing in a loop until all cutting results contain only one business activity.

[0135] Based on the relevant contents of S1031-S1033 above, it can be known that after obtaining the business log data of the target object, a business path set can be extracted based on the business log data, and the business path set can be cut. Among them, in order to accurately cut the business path set, the cutting process can be executed repeatedly until a fine cutting result is obtained (that is, a cutting result with business activities as cutting units), so that the business process analysis result can be obtained by further in-depth and meticulous data analysis based on this fine cutting result, and finally the business process prediction result of the target object can be obtained based on the prediction of the business process analysis result. In this way, compared with the directly extracted business log data, there is less information bias. Based on the data with reduced information bias (that is, the business process analysis result) for business process prediction, the accuracy of business process prediction can be improved, and the user experience can be improved.

[0136] In order to obtain accurate business process prediction results, the embodiment of the present application also provides an implementation method for obtaining the business process prediction results of the target object (that is, S105), which specifically includes S1051-S1054. S1051-S1054 are described below in combination with the embodiments and drawings:

[0137] Figure 6 A flowchart of an implementation method for obtaining a business process prediction result of a target object provided in an embodiment of the present application. Figure 6 As shown, S1051-S1054 may specifically include:

[0138] S1051: Determine influencing characteristics of business process analysis results based on business log data.

[0139] The influencing features of the business process analysis results refer to the features that can affect the business process analysis results. Taking the new employee onboarding scenario as an example, the features that affect the time it takes for the new employee to onboard may include whether the new employee can onboard on time, whether the company has sufficient housing when arranging dormitories, and whether the company has sufficient computer inventory.

[0140] S1052: extracting multiple business paths with consistent influencing characteristics from the business path set, and constructing a sample set corresponding to the consistent influencing characteristics based on the multiple business paths.

[0141] For example, if the business path set includes 7 business paths, and the business activity results of 4 of the business paths are affected by consistent impact features, then a sample set is constructed based on the 4 business paths. The number of consistent impact features can be one or more.

[0142] S1053: Determine the information gain of the consistent influencing feature based on the sample set and the consistent influencing feature.

[0143] The information gain of the influencing feature can characterize the decision-making ability of the influencing feature. The larger the information gain, the stronger the decision-making ability; the smaller the information gain, the weaker the decision-making ability. In addition, the embodiment of the present application does not specifically limit the method for determining the information gain of the consistent influencing feature. For ease of understanding, the embodiment of the present application is described in conjunction with a possible implementation method.

[0144] In a possible implementation manner, S1053 may specifically include steps 21-23:

[0145] Step 21: Obtain the entropy of the sample set and the conditional entropy of the consistent influencing features.

[0146] Among them, the process of obtaining the entropy of the sample set can specifically include: obtaining multiple business activity results corresponding to multiple business paths; determining the first occurrence probability of the multiple business activity results in the sample set; and obtaining the entropy of the sample set based on the first occurrence probability and a pre-set entropy calculation formula for the sample set.

[0147] The business activity result refers to the result of the business activity that is ultimately executed by the business path. In conjunction with the above example of the new employee onboarding scenario, if the business activity that is ultimately executed by the business path is sending an onboarding notification to the new employee, the business activity result may be the time when the new employee is notified of onboarding. If the business activity that is ultimately executed by the business path is verifying the new employee's materials, the business activity result may be whether the new employee is allowed to onboard.

[0148] The entropy calculation formula of the pre-set sample set can be found in the following formula (3).

[0149]

[0150] Where D represents the sample set, H(D) is the entropy of sample set D, |D| is the total number of business paths in sample set D, k represents the result of business activity, |C k | is the number of business paths corresponding to business activity result k, is the first occurrence probability of business activity result k in sample set D.

[0151] The process of obtaining the conditional entropy of a consistent influencing feature may specifically include: obtaining multiple different influencing results corresponding to the consistent influencing feature; respectively determining the entropies of the multiple different influencing results, and the second occurrence probabilities of the multiple different influencing results in the sample sets that respectively match the multiple different influencing results; obtaining the conditional entropy of the consistent influencing feature based on the entropies of the multiple influencing results and the multiple second occurrence probabilities and based on a pre-set conditional entropy calculation formula for the influencing feature.

[0152] In combination with the above-mentioned example of the new employee onboarding scenario, if the consistent influencing features include whether the new employee can start work on time, whether the company has sufficient housing when arranging dormitories, and whether the company has sufficient computer inventory, then the impact results corresponding to the influencing feature that the new employee can start work on time may include on time starting work and delayed starting work; the impact results corresponding to the influencing feature whether the company has sufficient housing when arranging dormitories may include no housing required, sufficient housing, and insufficient housing; the impact results corresponding to the influencing feature whether the company has sufficient computer inventory may include sufficient computer inventory and insufficient computer inventory.

[0153] In addition, the embodiment of the present application can determine the entropies of multiple different impact results based on a pre-set entropy calculation formula for the impact result. The pre-set entropy calculation formula for the impact result can be found in the following formula (4).

[0154]

[0155] Where D represents the sample set, T n Indicates the impact result, H(D|T n ) is the influence result T n The entropy of |T n | is the impact result T in sample set D n The total number of matching business paths, t represents the business activity result, |C t |To affect the results T n The number of business paths corresponding to the matching business activity result t, The result of the business activity t is related to the impact result T n The probability of occurrence in the matching business path.

[0156] In addition, the pre-set conditional entropy calculation formula for the influencing feature can be found in the following formula (5).

[0157]

[0158] Where D represents the sample set, T represents the consistent influencing feature, H(D|T) is the conditional entropy of the consistent influencing feature T, and T n Indicates the impact results, To influence the results T n Influencing the results T n The second occurrence probability in the matching sample set, H(D|T n ) is the influence result T n The entropy of .

[0159] Step 22: Determine the information gain of the consistent influencing feature based on the entropy of the sample set and the conditional entropy of the consistent influencing feature.

[0160] The embodiment of the present application can determine the information gain of the consistent influencing feature based on a pre-set information gain calculation formula. The pre-set information gain calculation formula can be found in the following formula (6).

[0161] Gain(D,T)=H(D)-H(D|T) (6)

[0162] Where D represents the sample set, T represents the consistent influencing feature, Gain(D,T) is the information gain of the consistent influencing feature T, H(D) is the entropy of the sample set D, and H(D|T) is the conditional entropy of the consistent influencing feature T.

[0163] Step 23: If the number of consistent influencing features is at least three, determine the target influencing feature from multiple consistent influencing features according to the maximum value in the determined information gain, and determine a new influencing feature other than the target influencing feature from multiple consistent influencing features; extract the business path matching the target influencing result corresponding to the target influencing feature from the sample set, construct a new sample set, and obtain the entropy of the new sample set and the conditional entropy of the new influencing feature, continue to determine the information gain of the new influencing feature with the entropy of the new sample set and the conditional entropy of the new influencing feature, and execute the loop until the number of new influencing features finally determined is one.

[0164] Combined with the above example of the new employee onboarding scenario, if the consistent influencing features include whether the new employee can start work on time, whether the company has sufficient housing when arranging dormitories, and whether the company has sufficient computer inventory, and the information gain of the influencing feature of whether the new employee can start work on time is the largest, then the influencing feature of whether the new employee can start work on time is used as the target influencing feature, and the two influencing features of whether the company has sufficient housing when arranging dormitories and whether the company has sufficient computer inventory are used as new influencing features. Further, the business path that matches the target influencing result of on-time onboarding is extracted from the sample set, a new sample set is constructed, and the entropy of the new sample set and the conditional entropy of the new influencing feature are obtained, so as to continue to determine the information gain of the new influencing feature using the entropy of the new sample set and the conditional entropy of the new influencing feature. Since the number of the above consistent influencing features is three, it is only necessary to perform the information gain determination process twice to determine the final information gain of the three consistent influencing features.

[0165] Based on the relevant contents of steps 21-23 above, it can be known that in an embodiment of the present application, by calculating the entropy of the sample set and the conditional entropy of the consistent influencing features, the information gain of the consistent influencing features is obtained, which facilitates the subsequent use of the information gain as the basis for constructing a decision tree model, accurately establishing a decision tree model, and thus obtaining a more accurate business process prediction result.

[0166] S1054: Based on information gain, a decision tree model for predicting the business process is established to predict the business process prediction result.

[0167] When the number of consistent influencing features is one, the decision tree model only needs to construct a decision node corresponding to the consistent influencing feature to predict the business process prediction result. When the number of consistent influencing features is multiple, the decision tree model needs to construct decision nodes in sequence according to the size relationship of information gain. In this regard, for ease of understanding, the embodiment of the present application is described in combination with a possible implementation method for the case where the number of consistent influencing features is multiple.

[0168] In one possible implementation, S1054 may specifically include: when there are multiple consistent influencing features, sequentially constructing multiple decision nodes of the consistent influencing features in order of information gain from large to small; generating a decision tree model based on multiple decision nodes and multiple different influencing results corresponding to the multiple consistent influencing features.

[0169] Specifically, refer to the above example of the new employee onboarding scenario and combine it with Figure 7 As shown, the embodiment of the present application provides a representation form of a decision tree model for predicting business processes based on information gain. Among them, multiple consistent influencing features include whether the new employee can start work on time, whether the company has sufficient housing when arranging dormitories, and whether the company's computer inventory is sufficient. If the information gain of the influencing feature of whether the new employee can start work on time is the largest, the influencing feature of whether the new employee can start work on time is used as the decision node for the initial construction. Further, the two influencing features of whether the company has sufficient housing when arranging dormitories and whether the company's computer inventory is sufficient are used as new influencing features, and the business path matching the target influencing result of on-time entry is extracted from the sample set to construct a new sample set. The entropy of the new sample set is obtained, as well as the conditional entropy of the two influencing features of whether the company has sufficient housing when arranging dormitories and whether the company's computer inventory is sufficient in the case of on-time entry, so as to continue to determine the information gain of the new influencing feature using the entropy of the new sample set and the conditional entropy of the new influencing feature (that is, the information gain of the two influencing features of whether the company has sufficient housing when arranging dormitories and whether the company's computer inventory is sufficient in the case of on-time entry). If the maximum value between the two is the information gain of the influencing feature of whether the company's computer inventory is sufficient when the employee starts work on time, then the information gain of the influencing feature of whether the company's computer inventory is sufficient when the employee starts work on time will be used as the decision node for the second construction. Correspondingly, the information gain of the influencing feature of whether the company has sufficient housing when arranging dormitories when the employee starts work on time will be used as the decision node for the third construction.

[0170] Based on the relevant contents of S1051-S1054 above, it can be known that after obtaining the data with reduced information bias (that is, the business process analysis results), business process prediction can be performed based on the business process analysis results, wherein the information gain of the influencing features is used as the basis for constructing the decision tree model to clarify the primary and secondary relationships between the influencing features, thereby further improving the accuracy of business process prediction and improving the user experience.

[0171] Based on the data analysis method provided in the above embodiment, the embodiment of the present application also provides a data analysis device, which is explained and illustrated below in conjunction with the accompanying drawings.

[0172] Figure 8 A schematic diagram of the structure of a data analysis device provided in an embodiment of the present application. Figure 8 As shown, the data analysis device 800 provided in the embodiment of the present application may include:

[0173] Data acquisition module 801, used to acquire business log data of the target object;

[0174] A path extraction module 802 is used to extract a business path set based on business log data; the business path set includes a business path, and the business path includes at least two business activities;

[0175] The path cutting module 803 is used to cut the business path set and obtain the cutting result with the business activity as the cutting unit;

[0176] The cutting result analysis module 804 is used to analyze the cutting result to obtain the business process analysis result corresponding to the business log data;

[0177] The process prediction module 805 is used to predict the business process prediction result of the target object based on the business process analysis result.

[0178] As an implementation method, in order to improve the accuracy of RPA products in business process prediction, the path cutting module 803 may specifically include:

[0179] A first cutting module is used to obtain the probabilities of multiple different cutting modes based on the business activities included in the business paths in the business path set and the execution relationship between the business activities; the cutting mode includes the execution relationship and the cutting position of the business activity set corresponding to the business path set; the business activity set includes all the business activities included in the business path set;

[0180] A second cutting module, used for determining a target cutting method from a plurality of different cutting methods according to a maximum value among the acquired probabilities;

[0181] The third cutting module is used to cut the business path set according to the target cutting method to obtain two or more cutting results corresponding to the business path set, and each cutting result only contains one business activity.

[0182] As an implementation method, in order to improve the accuracy of RPA products in business process prediction, the third cutting module is specifically used to:

[0183] When the cutting result contains a business path composed of multiple business activities, the cutting result is used as a new business path set and the target cutting method is re-determined;

[0184] The new business path set is cut continuously in the re-determined target cutting mode, and the process is repeated until all the cutting results contain only one business activity.

[0185] As an implementation mode, in order to improve the accuracy of RPA products in business process prediction, if the cutting position is set between the first business activity and the second business activity in the business activity set, the first cutting module may specifically include:

[0186] A fourth segmentation module, configured to determine a first execution probability between the first business activity and the second business activity under a plurality of different execution relationships;

[0187] a fifth cutting module, configured to determine, based on the cutting position and the business activity set, a first business activity set including the first business activity and a second business activity set including the second business activity;

[0188] A sixth cutting module, configured to obtain, based on the first execution probabilities under multiple different execution relationships, the first business activity set, and the second business activity set, a second execution probability between the first business activity set and the second business activity set under multiple different execution relationships;

[0189] The seventh cutting module is used to use the second execution probabilities under multiple different execution relationships as the probabilities of multiple different cutting methods.

[0190] As an implementation method, in order to improve the accuracy of RPA products in business process prediction, the fourth cutting module can be specifically used to:

[0191] Obtain an average value between the number of executions of the first business activity in the business path set and the number of executions of the second business activity in the business path set;

[0192] According to the average value and based on the multiple different execution relationships, the first execution probabilities under the multiple different execution relationships are calculated.

[0193] As an implementation method, in order to improve the accuracy of RPA products in business process prediction, the segmentation result analysis module 604 may specifically include:

[0194] The first analysis module is used to arrange all target cutting methods and all cutting results containing only one business activity in layers according to the order of cutting to obtain business process analysis results.

[0195] As an implementation method, in order to improve the accuracy of RPA products in business process prediction, the process prediction module 605 may specifically include:

[0196] A first prediction module, used to determine the impact characteristics of the business process analysis results based on the business log data;

[0197] The second prediction module is used to extract multiple business paths with consistent influencing characteristics from the business path set, and construct a sample set corresponding to the consistent influencing characteristics based on the multiple business paths;

[0198] A third prediction module is used to determine the information gain of the consistent impact feature based on the sample set and the consistent impact feature;

[0199] The fourth prediction module is used to establish a decision tree model for predicting business processes based on information gain, so as to predict business process prediction results.

[0200] As an implementation method, in order to improve the accuracy of RPA products in business process prediction, the fourth prediction module can be specifically used to:

[0201] When there are multiple consistent influencing features, multiple decision nodes of consistent influencing features are constructed in order from large to small information gain;

[0202] The decision tree model is generated based on multiple decision nodes and multiple different impact results corresponding to multiple consistent impact features.

[0203] As an implementation mode, in order to improve the accuracy of RPA products in business process prediction, the data analysis device 600 may further include:

[0204] The desensitizing module is used to respond to a desensitizing request for sensitive data in the business log data, perform desensitizing processing on the sensitive data, and obtain desensitized business log data.

[0205] Furthermore, an embodiment of the present application also provides a device, including: a processor, a memory, and a system bus;

[0206] The processor and the memory are connected via the system bus;

[0207] The memory is used to store one or more programs, and the one or more programs include instructions. When the instructions are executed by the processor, the processor executes any one of the implementation methods of the above-mentioned data analysis method.

[0208] Furthermore, an embodiment of the present application also provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed on a terminal device, the terminal device executes any one of the implementation methods of the above-mentioned data analysis method.

[0209] It can be known from the description of the above implementation mode that those skilled in the art can clearly understand that all or part of the steps in the above-mentioned embodiment method can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., including several instructions for enabling a computer device (which can be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.

[0210] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0211] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0212] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.< / d> < / c> < / d> < / c> < / e>

Claims

1. A data analysis method, characterized in that: include: Obtain the business log data of the target object; Extracting a business path set based on the business log data; the business path set includes a business path, the business path refers to a business activity actually performed for the target object, and the business path includes at least two business activities; Cutting the business path set to obtain a cutting result with business activities as cutting units; Analyzing the cutting result to obtain a business process analysis result corresponding to the business log data; Predicting a business process prediction result of the target object based on the business process analysis result; The segmenting of the business path set to obtain a segmentation result with business activities as the segmentation unit includes: Based on the business activities included in the business paths in the business path set and the execution relationship between the business activities, the probability of obtaining multiple different cutting modes; the cutting mode includes the execution relationship and the cutting position of the business activity set corresponding to the business path set; the business activity set includes all business activities included in the business path set; Determining a target cutting method from the multiple different cutting methods according to a maximum value among the acquired probabilities; The business path set is cut according to the target cutting method to obtain a plurality of cutting results corresponding to the business path set, and each of the cutting results only includes one business activity.

2. The method according to claim 1, characterized in that The cutting of the service path set according to the target cutting mode to obtain a plurality of cutting results corresponding to the service path set includes: When the cutting result includes a business path composed of multiple business activities, the cutting result is used as a new business path set and a target cutting method is re-determined; The new business path set is continuously cut in the re-determined target cutting manner, and the process is executed in a loop until all cutting results contain only one business activity.

3. The method according to claim 1, characterized in that If the cutting position is set between the first business activity and the second business activity in the business activity set, the probability of obtaining a plurality of different cutting modes based on the business activities included in the business paths in the business path set and the execution relationship between the business activities includes: determining a first execution probability between the first business activity and the second business activity under a plurality of different execution relationships; Based on the cutting position and the business activity set, determining a first business activity set including the first business activity and determining a second business activity set including the second business activity; Based on the first execution probabilities under the multiple different execution relationships, the first business activity set, and the second business activity set, obtaining second execution probabilities between the first business activity set and the second business activity set under the multiple different execution relationships; The second execution probabilities under the multiple different execution relationships are used as the probabilities of the multiple different cutting modes.

4. The method according to claim 3, characterized in that The determining a first execution probability between the first business activity and the second business activity under a plurality of different execution relationships includes: Obtaining an average value between the number of executions of the first business activity in the business path set and the number of executions of the second business activity in the business path set; According to the average value and based on the multiple different execution relationships, a first execution probability under the multiple different execution relationships is calculated.

5. The method according to claim 1, characterized in that The analyzing the segmentation result to obtain the business process analysis result corresponding to the business log data includes: All target cutting methods and all cutting results that only include one business activity are arranged in layers according to the order of cutting to obtain the business process analysis result.

6. The method according to any one of claims 1 to 5, characterized in that: The predicting of the business process prediction result of the target object based on the business process analysis result includes: Determine the impact characteristics of the business process analysis result based on the business log data; Extracting multiple business paths with consistent influencing characteristics from the business path set, and constructing a sample set corresponding to the consistent influencing characteristics according to the multiple business paths; Determining the information gain of the consistent influencing feature based on the sample set and the consistent influencing feature; Based on the information gain, a decision tree model for predicting the business process is established to predict the business process prediction result.

7. The method according to claim 6, characterized in that The step of establishing a decision tree model for predicting a business process based on the information gain includes: When there are multiple consistent influencing features, sequentially constructing multiple decision nodes of the consistent influencing features in descending order of information gain; The decision tree model is generated based on the plurality of decision nodes and the plurality of different impact results respectively corresponding to the plurality of consistent impact features.

8. The method according to any one of claims 1 to 5, characterized in that: Before obtaining the business log data of the target object, the data analysis method further includes: In response to a desensitization request for sensitive data in the business log data, the sensitive data is desensitized to obtain desensitized business log data.

9. A data analysis device, characterized in that: include: A data acquisition module is used to acquire the business log data of the target object; A path extraction module, configured to extract a business path set based on the business log data; the business path set includes a business path, the business path refers to a business activity actually performed for the target object, and the business path includes at least two business activities; A path cutting module, used to cut the business path set to obtain a cutting result with business activities as cutting units; A cutting result analysis module, used for analyzing the cutting result to obtain a business process analysis result corresponding to the business log data; A process prediction module, used to predict the business process prediction result of the target object based on the business process analysis result; Wherein, the path cutting module specifically includes: A first cutting module is used to obtain the probabilities of multiple different cutting modes based on the business activities included in the business paths in the business path set and the execution relationship between the business activities; the cutting mode includes the execution relationship and the cutting position of the business activity set corresponding to the business path set; the business activity set includes all the business activities included in the business path set; A second cutting module, used for determining a target cutting method from a plurality of different cutting methods according to a maximum value among the acquired probabilities; The third cutting module is used to cut the business path set according to the target cutting method to obtain two or more cutting results corresponding to the business path set, and each cutting result only contains one business activity.

10. A data analysis device, characterized in that: The device includes: a processor, a memory, and a system bus; The processor and the memory are connected via the system bus; The memory is used to store one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by the processor, the processor executes the method according to any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a terminal device, the terminal device executes the method according to any one of claims 1 to 8.

Images