CN116610814A - Knowledge graph-based RPA intelligent driving method and system - Google Patents
Knowledge graph-based RPA intelligent driving method and system Download PDFInfo
- Publication number
- CN116610814A CN116610814A CN202310616951.0A CN202310616951A CN116610814A CN 116610814 A CN116610814 A CN 116610814A CN 202310616951 A CN202310616951 A CN 202310616951A CN 116610814 A CN116610814 A CN 116610814A
- Authority
- CN
- China
- Prior art keywords
- activity
- rpa
- activities
- flow
- knowledge
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 138
- 230000000694 effects Effects 0.000 claims abstract description 874
- 230000008569 process Effects 0.000 claims abstract description 113
- 230000009471 action Effects 0.000 claims description 36
- 230000006399 behavior Effects 0.000 claims description 25
- 230000003993 interaction Effects 0.000 claims description 12
- 230000004927 fusion Effects 0.000 claims description 9
- 238000012163 sequencing technique Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 abstract description 13
- 241000699666 Mus <mouse, genus> Species 0.000 description 17
- 238000012544 monitoring process Methods 0.000 description 7
- 241000699670 Mus sp. Species 0.000 description 5
- 150000001875 compounds Chemical class 0.000 description 4
- 238000004883 computer application Methods 0.000 description 4
- 238000004801 process automation Methods 0.000 description 4
- FFBHFFJDDLITSX-UHFFFAOYSA-N benzyl N-[2-hydroxy-4-(3-oxomorpholin-4-yl)phenyl]carbamate Chemical compound OC1=C(NC(=O)OCC2=CC=CC=C2)C=CC(=C1)N1CCOCC1=O FFBHFFJDDLITSX-UHFFFAOYSA-N 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000013467 fragmentation Methods 0.000 description 1
- 238000006062 fragmentation reaction Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
An RPA intelligent driving method and system based on a knowledge graph, the method comprises the following steps: step one: acquiring an RPA business process, and constructing an activity knowledge graph according to the RPA business process; step two: according to the RPA business process name input by the user, matching the current activity corresponding to the RPA business process name from the activity knowledge graph; step three: determining whether the current activity is a key activity, if so, giving a next instruction by a user after the execution of the key activity is finished, and if not, repeating the second step after the execution of the current activity is finished; step four: according to the activity knowledge graph, utilizing knowledge reasoning to match an entity with an association relationship with the current activity; step five: and calculating the similarity between the automation requirement content input by the user and the entity with the association relation. The invention can be used for solving the risk problem encountered by the RPA technology in automation.
Description
Technical Field
The invention belongs to the technical field of artificial intelligence, and particularly relates to an RPA intelligent driving method and system based on a knowledge graph.
Background
With the rapid development of information technology, humans are moving into the information age, where computer applications play an important role. As the demand for information services by humans in every industry has increased dramatically, the scale of computer applications has been expanding, the types have been increasingly diversified, and the functions have been more complex. However, in the process of providing information service, various industries always generate some flows with simple operation rules and high repeatability, so that the user can repeatedly and fussy steps when operating the application program, and the manpower and time are wasted to a certain extent. The traditional flow operation often needs staff to frequently perform operations such as opening and closing of computer application programs, reading and inputting of data and the like, so that the work efficiency of enterprises is reduced, and the error rate of manual operation is improved.
The robot process automation (Robotic Process Automation, RPA) is a process automation management software that automates the operation of the business according to preset process steps. Along with the popularization of computer application programs, the RPA technology brings new opportunities for the operation of the traditional flow, and a plurality of enterprises introduce the RPA technology to improve the production flow so as to further improve the working efficiency, reduce the operation error rate and save manpower and material resources.
However, in the process of combining the conventional RPA technology with artificial intelligence, the intelligent processing when the behavior which does not meet the requirement is encountered in the RPA business process is lacking, so that the risk encountered in the automation of the RPA technology cannot be avoided or solved; the lack of an intelligent driving method for optimization from the point of execution logic of the RPA itself results in that each business process needs to independently formulate a dedicated robot.
Disclosure of Invention
The invention aims to provide an RPA intelligent driving method and system based on a knowledge graph, which solve the problems that the RPA technology encounters risks in automation and each business process needs to independently formulate a proprietary robot.
In order to achieve the above purpose, the present invention adopts the following technical scheme: an RPA intelligent driving method based on a knowledge graph comprises the following steps:
step one: and acquiring an RPA business process, and constructing an activity knowledge graph according to the RPA business process.
The activity knowledge graph comprises: a plurality of triples, each triplet including a head entity and a tail entity formed by an activity and a relationship formed by RPA business process names to which the activity belongs; the activity is obtained after splitting the RPA business process; the activities include critical activities.
Step two: and matching the current activity corresponding to the RPA business process name from the activity knowledge graph according to the RPA business process name input by the user.
Step three: determining whether the current activity is a key activity or not, if so, giving a next instruction by a user after the key activity is executed; and if the activity is not the key activity, repeating the step two after the current activity is finished.
Step four: and according to the activity knowledge graph, utilizing knowledge reasoning to match the entity with the association relation with the current activity.
Step five: and calculating the similarity between the automation requirement content input by the user and the entities with the association relationship, and sequencing the similarity according to the order from high to low, wherein the activities corresponding to the 5 entities with the association relationship with the similarity at the front are used as recommended activities.
Further, the first step includes:
and 11, acquiring an RPA service flow.
Step 12: splitting the RPA business process into a series of activities; wherein the series of activities includes a key activity, the key activity being a key activity obtained by selecting a portion of the series of activities as a key activity to be marked.
Step 13: constructing a triplet according to a series of activities, and generating an activity knowledge graph according to the triplet; in the triplets, the head entity and the tail entity are determined according to the execution sequence of a series of split activities, and the relation between the head entity and the tail entity is determined according to the RPA business process to which the activities belong.
Further, the second step includes:
step 21: and according to the RPA business process name input by the user, matching a head entity corresponding to the RPA business process name from the activity knowledge graph.
Step 22: and matching tail entities corresponding to the head entities according to the head entities and the RPA business process names, and taking the activities corresponding to the tail entities as the current activities.
Further, the next step instruction in the third step includes: continuing to execute the activity, withdrawing and re-searching the activity, and ending the RPA service flow; if the current automation behaviour meets the user's requirements, the user will give instructions to continue the execution of the activity, step 22 is repeated; if the current automation behavior does not meet the user requirement, the user gives a command of withdrawing and searching again, then withdrawing the automation behavior of the last key activity from the current automation behavior, and requesting the user to input the automation requirement content at the moment; if the current automation behavior meets the user requirement, the user gives an instruction for ending the RPA business flow, and all the automation behaviors are stopped.
Further, the third step further comprises: if not a critical activity, step 22 is repeated after the automated content of the activity is executed.
Further, in the fourth step, the entity having the association relationship with the current activity is matched by utilizing knowledge reasoning, and the corresponding matching rule is as follows:
if there are triples (Activity A, process n, activity B) and triples (Activity B, process m, activity C), then: the association relation exists between the activity A and the activity C, and the association relation is a flow m.
Further, the fifth step further comprises:
the user selects one activity from the recommended activities, marks the selected activity as a key activity, determines the next activity according to the activity knowledge graph, takes the next activity as the current activity, and repeats the step three; and stopping all automatic actions after the user gives an instruction for ending the RPA business flow until the selected recommended activity is executed.
The invention also adopts another technical scheme that: an RPA intelligent drive system based on a knowledge graph, comprising: the system comprises a knowledge storage module, a knowledge matching module, an intelligent interaction module, a knowledge reasoning module and a knowledge recommendation module.
The knowledge storage module is used for storing the activity knowledge graph; the activity knowledge graph is obtained according to the following method:
splitting the RPA business process into a series of activities, dividing the series of activities into a head entity and a tail entity according to the execution sequence, constructing a triplet by taking the RPA business process to which the activities belong as a relation, and constructing an activity knowledge graph according to the triplet; and meanwhile, the activities needing to be introduced into the intelligent interaction module are marked as key activities.
And the knowledge matching module is used for matching the current activity corresponding to the RPA business process name from the activity knowledge graph according to the RPA business process name input by the user.
The intelligent interaction module is used for determining whether the current activity is a key activity or not, and if the current activity is the key activity, after the execution of the key activity is finished, the user gives a next instruction; if not, the automation content of the current activity is executed.
And the knowledge reasoning module is used for matching the entity with the association relation with the current activity when the current activity which does not meet the user requirement appears in the RPA business process.
And the knowledge recommendation module is used for calculating the similarity by using a fuzzy matching algorithm according to the activity requirement provided by the user and the entities with the association relationship, sequencing the similarity according to the sequence from high to low, and taking the activities corresponding to the 5 entities with the association relationship with the similarity at the front as recommended activities.
Further, the knowledge storage module generates an active knowledge graph using a knowledge fusion algorithm for all triples.
Further, the instructions of the next step include: continue to perform activities, withdraw and re-search for instructions of the end of the activity or RPA business process.
The beneficial effects of the invention are as follows:
(1) The invention splits the RPA business process into a series of activities, takes the activities as entities and utilizes the activities to construct an activity knowledge graph. By combining intelligent interaction and activity knowledge graph in the RPA business process, the risk encountered by the RPA technology in automation, namely the behavior which does not meet the requirement, is solved.
(2) The invention splits the RPA business process into a series of activities and combines with the knowledge graph to realize the decomposition of the RPA business process into the form of fragmentation. When the behavior which does not meet the requirement appears in the RPA business flow, the activities meeting the requirement can be automatically recommended. Meanwhile, by combining the activities in various ways, RPA business processes meeting various functional requirements can be created, so that the requirements of an automation scene can be better met. Therefore, the risk encountered by the RPA technology in automation and the RPA business processes with different functional requirements are combined in a diversified way can be solved by storing the automation behavior into the knowledge graph and combining the intelligent processing in the automation operation.
Drawings
FIG. 1 is a diagram of an RPA intelligent drive system based on a knowledge graph;
FIG. 2 is a flow chart of a method of the knowledge-based RPA intelligent drive system of the present invention;
Fig. 3 is an active knowledge graph of an embodiment of the present invention.
Detailed Description
The following description of the technical solutions in the embodiments of the present invention will be clear and complete, and it is obvious that the described embodiments are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, the present invention provides an RPA intelligent driving system based on a knowledge graph, including: the system comprises a knowledge storage module, a knowledge matching module, an intelligent interaction module, a knowledge reasoning module and a knowledge recommendation module.
The knowledge storage module is used for storing the activity knowledge graph; the activity knowledge graph is obtained according to the following method:
splitting the RPA business process into a series of activities, dividing the series of activities into a head entity and a tail entity according to the execution sequence, constructing a triplet by taking the RPA business process to which the activities belong as a relation, and constructing an activity knowledge graph according to the triplet; and meanwhile, the activities needing to be introduced into the intelligent interaction module are marked as key activities.
And the knowledge matching module is used for matching the current activity corresponding to the RPA business process name from the activity knowledge graph according to the RPA business process name input by the user.
The intelligent interaction module is used for determining whether the current activity is a key activity or not, and if the current activity is the key activity, after the execution of the key activity is finished, the user gives a next instruction; if not, the automation content of the current activity is executed.
And the knowledge reasoning module is used for matching the entity with the association relation with the current activity when the current activity which does not meet the user requirement appears in the RPA business process.
And the knowledge recommendation module is used for calculating the similarity by using a fuzzy matching algorithm according to the activity requirement provided by the user and the entities with the association relationship, sequencing the similarity according to the sequence from high to low, and taking the activities corresponding to the 5 entities with the association relationship with the similarity at the front as recommended activities.
As shown in fig. 2, an RPA intelligent driving method based on a knowledge graph specifically includes the following steps:
step one: and acquiring an RPA business process, and constructing an activity knowledge graph according to the RPA business process.
Specifically, the RPA business process is split into a series of activities, the series of activities are divided into a head entity and a tail entity according to the execution sequence, triples are constructed by taking the RPA business process to which the activities belong as a relation, then a knowledge fusion algorithm is used for all triples to generate an activity knowledge graph, and finally key activities are marked in the series of activities of the RPA business process. The final constructed active knowledge graph is shown in fig. 3.
The following details how step one is implemented, which includes the steps of:
and 11, acquiring an RPA service flow.
Specifically, the RPA service flow is a preset flow step, and may specifically include a plurality of flows, for example: scheme 1, scheme 2, scheme 3 … scheme n.
Step 12: splitting the RPA business process into a series of activities; wherein the series of activities includes a key activity, the key activity being a key activity obtained by selecting a portion of the series of activities as a key activity to be marked.
Specifically, table 1 is an activity sequence in the RPA business process, for example: the invention splits the process 1 into a series of activities which can be specifically: activity a, activity b, activity c, activity d, activity e, activity f, activity g. Splitting the flow 2 into an activity a, an activity d, an activity h, an activity t, an activity i, an activity w and an activity o.
Then, a part of activities are selected from a series of activities corresponding to the flow 1, the flow 2 and the flow 3 to be marked as key activities, for example: activity c, activity d, activity m, activity t, activity w, activity g, activity o, activity q are labeled as critical activities.
Table 1 active sequence in RPA business process
Step 13: constructing a triplet according to a series of activities, and generating an activity knowledge graph according to the triplet; in the triplets, the head entity and the tail entity are determined according to the execution sequence of a series of split activities, and the relation between the head entity and the tail entity is determined according to the RPA business process to which the activities belong.
Specifically, the expression of the triplet may be: (Activity A, process n, activity B), wherein Activity A, activity B are head entity and tail entity obtained according to activity in Process n in order of execution; the flow n is the corresponding relation between the activity A and the activity B, and the relation is determined according to the RPA business flow names of the activity A and the activity B. The business process name is: one of the flow 1, the flow 2, and the flow 3 … flow n is a flow n corresponding to the business flow name in the present embodiment.
Table 2 shows the expression form of the triplet set in each RPA service flow, and as shown in Table 2, the triplet is constructed in the following way: the activities in the flow 1, the flow 2 and the flow 3 are respectively used as a head entity and a tail entity according to the execution sequence, and the RPA business flow names of the two activities are used as the relation of the two entities, so as to construct the triples.
TABLE 2 triplet sets in each RPA service flow
The independent operations of the 20 times of mouse and keyboard in a series of independent operations of the mouse and keyboard obtained through monitoring are sequentially split and stored as an action attribute in one entity created by python by using the py2neo library (when the rest is less than 20 times of independent operations, all the rest is stored as an action attribute in one entity), and for simplicity, the entity names are exemplified by an activity a, an activity b and the like. The sequence of activities in the RPA traffic is shown in table 1.
The activities in the flow 1, the flow 2 and the flow 3 are respectively used as a head entity and a tail entity according to the execution sequence, and the RPA business flow names of the two activities are used as the relation of the two activities, so as to construct a triplet, as shown in the table 2.
The triples in table 2 are used to generate an active knowledge graph using a knowledge fusion algorithm, as shown in fig. 2. The knowledge fusion specifically includes calculating the similarity between the names of the entities through the Levenshtein distance (when the names of the entities are English, the names of the entities are automatically translated into Chinese by using a translation library), namely, the minimum editing distance, namely, performing entity alignment on the entities with the minimum editing distance of 0, wherein the minimum editing distance refers to the minimum editing operation times required by converting one character string into another character string, and the editing operation includes inserting one character, deleting one character and replacing one character; when there are multiple relationships between two entities, a relationship merge is required, such as (entity 1, relationship 1, entity 2), (entity 1, relationship 2, entity 2) … (entity 1, relationship n, entity 2), and the relationship merge is (entity 1, relationship 1/relationship 2/./relationship n, entity 2). Wherein for two strings A, B, the Levenshtein distance for the first i characters of string a and the first j characters of string B satisfy the following formula:
Wherein I (A) i ≠B j ) Is an indication function, and when the ith character of the character string A and the jth character of the character string B are different, the value is 1; otherwise, 0.
Each RPA business process stored by the activity knowledge graph in the neo4j database is sequentially executed by using a python pyautogui, pynput database, namely, the automation content of the action attribute of the entity contained in each RPA business process is executed, and after each activity is executed, the operation is stopped for 60s, the withdrawal operation of the current activity is completed in the period, meanwhile, the independent operation of a mouse and a keyboard is monitored by using a pyautogui, pynput database, the independent operation of the mouse and the keyboard is stored in the withdraw attribute of the current entity, and the automation content of the action attribute of the current activity is executed again after 60 s.
The activity needing to be introduced into the intelligent interaction module is marked as a key activity in the activity knowledge graph, and particularly when the action attribute of the entity contains a key automatic behavior, namely, when manual inspection is needed after the execution of the activity is finished, the entity is added with an identifier attribute, and the attribute value is True. Here, add an identifier to activity c, activity d, activity m, activity t, activity w, activity g, activity o, and activity q, and the attribute value is True.
The operation flow corresponding to the execution step one is as follows:
s11: the implementation mode of RPA automation can use UiPath, blue Prism, automation Anywhere and other automation tools, and can also use pyautogui, pynput and other libraries in python to monitor a mouse and a keyboard to realize RPA automation. The invention realizes the RPA business process automation by monitoring the mouse and the keyboard through pyautogui, pynput and other libraries and stores the RPA automation operation in neo4j database. The method specifically comprises the following steps: firstly, inputting the established RPA business flow name, and then realizing the independent operation of monitoring a series of continuous mice and keyboards through pyautogui, pynput and other libraries. Then, the independent operations of the 20 times of mouse and keyboard obtained by monitoring are sequentially split and stored in an entity as an action attribute (when the rest is less than 20 times of independent operations, the rest is stored in an entity as an action attribute), and the entity name is manually input (the entity name is named by the automation content which can be realized by the action attribute of the entity). The RPA business flow names are used as relations, and the entities are sequentially connected according to the sequence of the automatic operation. And splitting and storing the operation content obtained by monitoring the complete operation flow of the mouse and the keyboard, namely splitting the RPA service flow. The purpose is that after the operation is stored in this way, when the subsequent RPA automation is wanted to be executed, the action attribute of the entity is read, and the automation behavior can be completed by utilizing pyautogui, pynput library. The stored information for a single mouse-keyboard independent operation is represented in table 3, and the pyautogui, pynput library can complete the automation operation by using the stored information. Each split entity contains a series of consecutive pieces of these independent operation information. Meanwhile, the first entity in a series of entities constructed by each RPA business process is used as a tail entity and is connected with a head entity named as 'start', and the relation between the first entity and the head entity is the name of the corresponding RPA business process. The independent operation of a series of mice and keyboards contained in the action attribute in each entity is referred to as an activity. An independent operation of the mouse keyboard is shown in table 1 (mouse movement is marked 0, click mouse is marked 10, release mouse is marked 11, scroll wheel is marked 2, special key press is marked 300, character key press is marked 301, special key release is marked 310, and character key release is marked 311). Character keys include letter, number, symbol keys in a keyboard, such as: "a", "b", "1", "@etc., the special keys are all keys except character keys, such as: left Alt key, right Alt key, shift key, etc.
Table 3 stored information for single mouse and keyboard independent operation
S12: and (3) using a knowledge fusion algorithm to generate an active knowledge graph by using the triples contained in all the RPA business processes constructed in the step (S11). Knowledge fusion can be achieved by calculating the similarity between entities, such as by calculating the similarity between entities through a Dice coefficient, a Jaccard coefficient or a Levenshtein distance, and in the invention, the similarity between the names of the entities (when the names of the entities are English, the names of the entities are automatically translated into Chinese by using a translation library), namely the minimum editing distance, is calculated through the Levenshtein distance, and meanwhile, entity alignment is carried out on the entity with the minimum editing distance of 0. The minimum edit distance refers to the minimum number of editing operations required to convert one character string into another, including inserting one character, deleting one character, and replacing one character; when there are multiple relationships between two entities, a relationship merge is required, such as (entity 1, relationship 1, entity 2), (entity 1, relationship 2, entity 2) … (entity 1, relationship n, entity 2), and the relationship merge is (entity 1, relationship 1/relationship 2/./relationship n, entity 2). Wherein for two strings A, B, the Levenshtein distance for the first i characters of string a and the first j characters of string B satisfy the following formula:
Wherein I (A) i ≠B j ) Is an indication function, and when the ith character of the character string A and the jth character of the character string B are different, the value is 1; otherwise, 0.
S13: each RPA business process stored by the activity knowledge graph in the neo4j database is sequentially executed by using a python pyautogui, pynput database, namely, the automation content of the action attribute of the entity contained in each RPA business process is executed, and after each activity is executed, the operation is stopped for 60s, the withdrawal operation of the current activity is completed in the period, meanwhile, the independent operation of a mouse and a keyboard is monitored by using a pyautogui, pynput database, the independent operation of the mouse and the keyboard is stored in the withdraw attribute of the current entity, and the automation content of the action attribute of the current activity is executed again after 60 s.
S14: the activity needing to be introduced into the intelligent interaction module is marked as a key activity in the activity knowledge graph, specifically, when the action attribute of the entity comprises a key automation link, namely, when the artificial examination is needed after the execution of the key activity is finished, the entity is added with an important attribute, and the attribute value is True according to the activity knowledge graph obtained in the step S13.
Step two: and matching the current activity corresponding to the RPA business process name from the activity knowledge graph according to the RPA business process name input by the user.
The following describes how to implement the scheme corresponding to the second step, which specifically includes the following steps:
step 21: and according to the RPA business process name input by the user, matching a head entity corresponding to the RPA business process name from the activity knowledge graph.
Step 22: and matching tail entities corresponding to the head entities according to the head entities and the RPA business process names, and taking the activities corresponding to the tail entities as the current activities.
Step three: determining whether the current activity is a key activity or not, if so, giving a next instruction by a user after the key activity is executed; and if the activity is not the key activity, repeating the step two after the current activity is finished.
Specifically, after the execution of the key activity is finished, a next instruction is given by the user, and the next instruction includes: continuing to perform the activity, withdrawing and rescuing the activity or ending the RPA business process. If the current automation behaviour meets the user's requirements, the user will give instructions to continue the execution of the activity, step 22 is repeated; if the current automation behavior does not meet the user requirement, the user gives a command of withdrawing and searching again, then withdrawing the automation behavior of the last key activity from the current automation behavior, and requesting the user to input the automation requirement content at the moment; if the current automation behavior meets the user requirement, the user gives an instruction for ending the RPA business flow, and all the automation behaviors are stopped. If not a critical activity, step 22 is repeated after the automated content of the activity is executed.
The operation flow corresponding to the second and third execution steps is as follows:
s21: according to the activity knowledge graph of step 1, the user inputs the RPA business process name (the concrete expression form is a character string) to be executed. Querying whether a tail entity exists in the active knowledge graph of the step 1 by using a py2neo library through python, wherein the relation with the start entity is the RPA business process name required to be executed. When the entity is queried through the relationship, the relationship existing between the entities is divided by '/' characters when the compound relationship exists, such as: (entity A, flow 1/flow 2/flow 3, entity B), the relationship of entity A and entity B will be divided by the '/' character into relationship 1, relationship 2, relationship 3, three relationships in total. If the entity does not exist, re-inputting the RPA business process name to be executed; if an entity exists, then automated content of the entity's action attribute is performed through python using a library pyautogui, pynput or the like.
S22: querying whether a tail entity exists in the active knowledge graph of the step 1 by using a py2neo library through python, wherein the relation between the tail entity and the currently active head entity is the RPA business process name to be executed. When the entity is queried through the relationship, the relationship existing between the entities is divided by '/' characters when the compound relationship exists, such as: (entity A, flow 1/flow 2/flow 3, entity B), the relationship of entity A and entity B will be divided by the '/' character into relationship 1, relationship 2, relationship 3, three relationships in total.
S31, if the entity does not belong to the key activity, executing the automation content of the action attribute of the entity by using a pyautogui, pynput library through python and the like, and repeating S22; if the entity belongs to the key activity, executing the automation content of the action attribute of the entity by using a pyautogui, pynput library through python and the like, and giving a next instruction by a user; if the entity does not exist, the activity is the last activity, and the user gives a next instruction. The next step of instruction comprises: continuing to perform the activity, withdrawing and rescuing the activity or ending the RPA business process.
S32: when the current automation activity content meets the user requirement, S22 is repeated after receiving the continuous execution activity instruction in S31.
S33: when the current activity content does not meet the user' S requirement, the withdrawal and re-search activity command in S31 is received, and then the automated content of their widthwart attribute (the widthwart attribute of the entity excluding the last critical activity, withdrawal operation) is sequentially executed from the current entity to the last critical activity entity through the python using pyautogui, pynput or the like library, and the user is required to input the activity requirement content at this time.
S34: when the execution reaches the current activity and meets the user requirement, all the automation actions are stopped after the RPA business flow ending instruction in S31 is received.
Step four: according to the activity knowledge graph, utilizing knowledge reasoning to match an entity with an association relationship with the current activity;
specifically, the invention matches the entity with the association relation with the current activity by utilizing the knowledge reasoning technology.
The operation flow corresponding to the fourth step is as follows:
step 4: and according to the activity knowledge graph, utilizing knowledge reasoning to match the entity with the association relation with the current activity.
S41: after receiving the instruction of withdrawing and searching for the activity again according to S31 and completing the withdrawing operation of S32, using the py2neo library to match the entity having the association relation with the current activity to the activity knowledge graph in the step 1 by using the knowledge reasoning technology based on rules through python, wherein the established rules are as follows:
if triples (activity A, flow n, activity B) and triples (activity B, flow m, activity C) the n triples (activity A, flow m, activity C), namely, activity A and activity C, have an association relationship, wherein the association relationship is flow m.
Step five: and calculating the similarity between the automation requirement content input by the user and the entities with the association relationship, and sequencing the similarity according to the order from high to low, wherein the activities corresponding to the 5 entities with the association relationship with the similarity at the front are used as recommended activities.
Specifically, the provided activity demands are subjected to fuzzy matching with the entities with association relations, and then the recommendation of the next activity is performed based on the similarity.
The operation flow corresponding to the fifth step is as follows:
s51: the fuzzy matching can calculate the similarity between the activity requirement content received in the step 3 and the entity with the association relation in the step 4 through the Dice coefficient, the Jaccard coefficient or the Levenshtein distance. The similarity between the activity demand content received in the step 3 and the entity with the association relation in the step 4 is calculated through the Levenshtein distance.
S52: and (5) sorting from high to low according to the similarity obtained in the step (S51), and recommending the first 5 activities with the highest similarity.
S53: and according to the 5 activities obtained in the step S52, selecting an activity which meets the requirement most (the activity is marked as a key activity, namely, the importent attribute of the entity of the activity is added, the attribute value is True), taking the association relationship between the current entity and the entity which meets the requirement most as the name of the RPA business process which needs to be executed currently, repeating the step S31 until the automatic behavior is stopped after receiving the instruction of ending the RPA business process.
The mouse and keyboard monitoring information of the RPA service flow in the invention is as follows:
And realizing the process 1, the process 2 and the process 3 in the construction of the RPA service process by monitoring a mouse and a keyboard through pyautogui, pynput and other libraries. The independent operation of a series of successive mice and keyboards in the RPA business process is shown in table 4 (since only a very small portion of the mice and keyboards are shown in the content restriction table).
TABLE 4 independent operation of a series of successive mice and keyboards in RPA business process
Example 1
Taking the user input process 1 as an example (scenario: user actual demand is the implementation of activity w).
Step 1, splitting an RPA business process into a series of activities, dividing the activities into a head entity and a tail entity according to an execution sequence, constructing triples by taking the RPA business process to which the activities belong as a relation, generating an activity knowledge graph by using a knowledge fusion algorithm on all triples, and finally marking key activities in the series of activities of the RPA business process, wherein the activities c, the activities d, the activities m, the activities t, the activities w, the activities g, the activities o and the activities q are marked as key activities.
The step 1 specifically comprises the following steps: an active knowledge graph embodiment process constructed with reference to fig. 3.
And 2, matching the current activity corresponding to the RPA business process name from the activity knowledge graph according to the RPA business process name input by the user.
Step 21: and according to the RPA business process name input by the user, matching a head entity corresponding to the RPA business process name from the activity knowledge graph.
Step 22: and matching tail entities corresponding to the head entities according to the head entities and the RPA business process names, and taking the activities corresponding to the tail entities as the current activities.
The step 2 specifically comprises the following steps:
s21: the user enters flow 1 by python using the py2neo library to query the active knowledge-graph shown in fig. 3 for the presence of a tail entity, the relationship to the "start" of the head entity is flow 1. When the entity is queried through the relation, when the compound relation exists, the relation existing between the entities is segmented by the '/' characters, a triplet is queried ("start", flow 1/flow 2/flow 3, activity a), and the automation content of the action attribute of the activity a is executed through python by using a pyautogui, pynput library and the like.
S22: querying whether a tail entity exists in the active knowledge graph of the step 1 by using a py2neo library through python, wherein the relation between the tail entity and the currently active head entity is the RPA business process name to be executed.
Step 3, determining whether the current activity is a key activity, if so, giving a next instruction by a user after the key activity is executed; and if the activity is not the key activity, repeating the step two after the current activity is finished.
The step 3 specifically comprises the following steps: the relationship with the head entity of activity a is flow 1 by using python to query the active knowledge-graph as shown in fig. 3 for the presence of a tail entity using the py2neo library. Querying out triples (activity a, flow 1, activity b), wherein activity b does not belong to a key activity, executing automation content of action attribute of activity b, and repeating S22; querying the entity corresponding to the activity b through the relation to obtain a triplet (activity b, flow 1, activity c), wherein the activity c belongs to a key activity, executing the automation content of the action attribute of the activity c, and repeating S22, wherein the automation content of the activity c accords with the requirement of a user, and the user selects an instruction for continuously executing the activity; querying an entity corresponding to the activity c through the relation to obtain a triplet (activity c, a flow 1 and activity d), wherein the activity d belongs to a key activity, executing the automation content of the action attribute of the activity d, and repeating S22, wherein the automation content of the activity d accords with the requirement of a user, and the user selects an instruction for continuously executing the activity; querying the entity corresponding to the activity d through the relation to obtain a triplet (activity d, flow 1, activity e), wherein the activity e does not belong to a key activity, executing the automation content of the action attribute of the activity e, and repeating S22; querying an entity corresponding to the activity e through the relation to obtain a triplet (activity e, flow 1, activity f), wherein the activity f does not belong to a key activity, executing automation content of the action attribute of the activity f, and repeating S22; and querying the entity corresponding to the activity f through the relation, querying out a triplet (activity f, flow 1, activity g), wherein the activity g belongs to a key activity, executing the automation content of the action attribute of the activity g, and sequentially executing the automation content of the activity g, the activity f, the activity e and the withdraw attribute of the activity g by using a pyautogui, pynput library through python and the like and simultaneously inputting the automation requirement content (namely, the activity w) by a user.
And 4, according to the activity knowledge graph, utilizing knowledge reasoning to match out the entity with the association relation with the current activity.
The step 4 specifically comprises the following steps: and 3, receiving an instruction for withdrawing and searching the activity again according to the step 3, and after the withdrawing operation of the step 3 is completed, matching the entity which has an association relation with the current activity d with the activity knowledge graph in the step 1 by using a py2neo library through python and using a rule-based knowledge reasoning technology. From the triplet (activity d, flow 1, activity e) it can be derived that activity d has an association with activity e: flow 1, simultaneously store ((activity d, flow 1, activity e)) in the now_way attribute of activity e; the association between activity d and activity f, which can be derived from triples (activity d, flow 1, activity e) and (activity e, flow 1, activity f), exists: flow 1, a triplet (activity d, flow 1, activity f), while storing ((activity d, flow 1, activity e), (activity e, flow 1, activity f)) in the now_way attribute of activity f; from the triplet (activity d, flow 1, activity f) and triplet (activity f, flow 1, activity g), it can be derived that there is an association between activity d and activity g: flow 1, a triplet (activity d, flow 1, activity g), while storing ((activity d, flow 1, activity e), (activity e, flow 1, activity f), (activity f, flow 1, activity g)) in the now_way attribute of activity g; from the triplet (activity d, flow 2, activity h), it can be obtained that there is an association relationship between activity d and activity h: flow 2, simultaneously store ((activity d, flow 2, activity h)) in the now_way attribute of activity h; from the triples (activity d, flow 2, activity h) and triples (activity h, flow 2, activity t), it can be derived that there is an association between activity d and activity t: flow 2, a triplet (activity d, flow 2, activity t), while storing ((activity d, flow 2, activity h), (activity h, flow 2, activity t)) in the now_way attribute of activity t; from the triplet (activity d, flow 2, activity t) and triplet (activity t, flow 3, activity m), it can be derived that there is an association between activity d and activity m: flow 3, a triplet (activity d, flow 3, activity m), while storing ((activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 3, activity m)) in the now_way attribute of activity m; from the triplet (activity d, flow 2, activity t) and triplet (activity t, flow 2, activity i), it can be derived that there is an association between activity d and activity t: flow 2, a triplet (activity d, flow 2, activity i), while storing ((activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i)) in the now_way attribute of activity i; from the triples (activity d, flow 2, activity i) and triples (activity i, flow 2/flow 3, activity w) it can be derived that there is an association between activity d and activity w: flow 2/flow 3, i.e., triples (activity d, flow 2/flow 3, activity w), while storing ((activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i), (activity i, flow 2/flow 3, activity w)) in the now_way attribute of activity w; from the triplet (activity d, flow 2/flow 3, activity w) and triplet (activity w, flow 2, activity o), it can be derived that there is an association between activity d and activity o: flow 2, i.e., triplet (activity d, flow 2, activity o), while storing ((activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i), (activity i, flow 2/flow 3, activity w), (activity w, flow 2, activity o)) in the now_way attribute of activity o; from the triplet (activity d, flow 2/flow 3, activity w) and triplet (activity w, flow 3, activity q) it can be derived that there is an association between activity d and activity q: flow 3, i.e., triplet (activity d, flow 3, activity q), is stored in the now_way attribute of activity q at the same time ((activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i), (activity i, flow 2/flow 3, activity w), (activity w, flow 3, activity q)). Namely, the activity having the association relationship with the activity d includes: activity e, activity f, activity g, activity h, activity t, activity i, activity w, activity o, activity q, activity m.
And 5, calculating the similarity between the automation requirement content input by the user and the entities with the association relationship, and sequencing the similarity according to the sequence from high to low, wherein the activities corresponding to the 5 entities with the association relationship with the similarity at the front are used as recommended activities.
The step 5 specifically comprises the following steps:
s51: and calculating the similarity between the activity demand content received in the step 3 and the entity with the association relation in the step 4 through the Levenshtein distance.
S52: the similarity obtained according to step S51 is ranked from high to low, and the first 5 activities with the highest similarity (activity w is the activity with the highest similarity) are recommended.
S53: after selecting activity w according to the 5 activities obtained in step S52 (setting the value of the important attribute of activity w to True), step 3 is sequentially repeated according to the reasoning path sequence in the non-way attribute of activity w ((activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i), (activity i, flow 2/flow 3, activity w)). The method specifically comprises the following steps: querying the entity corresponding to the activity d through the relation (flow 2), querying out a triplet (activity d, flow 2, activity h), wherein the activity h does not belong to a key activity, executing the automation content of the action attribute of the activity h, and repeating S22; querying an entity corresponding to the activity h through a relation (flow 2), querying a triplet (activity h, flow 2, activity t), wherein the activity t belongs to a key activity, executing the automation content of the action attribute of the activity t, and repeating S22, wherein the automation content of the activity t accords with the requirement of a user, and the user selects an instruction for continuously executing the activity; querying an entity corresponding to the activity t through the relation (flow 2), querying out a triplet (activity t, flow 2, activity i), wherein the activity i does not belong to a key activity, executing the automation content of the action attribute of the activity i, and repeating S22; and querying the entity corresponding to the activity i through the relation (flow 2/flow 3), querying out the triples (activity i, flow 2/flow 3, activity w), wherein the activity w belongs to a key activity, executing the automation content of the action attribute of the activity w, and stopping all automation actions by the user selecting the RPA business flow ending instruction because the automation content executed to the activity w meets the user requirement.
Example 2
Taking the user input process 3 as an example (scenario: user actual demand is to implement activity d).
Step 1, splitting an RPA business process into a series of activities, dividing the activities into a head entity and a tail entity according to an execution sequence, constructing triples by taking the RPA business process to which the activities belong as a relation, generating an activity knowledge graph by using a knowledge fusion algorithm on all triples, and finally marking key activities in the series of activities of the RPA business process, wherein the activities c, the activities d, the activities m, the activities t, the activities w, the activities g, the activities o and the activities q are marked as key activities.
The step 1 specifically comprises the following steps: an active knowledge graph embodiment process constructed with reference to fig. 3.
And 2, matching the current activity corresponding to the RPA business process name from the activity knowledge graph according to the RPA business process name input by the user.
The step 2 specifically comprises the following steps:
s21: the user enters flow 3 by python using the py2neo library to query the active knowledge-graph as shown in fig. 3 for the presence of a tail entity, the relationship "beginning" with the head entity is flow 3. When the entity is queried through the relation, when the compound relation exists, the relation existing between the entities is segmented by the '/' characters, a triplet is queried ("start", flow 1/flow 2/flow 3, activity a), and the automation content of the action attribute of the activity a is executed through python by using a pyautogui, pynput library and the like.
S22: querying whether a tail entity exists in the active knowledge graph of the step 1 by using a py2neo library through python, wherein the relation between the tail entity and the currently active head entity is the RPA business process name to be executed.
Step 3, determining whether the current activity is a key activity, if so, giving a next instruction by a user after the key activity is executed; and if the activity is not the key activity, repeating the step two after the current activity is finished.
The step 3 specifically comprises the following steps: the relationship with the head entity of activity a is flow 3 by querying if there is a tail entity in the activity knowledge-graph as shown in fig. 3 using the py2neo library by python. Querying out a triplet (activity a, flow 3, activity c), wherein the activity c belongs to a key activity, executing automation content of the action attribute of the activity c, and repeating S22, wherein the automation content of the activity c accords with user requirements, and the user selects an instruction for continuously executing the activity; and querying the entity corresponding to the activity c through the relation, namely querying out a triplet (activity out c, flow 3 and activity t), wherein the activity t belongs to a key activity, executing the automation content of the action attribute of the activity t, executing the automation content of the withdraw attribute of the activity t through python by using a pyautogui, pynput library and the like, and simultaneously inputting the automation requirement content (namely, activity d) by a user.
And 4, according to the activity knowledge graph, utilizing knowledge reasoning to match out the entity with the association relation with the current activity.
The step 4 specifically includes: and 3, receiving an instruction for withdrawing and searching the activity again according to the step 3, and after the withdrawing operation of the step 3 is completed, matching the entity which has an association relation with the current activity c with the activity knowledge graph in the step 1 by using a py2neo library through python and using a rule-based knowledge reasoning technology. From the triplet (activity c, flow 1, activity d), it can be obtained that there is an association relationship between activity c and activity d: flow 1, simultaneously store ((activity c, flow 1, activity d)) in the now_way attribute of activity d; from the triplet (activity c, flow 1, activity d) and triplet (activity d, flow 1, activity e), it can be derived that there is an association between activity c and activity e: flow 1, a triplet (activity c, flow 1, activity e), while storing ((activity c, flow 1, activity d), (activity d, flow 1, activity e)) in the now_way attribute of activity e; the association between activity c and activity f, which can be derived from triples (activity c, flow 1, activity e) and (activity e, flow 1, activity f), exists: flow 1, i.e., triplet (activity c, flow 1, activity f), while storing ((activity c, flow 1, activity d), (activity d, flow 1, activity e), (activity e, flow 1, activity f)) in the now_way attribute of activity f; from the triplet (activity c, flow 1, activity f) and triplet (activity f, flow 1, activity g), it can be derived that there is an association between activity c and activity g: flow 1, i.e., triplet (activity c, flow 1, activity g), while storing ((activity c, flow 1, activity d), (activity d, flow 1, activity e), (activity e, flow 1, activity f), (activity f, flow 1, activity g)) in the now_way attribute of activity g; from the triples (activity c, flow 1, activity d) and triples (activity d, flow 2, activity h), it can be obtained that there is an association between activity c and activity h: flow 2, a triplet (activity c, flow 2, activity h), while storing ((activity c, flow 1, activity d), (activity d, flow 2, activity h)) in the now_way attribute of activity h; from the triplet (activity c, flow 2, activity h) and triplet (activity h, flow 2, activity t), it can be derived that there is an association between activity c and activity t: flow 2, a triplet (activity c, flow 2, activity t), while storing ((activity c, flow 1, activity d), (activity d, flow 2, activity h), (activity h, flow 2, activity t)) in the now_way attribute of activity t; from the triplet (activity c, flow 2, activity t) and triplet (activity t, flow 3, activity m), it can be derived that there is an association between activity c and activity m: flow 3, a triplet (activity c, flow 3, activity m), while storing ((activity c, flow 1, activity d), (activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 3, activity m)) in the now_way attribute of activity m; from the triplet (activity c, flow 2, activity t) and triplet (activity t, flow 2, activity i), it can be derived that there is an association between activity c and activity i: flow 2, i.e., triplet (activity c, flow 2, activity i), while storing ((activity c, flow 1, activity d), (activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i)) in the now_way attribute of activity m; from the triples (activity c, flow 2, activity i) and triples (activity i, flow 2/flow 3, activity w) it can be derived that there is an association between activity c and activity w: flow 2/flow 3, i.e., triples (activity c, flow 2/flow 3, activity w), while storing ((activity c, flow 1, activity d), (activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i), (activity i, flow 2/flow 3, activity w)) in the now_way attribute of activity w; from the triplet (activity c, flow 2/flow 3, activity w) and triplet (activity w, flow 2, activity o), it can be derived that there is an association between activity c and activity o: flow 2, i.e., triplet (activity c, flow 2, activity o), while storing ((activity c, flow 1, activity d), (activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i), (activity i, flow 2/flow 3, activity w), (activity w, flow 2, activity o)) in the now_way attribute of activity o; from the triplet (activity c, flow 2/flow 3, activity w) and triplet (activity w, flow 3, activity q) it can be derived that there is an association between activity c and activity q: flow 3, i.e., triplet (activity c, flow 3, activity q), is simultaneously stored ((activity c, flow 1, activity d), (activity d, flow 2, activity h), (activity h, flow 2, activity t), (activity t, flow 2, activity i), (activity i, flow 2/flow 3, activity w), (activity w, flow 3, activity q)) in the now_way attribute of activity q. Namely, the activity having the association relationship with the activity c includes: activity d, activity e, activity f, activity g, activity h, activity t, activity i, activity w, activity o, activity q, activity m.
Step 5, calculating the similarity between the automation requirement content input by the user and the entities with the association relationship, and sequencing the similarity according to the order from high to low, wherein the activities corresponding to the 5 entities with the association relationship with the similarity at the front are used as recommended activities
The step 5 specifically comprises the following steps:
s51: and calculating the similarity between the activity demand content received in the step 32 and the entity with the association relation in the step 4 through the Levenshtein distance.
S52: the obtained similarity is ranked from high to low according to S51, and the first 5 activities with the highest similarity (activity d is the activity with the highest similarity) are recommended.
S53: after selecting the activity d (setting the value of the important attribute of the activity d to True) according to the 5 activities obtained in step S52, step 3 is sequentially repeated in the order of the inference paths in the non_way attribute of the activity d ((activity c, flow 1, activity d)). The method specifically comprises the following steps: and querying an entity corresponding to the activity c through a relation (flow 1), querying out a triplet (activity c, flow 1, activity d), wherein the activity d belongs to a key activity, executing the automation content of the action attribute of the activity d, and stopping all automation actions as the automation content executed to the activity d meets the requirement of a user, and the user selects an RPA business flow ending instruction.
While the present invention has been described in detail through the foregoing description of the preferred embodiment, it should be understood that the foregoing description is not to be considered as limiting the invention. Many modifications and substitutions of the present invention will become apparent to those of ordinary skill in the art upon reading the foregoing. Accordingly, the scope of the invention should be limited only by the attached claims.
Claims (10)
1. The RPA intelligent driving method based on the knowledge graph is characterized by comprising the following steps of:
step one: acquiring an RPA business process, and constructing an activity knowledge graph according to the RPA business process;
the activity knowledge graph comprises: a plurality of triples, each triplet including a head entity and a tail entity formed by an activity, and a relationship formed by RPA business process names to which the activity belongs; the activity is obtained after splitting the RPA business process; the activities include critical activities;
step two: according to the RPA business process name input by the user, matching the current activity corresponding to the RPA business process name from the activity knowledge graph;
step three: determining whether the current activity is a key activity or not, if so, giving a next instruction by a user after the key activity is executed; if the activity is not the key activity, repeating the second step after the current activity is executed;
Step four: according to the activity knowledge graph, utilizing knowledge reasoning to match out an entity with an association relation with the current activity;
step five: and calculating the similarity between the automation requirement content input by the user and the entities with the association relationship, and sequencing the similarity according to the sequence from high to low, wherein the activities corresponding to the 5 entities with the association relationship with the similarity at the front are used as recommended activities.
2. The knowledge-graph-based RPA intelligent driving method according to claim 1, wherein the first step includes:
step 11, obtaining an RPA service flow;
step 12: splitting the RPA business process into a series of activities; wherein the series of activities includes a key activity, and the key activity is obtained by selecting a part of activities in the series of activities as key activities to mark;
step 13: constructing a triplet according to the series of activities, and generating an activity knowledge graph according to the triplet; in the triplet, a head entity and a tail entity are determined according to the split series of activities and the execution sequence, and the relation between the head entity and the tail entity is determined according to the RPA business process to which the activities belong.
3. The knowledge-graph-based RPA intelligent driving method according to claim 1, wherein the second step includes:
step 21: according to the RPA business process name input by the user, matching a head entity corresponding to the RPA business process name from the activity knowledge graph;
step 22: and matching tail entities corresponding to the head entities according to the head entities and the RPA business process names, and taking the activities corresponding to the tail entities as current activities.
4. The knowledge-graph-based RPA intelligent driving method according to claim 1, wherein the next step instruction in the third step includes: continuing to execute the activity, withdrawing and re-searching the activity, and ending the RPA service flow; if the current automation behaviour meets the user's requirements, the user will give instructions to continue the execution of the activity, step 22 is repeated; if the current automation behavior does not meet the user requirement, the user gives a command of withdrawing and searching again, then withdrawing the automation behavior of the last key activity from the current automation behavior, and requesting the user to input the automation requirement content at the moment; if the current automation behavior meets the user requirement, the user gives an instruction for ending the RPA business flow, and all the automation behaviors are stopped.
5. The knowledge-graph-based RPA intelligent driving method according to claim 1, wherein the third step further comprises: if not a critical activity, step 22 is repeated after the automated content of the activity is executed.
6. The knowledge graph-based RPA intelligent driving method according to claim 1, wherein in the fourth step, the entity having the association relationship with the current activity is matched by using knowledge reasoning, and the matching rule corresponding to the entity is:
if there are triples (Activity A, process n, activity B) and triples (Activity B, process m, activity C), then: the association relation exists between the activity A and the activity C, and the association relation is a flow m.
7. The knowledge-graph-based RPA intelligent driving method according to claim 1, wherein the fifth step further comprises:
selecting one activity from the recommended activities by the user, marking the selected activity as a key activity, determining the next activity according to the activity knowledge graph, taking the next activity as the current activity, and repeating the step three; and stopping all automatic actions after the user gives an instruction for ending the RPA business flow until the selected recommended activity is executed.
8. A driving system of the knowledge-graph-based RPA intelligent driving method according to any one of claims 1 to 7, comprising: the system comprises a knowledge storage module, a knowledge matching module, an intelligent interaction module, a knowledge reasoning module and a knowledge recommendation module;
The knowledge storage module is used for storing an active knowledge graph; the activity knowledge graph is obtained according to the following method:
splitting an RPA business process into a series of activities, dividing the series of activities into a head entity and a tail entity according to an execution sequence, constructing a triplet by taking the RPA business process to which the activities belong as a relation, and constructing the activity knowledge graph according to the triplet; meanwhile, marking the activities needing to be introduced into the intelligent interaction module as key activities;
the knowledge matching module is used for matching the current activity corresponding to the RPA business process name from the activity knowledge graph according to the RPA business process name input by the user;
the intelligent interaction module is used for determining whether the current activity is a key activity or not, and if the current activity is the key activity, after the execution of the key activity is finished, the user gives a next instruction; if not, executing the automation content of the current activity;
the knowledge reasoning module is used for matching an entity with an association relation with the current activity when the current activity which does not meet the user requirement appears in the RPA business process;
and the knowledge recommendation module is used for calculating the similarity by using a fuzzy matching algorithm according to the activity requirement provided by the user and the entities with the association relationship, sequencing the similarity according to the sequence from high to low, and taking the activities corresponding to the 5 entities with the association relationship with the similarity at the front as recommended activities.
9. The knowledge-based RPA intelligent drive system of claim 8, wherein the knowledge storage module generates an active knowledge-graph using a knowledge fusion algorithm for all triples.
10. The knowledge-graph based RPA intelligent drive system of claim 8, wherein the next step of instructions comprises: continue to perform activities, withdraw and re-search for instructions of the end of the activity or RPA business process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310616951.0A CN116610814A (en) | 2023-05-29 | 2023-05-29 | Knowledge graph-based RPA intelligent driving method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310616951.0A CN116610814A (en) | 2023-05-29 | 2023-05-29 | Knowledge graph-based RPA intelligent driving method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116610814A true CN116610814A (en) | 2023-08-18 |
Family
ID=87679753
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310616951.0A Pending CN116610814A (en) | 2023-05-29 | 2023-05-29 | Knowledge graph-based RPA intelligent driving method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116610814A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117592622A (en) * | 2024-01-19 | 2024-02-23 | 西南科技大学 | Robot flow automation-oriented behavior sequence prediction method and system |
-
2023
- 2023-05-29 CN CN202310616951.0A patent/CN116610814A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117592622A (en) * | 2024-01-19 | 2024-02-23 | 西南科技大学 | Robot flow automation-oriented behavior sequence prediction method and system |
| CN117592622B (en) * | 2024-01-19 | 2024-04-30 | 西南科技大学 | Robot flow automation-oriented behavior sequence prediction method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cai et al. | An encoder-decoder framework translating natural language to database queries | |
| US7716253B2 (en) | Centralized KPI framework systems and methods | |
| CN110532019B (en) | Method for tracing history of software code segment | |
| CN113379399B (en) | RPA component recommendation method based on state transition probability model | |
| WO2014200725A1 (en) | Smart selection engine | |
| CN116610814A (en) | Knowledge graph-based RPA intelligent driving method and system | |
| CN110362596A (en) | A kind of control method and device of text Extracting Information structural data processing | |
| CN113254507A (en) | Intelligent construction and inventory method for data asset directory | |
| Das et al. | MyNLIDB: a natural language interface to database | |
| CN116244333A (en) | Database query performance prediction method and system based on cost factor calibration | |
| US20180157706A1 (en) | Joining tables by leveraging transformations | |
| TWI436222B (en) | Real - time multi - dimensional analysis system and method on cloud | |
| de Moura et al. | Extracting new metrics from version control system for the comparison of software developers | |
| Zehtaban et al. | Systematic functional analysis methods for design retrieval and documentation | |
| CN115334179B (en) | Unknown protocol reverse analysis method based on named entity recognition | |
| Ilieva et al. | AI Integration in Business Processes Management | |
| CN114399277B (en) | System for assisting concept designer in customizing product concept design | |
| CN109344239A (en) | A kind of business process model querying method and inquiry system based on temporal aspect | |
| CN115329753A (en) | Intelligent data analysis method and system based on natural language processing | |
| CN113885937A (en) | RPA element repairing method based on universality | |
| Padhi et al. | FlashProfile: Interactive Synthesis of Syntactic Profiles. | |
| CN116010444B (en) | Low-code interactive graph query statement construction method | |
| CN113486630B (en) | Supply chain data vectorization and visualization processing method and device | |
| US20220156022A1 (en) | Print recording system and method for controlling print recording system | |
| Atanassov | Comparative analysis of contemporary case based reasoning software frameworks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |