CN115167846A - Recommendation method of downstream operator, electronic device and computer-readable storage medium - Google Patents
Recommendation method of downstream operator, electronic device and computer-readable storage medium Download PDFInfo
- Publication number
- CN115167846A CN115167846A CN202211091774.0A CN202211091774A CN115167846A CN 115167846 A CN115167846 A CN 115167846A CN 202211091774 A CN202211091774 A CN 202211091774A CN 115167846 A CN115167846 A CN 115167846A
- Authority
- CN
- China
- Prior art keywords
- operator
- target
- template
- operators
- determining
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000012216 screening Methods 0.000 claims abstract description 20
- 239000013598 vector Substances 0.000 claims description 41
- 238000004590 computer program Methods 0.000 claims description 9
- 238000004891 communication Methods 0.000 claims description 6
- 238000011144 upstream manufacturing Methods 0.000 claims description 6
- 230000004044 response Effects 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 10
- 238000013473 artificial intelligence Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 8
- 238000001514 detection method Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/30—Creation or generation of source code
- G06F8/35—Creation or generation of source code model driven
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- User Interface Of Digital Computer (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application discloses a recommendation method of a downstream operator, electronic equipment and a computer readable storage medium, wherein the recommendation method comprises the following steps: acquiring a first target operator with an output flow to be determined in at least one operator for establishing a connection relation, wherein the connection relation between the at least one operator is input externally; determining the recommendation score of each template operator in the operator set, wherein the recommendation score of the template operator is determined according to the first target operator and the template operator; screening template operators in the operator set according to the recommendation scores; and generating an operator recommendation list for indicating downstream operators which can be connected with the first target operator according to the screened template operators. The recommendation method provided by the application can generate the operator recommendation list, so that a designer can preferentially search the downstream operator of the first target operator from the operator recommendation list, the search amount of the designer can be reduced, and errors of the designer during search are reduced.
Description
Technical Field
The application belongs to the technical field of artificial intelligence, and particularly relates to a recommendation method of a downstream operator, electronic equipment and a computer-readable storage medium.
Background
In the field of algorithm scheme modeling, generally, one algorithm scheme is realized by forming one or even dozens of different operators through a certain rule, and the operators bear respective business functions and influence each other.
In the process of constructing the scheme, a designer needs to select a proper operator first and then establish connection among the operators, when the operators are more, the process of constructing the scheme needs to spend a large amount of time, errors are easy to occur, and the working efficiency is reduced.
Disclosure of Invention
The application provides a recommendation method of a downstream operator, an electronic device and a computer-readable storage medium, which can intelligently generate an operator recommendation list for a designer to select a proper downstream operator, and reduce the operation complexity of the designer.
A first aspect of an embodiment of the present application provides a method for recommending a downstream operator, where the method includes: acquiring a first target operator with an output flow to be determined in at least one operator for establishing a connection relation, wherein the connection relation among the at least one operator is input externally; determining a recommendation score for each template operator in a set of operators, wherein the recommendation score for the template operator is determined from the first target operator and the template operator; screening the template operators in the operator set according to the recommendation scores; and generating an operator recommendation list for indicating downstream operators which can be connected with the first target operator according to the screened template operators.
A second aspect of the embodiments of the present application provides an electronic device, which includes a processor, a memory, and a communication circuit, where the processor is respectively coupled to the memory and the communication circuit, the memory stores program data, and the processor executes the program data in the memory to implement the steps in the foregoing method.
A third aspect of embodiments of the present application provides a computer-readable storage medium, which stores a computer program, the computer program being executable by a processor to implement the steps in the above method.
The beneficial effects are that: according to the method, the recommendation score of each template operator in the operator set is generated firstly, then the template operators in the operator set are screened according to the recommendation scores, finally, the operator recommendation list is generated according to the screened template operators, so that a designer can preferentially search the downstream operator of the first target operator from the operator recommendation list, the search amount of the designer can be reduced, and errors of the designer in searching can be reduced.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings used in the description of the embodiments are briefly introduced below, it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without inventive efforts, wherein:
FIG. 1 is an interface diagram of an artificial intelligence platform of the present application;
FIG. 2 is a schematic flow chart diagram illustrating an embodiment of a method for recommending downstream operators according to the present application;
FIG. 3 is a schematic flow chart of step S120 in FIG. 2;
FIG. 4 is a schematic flow chart of step S140 in FIG. 2;
FIG. 5 is a schematic flow chart of step S142 in FIG. 4;
FIG. 6 is a schematic diagram of an embodiment of an electronic device of the present application;
FIG. 7 is a schematic diagram of another embodiment of an electronic device of the present application;
FIG. 8 is a schematic structural diagram of an embodiment of a computer-readable storage medium of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without making any creative effort belong to the protection scope of the present application.
It should be noted that the terms "first" and "second" in the present application are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implying any number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present application, "plurality" means at least two, e.g., two, three, etc., unless explicitly specifically limited otherwise. Furthermore, the terms "include" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements listed, but may alternatively include other steps or elements not listed, or inherent to such process, method, article, or apparatus.
Before introducing the scheme of the application, firstly, the artificial intelligence platform of the application is briefly introduced:
firstly, a designer can construct a corresponding scheme by using a plurality of algorithm components on an artificial intelligence platform (which is a platform for developing a machine learning model) according to task requirements, wherein each algorithm component generally comprises information such as a model with a specific algorithm function, a modeling code of the model, a data set used by a training model, a deployment code of the model and the like, and the algorithm components are called operators in the application. For example, a target detection function, a target tracking function, a target recognition function, or the like may be realized by one operator.
Meanwhile, the application provides a dragging type artificial intelligence platform, and in combination with fig. 1, when a designer builds a scheme on the artificial intelligence platform, firstly a scheme creation instruction is input, and a creation interface (indicated by reference numeral 10 in fig. 1) is entered. The creating interface is divided into a plurality of sub-regions, wherein the sub-region with the reference number of 101 is an information configuration region, in the sub-region, a designer can configure the scheme information and the operator information in detail, wherein when the scheme information is selected to configure the scheme information, the name of the scheme and the service application scene of the configuration scheme can be configured, for example, the service application scene of the configuration scheme is access detection, attendance opening or an intelligent cell, or an equipment platform supported or selected by the scheme can be configured (the equipment platform supported by the scheme is an equipment platform capable of applying the scheme, and the equipment platform selected by the scheme is an equipment platform to which the scheme is to be applied), or an authorized person of the scheme can be configured, that is, a person allowed to use the scheme is configured. When the selected operator information column is used for carrying out information configuration on the selected operator, information such as a developer, a maintainer, an application scene, input data, output data and the like of the operator can be configured. Of course, other information of the schemes and operators may also be configured, and are not illustrated here.
The sub-region denoted by reference numeral 102 shows an operator list, which may include all operators, or only the device platform selected by the solution or the operators supported by the device platforms supported by the solution. Wherein the operators in the operator list may have been classified according to the relevant function, for example in fig. 1, the operator list comprises 5 sets of operators, each set of operators being capable of implementing a different function, wherein there are operator 1, operator 2 and operator 3 implementing the target tracking function.
When a designer builds a scheme, operators are selected from the sub-regions with the reference number of 102 in sequence, a dragging operation is performed, the selected operators are dragged to the sub-regions with the reference number of 103, and after the designer drags the operators to the sub-regions with the reference number of 103 each time, the connection relation between the just dragged operator and the operator dragged before needs to be established, namely the operator dragged just receives output data of the operator. It should be noted that after the operator is dragged from the sub-region denoted by reference numeral 102 to the sub-region denoted by reference numeral 103, the operator can still be displayed in the sub-region denoted by reference numeral 102, and in the following dragging process, the designer can still drag the operator from the sub-region denoted by reference numeral 102 to the sub-region denoted by reference numeral 103 again.
That is, the flow of the designer building scheme is:
the first step is as follows: the designer selects an operator from the sub-region denoted by reference numeral 102 and drags the operator to the sub-region denoted by reference numeral 103, taking the operator as a starting node of the entire scheme, that is, the operator receives input data of the scheme;
the second step is that: the designer selects an operator from the sub-region denoted by reference numeral 102;
the third step: dragging the selected operator to a sub-region with the reference number 103 by a designer, and establishing a connection relation between the operator and the operator in the sub-region with the reference number 103, namely determining which operator in the sub-region with the reference number 103 receives the output data of the operator;
and repeating the second step and the third step until all the operators dragged and connected form a complete scheme, namely the operator at the tail end outputs output data of the scheme.
Through the scheme, when all the operators which are dragged and connected are established to form a complete scheme, a designer packages files, wherein the packaged files comprise all the operators in the complete scheme and the connection relations which are established among all the operators by the designer. And then the designer sends the packaged file to corresponding service equipment, such as a camera, and the service equipment runs the file after receiving the file, so that the complete scheme runs on the service equipment, and further corresponding functions are realized.
It can be seen from the above that, each time the designer needs to select and drag an operator from the sub-region denoted by reference numeral 102, when there are many operators displayed in the sub-region denoted by reference numeral 102, the designer needs to spend a certain time searching for a proper operator, which is time-consuming and prone to error for the designer.
Therefore, in order to avoid the above drawbacks, the present application provides the following:
referring to fig. 2, in an embodiment of the present application, a method for recommending a downstream operator includes:
s110: and acquiring a first target operator with an output flow to be determined in at least one operator for establishing a connection relation, wherein the connection relation between at least one operator is input externally.
Specifically, the first target operator refers to an operator whose output flow is to be determined in at least one operator for which a connection relationship is established, that is, the operator for receiving the output data of the first target operator is selected and dragged next by the designer. The object of the present application is to recommend to the designer an operator that is likely to receive the output data of the first target operator in the present scheme, that is, a downstream operator of the first target operator.
The designer drags at least one operator to the corresponding region in advance, and connection relations are established among the at least one operator.
In at least one operator for establishing the connection relation, the output flow can be one or more than one first target operator to be determined. When the first target operator is multiple, the next steps are executed for each first target operator, and then an operator recommendation list is generated for each first target operator. However, for convenience of explanation, the number of the first target operators is set as one in the following description.
S120: and determining the recommendation score of each template operator in the operator set, wherein the recommendation score of the template operator is determined according to the first target operator and the template operator.
Specifically, any one of the operators that can be selected by the designer and dragged into the sub-region labeled 103 is in the set of operators. In this embodiment, the set of operators includes the operators shown in the sub-region denoted by reference numeral 102. Wherein for convenience of illustration, the operators in the operator set are defined as template operators.
Since the operators downstream of the first target operator need to be recommended, which operators are recommended to the designer are relevant to the first target operator. Therefore, for any template operator, the probability that the template operator receives the first target operator output data in the scheme can be accurately represented according to the first target operator and the recommendation score generated by the template operator, wherein the higher the recommendation score of the template operator is, the higher the probability that the template operator receives the first target operator output data in the scheme is.
S130: and screening the template operators in the operator set according to the recommendation scores.
Specifically, the recommendation score of the template operator can indicate the probability that the template operator receives the output data of the first target operator in the scheme, so that the template operator which has a higher probability of receiving the output data of the first target operator in the scheme can be screened out according to the recommendation score.
The screening process in step S130 may be: and screening out the template operators with the recommendation scores exceeding the score threshold value, or screening out a plurality of positions with the highest recommendation scores. The present application does not specifically limit step S130.
S140: and generating an operator recommendation list according to the screened template operators, wherein the operator recommendation list indicates downstream operators which can be connected with the first target operator.
Specifically, after generating the operator recommendation list, the designer may preferentially select a suitable template operator from the operator recommendation list, and drag the selected template quantum to the sub-region denoted by reference numeral 103.
It will be appreciated that when the designer cannot select a suitable template operator from the operator recommendation list, it can select a suitable template operator from the operator set, that is, in fig. 1, the designer can still select a suitable template operator from the sub-region denoted by reference numeral 102.
It can be understood that the operator recommendation list is generated according to the screened template operators, so that the number of the template operators in the operator recommendation list is certainly smaller than the number of the template operators in the operator set, and therefore for a designer, the difficulty of selecting a proper template operator from the operator recommendation list is smaller than that of selecting a proper template operator from the operator set, and an error is not easy to occur during searching.
In this embodiment, in order to enable a designer to select more quickly, when the generated operator recommendation list is displayed to the designer, recommendation scores corresponding to each template operator, operator information corresponding to each template operator, and the like may also be displayed in the operator recommendation list, where the operator information of the template operator includes, but is not limited to, an application scenario supportable by the template operator, a type of template operator input data, a type of template operator output data, a developer of the template operator, and the like.
Meanwhile, the template operators displayed in the operator recommendation list can be arranged according to the recommendation scores from high to low.
And if the template operator added into the operator recommendation list finally has a plurality of different versions, adding the version with the optimal test result into the operator recommendation list.
Referring to fig. 3, in the present embodiment, step S120 specifically includes:
s121: screening out a target historical scheme from a recorded historical scheme library aiming at each template operator; the target history scheme comprises a first target operator, and in the target history scheme, a downstream operator directly connected with the first target operator is a template operator; and counting the number of the target historical schemes, and determining the association degree of the template operator and the first target operator based on the number of the target historical schemes.
Specifically, the history scheme library stores a plurality of history schemes that have been previously run, each of which is capable of fulfilling a service requirement.
The target history scheme corresponding to each template operator meets the following conditions: the target history scheme includes a template operator and a first target operator, and in the target history scheme, a downstream operator directly connected to the first target operator is the template operator, that is, in the target history scheme, the template operator receives data output by the first target operator.
The method aims to recommend the downstream operator of the first target operator, so that the relevance determined according to the number of the target historical schemes corresponding to the template operator can accurately represent the probability of the template operator receiving the output data of the first target operator in the scheme.
In an application scenario, for any template operator, determining the association degree between a first target operator and the template operator according to the following steps:
s1211: and determining the association degree of the template operator and the first target operator as a preset value in response to the number of the target historical schemes not exceeding a first number threshold.
S1212: in response to the number of the target historical schemes exceeding a first number threshold, determining the association degree between the template operator and a first target operator according to the ratio of the number of the target historical schemes to the total number of targets, wherein the total number of the targets is the sum of the number of all the target historical schemes exceeding the first number threshold, and the association degree is in direct proportion to the ratio.
Specifically, if the number of the target historical schemes corresponding to the template operator does not exceed the first number threshold, it is determined that in the scheme of this time, the probability that the template operator receives the output data of the first target operator is very low, and in order to reduce the amount of calculation, the association between the first target operator and the template operator is directly determined as a preset value. The preset value can be set according to actual requirements, but it is required to ensure that the association degree between the first target operator and any template operator corresponding to the target history scheme with the number exceeding the first number threshold is smaller than the preset value.
And if the number of the target historical schemes corresponding to the template operator exceeds a first number threshold, judging that the probability that the template operator receives the output data of the first target operator is higher in the scheme, and determining the association degree between the first target operator and the template operator according to the ratio of the number of the target historical schemes to the total number of the targets.
Wherein the process of determining the total number of targets is: and adding the number of all the target historical schemes of which the number exceeds the first number threshold value to obtain the total number of the targets.
When the number of the target history schemes corresponding to the template operator exceeds the first number threshold, the ratio of the number of the target history schemes corresponding to the template operator to the total number of the targets can be directly determined as the association degree between the first target operator and the template operator, or the ratio of the number of the target history schemes corresponding to the template operator to the total number of the targets can be operated to obtain the association degree between the first target operator and the template operator, wherein the association degree is higher as long as the higher the ratio is ensured.
Wherein, the first quantity threshold value can be set according to actual requirements.
S122: and determining the similarity between the first target operator and the template operator aiming at each template operator.
Specifically, there should be a certain similarity between the template operator capable of receiving the output data of the first target operator and the first target operator, for example, the type of the input data of the template operator capable of receiving the output data of the first target operator and the type of the output data of the first target operator should be the same (for example, both are target frame types), or the scenarios to which both can be applied should be similar, or the developers or maintainers of both should be similar.
Therefore, the higher the similarity between the first target operator and the template operator is, which indicates that in the scheme, the higher the possibility that the template operator receives the output data of the first target operator is. That is, the similarity between the first target operator and the template operator can also represent the probability that the template operator receives the output data of the first target operator in the present scheme.
In this embodiment, for any template operator, the similarity between the first target operator and the template operator is determined according to the following steps:
s1221: and determining a first feature vector of the first target operator according to the operator attribute of the first target operator.
S1222: and aiming at each template operator, determining a second feature vector of the template operator according to the operator attribute of the template operator.
S1223: and for each template operator, determining the similarity of the first target operator and the template operator based on the first characteristic vector of the first target operator and the second characteristic vector of the template operator.
Specifically, the first target operator and each template operator are an algorithm component, the algorithm component includes information such as a model with a specific algorithm function, a modeling code of the model, a data set used by a training model, a deployment code of the model, and the like, and in order to determine the similarity between the first target operator and the template operator, in the embodiment, according to an operator attribute of the first target operator, a first feature vector is generated for the first target operator, according to an operator attribute of the template operator, a second feature vector is generated for the template operator, and then the similarity between the first feature vector and the template operator is obtained by comparing the similarities between the first feature vector and the second feature vector.
The similarity between the first target operator and the template operator can be determined according to the euclidean distance or the cosine value between the first feature vector and the second feature vector, and the process is not particularly limited in the present application. It will be appreciated that, since the first feature vector and the second feature vector need to be calculated, the first feature vector and the second feature vector are the same length, i.e. comprise the same number of values.
In this embodiment, considering that the template operator can receive the first target operator output data under the precondition that the type of the template operator input data is the same as the type of the first target operator output data, the operator attribute of the first target operator in step S1221 is set to at least include the type of the first target operator output data, and the operator attribute of the template operator in step S1222 at least includes the type of the template operator input data. That is, a first feature vector of the first target operator is determined at least based on the type of the first target operator output data, and a second feature vector of the template operator is determined at least based on the type of the template operator input data.
The type of the data refers to the meaning of the data representation, for example, if the data representation is a target frame on the image, the type of the data is a target frame type, if the data representation is the probability that a certain target object in the image is a person, the type of the data is a confidence level type, and if the data representation is the feature information of the certain target object on the image, the type of the data is a target feature type.
Specifically, a corresponding vector may be set in advance for each type of data, for example, a target frame type is represented by a vector 1, a feature type is represented by a vector 2, an image type is represented by a vector 3, and the like. The type of the output data of the first target operator may be one or multiple, if the type is one, the vector corresponding to the type is directly used as the first feature vector of the first target operator, but if the type is multiple, the vector corresponding to each type is obtained first, then the vectors corresponding to all types are fused into one vector, and the first feature vector of the first target operator is obtained, wherein the length of the vector obtained by fusion and the length of each vector before fusion need to be kept consistent. The process of determining the second feature vector of the template operator is similar to the process of determining the first feature vector of the first target operator, and is not repeated herein.
Or a fixed-length basis vector can be preset, wherein when a first feature vector of a first target operator is generated, the value of a first preset position in the basis vector is modified to correspond to the type of the output data of the first target operator according to the type of the output data of the first target operator, and when a second feature vector of the template operator is generated, the value of the first preset position in the basis vector is modified to correspond to the type of the input data of the template operator according to the type of the input data of the template operator, wherein the first preset position can be any position in the basis vector.
It should be noted that, in other embodiments, the operator attribute of the first target operator in step S1221 may further include parameters of an application scenario, a maintainer, a supported device platform, and the like of the first target operator, and correspondingly, the operator attribute of the template operator in step S1222 may also include parameters of an application scenario, a maintainer, a supported device platform, and the like. Wherein, if the operator attribute of the first target operator further includes other types of parameters, the operator attribute of the template operator also needs to further include the type of parameters. For example, if the operator attribute of the first target operator in step S1221 further includes the application scenario of the first target operator, the operator attribute of the template operator in step S1222 also needs to further include the application scenario of the template operator.
S123: and determining the recommendation score of the template operator based on the association degree and the similarity of the first target operator and the template operator for each template operator.
For each template operator, the corresponding relevance and similarity can represent the probability that the template operator receives the output data of the first target operator in the scheme, so that the recommendation score determined according to the relevance and similarity corresponding to the template operator can also represent the probability that the template operator receives the output data of the first target operator in the scheme. Meanwhile, the template operator is evaluated according to the corresponding association degree and similarity of the template operator, so that the generated recommendation score can show the conditions of the template operator in multiple aspects, and the subsequent accurate recommendation is ensured.
In this embodiment, the relevance and similarity corresponding to the template operator may be subjected to weighted summation to obtain the recommendation score of the template operator. At the moment, the weights corresponding to the association degree and the similarity can be set according to actual requirements, and the flexibility is high.
In other embodiments, other operations such as averaging may be performed on the relevance and similarity corresponding to the template operator to obtain the recommendation score of the template operator.
Referring to fig. 4, in the present embodiment, step S140 specifically includes:
s141: and determining a second target operator from at least one operator for establishing the connection relation, wherein the first target operator receives output data of the second target operator.
Specifically, the second target operator is an operator that outputs data to the first target operator, among at least one operator that establishes a connection relationship, that is, the second target operator is an upstream operator of the first target operator.
S142: and re-screening the screened template operator according to the second target operator.
Specifically, the second target operator is an upstream operator of the first target operator, and the above environment of the first target operator is displayed, so that the screened template operator is screened again according to the second target operator, and subsequent accurate recommendation can be further ensured.
S143: and generating an operator recommendation list according to the template operators screened out again.
The template operator screened again can be directly added into the operator recommendation list, or the template operator screened again can be further screened, for example, the template operator screened again is further screened according to an upstream operator of a second target operator.
Referring to fig. 5, in the present embodiment, step S142 specifically includes:
s1421: generating a first target combination corresponding to the template operator for each screened template operator; the first target combination includes: a first target operator, a second target operator and a template operator; in the first target combination, the upstream operator of the first target operator is the second target operator, and the downstream operator of the first target operator is the template operator.
Specifically, for each template operator screened out, a first target combination is generated. For any screened template operator, in a corresponding first target combination, an upstream operator of the first target operator is a second target operator, and a downstream operator of the first target operator is a template operator, that is, in the first target combination, the first target operator receives output data of the second target operator, and the template operator receives output data of the first target operator.
It can be understood that, if the designer uses a certain template operator screened out as the operator to be dragged next, the scheme also includes the first target combination corresponding to the template operator. Therefore, in order to realize more accurate recommendation, the probability that the first target combination corresponding to each template operator appears in the scheme is analyzed, the first target combination with high probability is screened out, and finally the template operator in the screened first target combination is used as the template operator screened out again.
S1422: and screening the first target combinations corresponding to all the template operators according to the historical scheme library.
S1423: and taking the template operator corresponding to the screened first target combination as the screened template operator again.
In this embodiment, step S1422 specifically includes: counting the occurrence times of each first target combination in the historical scheme library; and screening out a first target combination with the occurrence frequency meeting a preset requirement.
Specifically, the higher the occurrence frequency of the first target combination in the historical solution library, the higher the probability that the first target combination appears in the solution, and the higher the probability that the template operator corresponding to the first target combination receives the output data of the first target operator.
The first target combinations with the occurrence frequency exceeding the threshold value can be screened out, or the first target combinations can be sorted according to the occurrence frequency from high to low, and then the first target combinations with the preset number arranged in the front are screened out.
In other application scenarios, step S1422 may also filter the first target combinations corresponding to all template operators according to other rules, for example, according to a historical solution library, count the latest dates used by the first target combinations, sort the first target combinations according to a sequence from near to far from the current time, and then filter out a plurality of first target combinations ranked in front.
In other embodiments, the process of S142 performing re-screening on the screened template operator may also be: and for the screened template operators, counting the times of the second target operator and the template operator appearing in the same historical scheme in the historical scheme library respectively, and after the times corresponding to each screened template operator are obtained, screening the screened template operators again according to the times.
Referring to fig. 6, fig. 6 is a schematic structural diagram of an embodiment of an electronic device according to the present application. The electronic device 200 includes a processor 210, a memory 220, and a communication circuit 230, wherein the processor 210 is coupled to the memory 220 and the communication circuit 230, respectively, the memory 220 stores program data, and the processor 210 implements the steps of the method according to any of the above embodiments by executing the program data in the memory 220, wherein detailed steps can refer to the above embodiments and are not described herein again.
The electronic device 200 may be any device with algorithm processing capability, such as a computer and a mobile phone, and an artificial intelligence platform may be integrated thereon.
Referring to fig. 7, fig. 7 is a schematic structural diagram of another embodiment of the electronic device of the present application. The electronic device 300 includes an acquisition module 310, a determination module 320, a filtering module 330, and a generation module 340.
The obtaining module 310 is configured to obtain, in at least one operator for establishing a connection relationship, a first target operator whose output flow is to be determined, where the connection relationship between the at least one operator is input externally.
The determining module 320 is connected to the obtaining module 310, and is configured to determine a recommendation score of each template operator in the operator set, where the recommendation score of the template operator is determined according to the first target operator and the template operator.
The screening module 330 is connected to the determining module 320, and is configured to screen the template operators in the operator set according to the recommendation score.
The generating module 340 is connected to the screening module 330, and configured to generate an operator recommendation list indicating downstream operators connectable to the first target operator according to the screened template operator.
The electronic device 300 may be any device with algorithm processing capability, such as a computer or a mobile phone, and may have an artificial intelligence platform integrated thereon.
Meanwhile, the electronic device 300 performs the steps in any of the above embodiments when in operation, and the detailed steps can be referred to the above related contents, which are not described herein again.
Referring to fig. 8, fig. 8 is a schematic structural diagram of an embodiment of a computer-readable storage medium of the present application. The computer-readable storage medium 400 stores a computer program 410, the computer program 410 being executable by a processor to implement the steps of any of the methods described above.
The computer-readable storage medium 400 may be a device that can store the computer program 410, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk, or may be a server that stores the computer program 410, and the server may send the stored computer program 410 to another device for operation, or may self-operate the stored computer program 410.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings, or which are directly or indirectly applied to other related technical fields, are intended to be included within the scope of the present application.
Claims (10)
1. A method for recommending downstream operators, the method comprising:
acquiring a first target operator with an output flow to be determined in at least one operator for establishing a connection relation, wherein the connection relation among the at least one operator is input externally;
determining a recommendation score for each template operator in a set of operators, wherein the recommendation score for the template operator is determined from the first target operator and the template operator;
screening the template operators in the operator set according to the recommendation scores;
and generating an operator recommendation list for indicating downstream operators which can be connected with the first target operator according to the screened template operators.
2. The method of claim 1, wherein the step of determining a recommendation score for each template operator in the set of operators comprises:
screening out a target historical scheme from a recorded historical scheme library aiming at each template operator; the target history scheme comprises the first target operator, and in the target history scheme, a downstream operator directly connected with the first target operator is the template operator; counting the number of the target historical schemes, and determining the association degree of the template operator and the first target operator based on the number of the target historical schemes;
for each template operator, determining the similarity between the first target operator and the template operator;
for each template operator, determining the recommendation score for the template operator based on the association and the similarity of the first target operator to the template operator.
3. The method of claim 2, wherein the step of determining the degree of association of the template operator with the first target operator based on the number of target history schemes comprises:
determining the degree of association of the template operator and the first target operator as a preset value in response to the number of the target history schemes not exceeding a first number threshold;
in response to the number of the target historical solutions exceeding the first number threshold, determining the association degree between the template operator and the first target operator according to a ratio of the number of the target historical solutions to a target total number, wherein the target total number is the sum of the numbers of all the target historical solutions exceeding the first number threshold, and the association degree is in direct proportion to the ratio.
4. The method of claim 2, wherein said step of determining, for each of said template operators, a similarity between said first target operator and said template operator comprises:
determining a first feature vector of the first target operator according to the operator attribute of the first target operator;
for each template operator, determining a second feature vector of the template operator according to the operator attribute of the template operator;
for each template operator, determining the similarity of the first target operator and the template operator based on a first feature vector of the first target operator and a second feature vector of the template operator.
5. The method of claim 4, wherein the operator attribute of the first target operator comprises at least a type of the first target operator output data; the operator attribute of the template operator comprises at least a type of the template operator input data.
6. The method according to claim 1, wherein the step of generating an operator recommendation list indicating downstream operators connectable by the first target operator according to the screened template operator comprises:
determining a second target operator from the at least one operator for establishing the connection relation, wherein the first target operator receives output data of the second target operator;
re-screening the screened template operator according to the second target operator;
and generating the operator recommendation list according to the template operator screened again.
7. The method according to claim 6, wherein the step of re-screening the screened template operator according to the second target operator comprises:
aiming at each screened template operator, generating a first target combination corresponding to the template operator; the first target combination includes: the first target operator, the second target operator, and the template operator; in the first target combination, the upstream operator of the first target operator is the second target operator, and the downstream operator of the first target operator is the template operator; screening first target combinations corresponding to all the template operators according to a historical scheme library;
and taking the template operator contained in the screened first target combination as the template operator screened again.
8. The method according to claim 7, wherein the step of filtering the first target combinations corresponding to all the template operators according to the historical solution library comprises:
counting the occurrence times of each first target combination in the historical scheme library;
and screening the first target combination with the occurrence frequency meeting the preset requirement.
9. An electronic device, comprising a processor, a memory and a communication circuit, wherein the processor is coupled to the memory and the communication circuit, respectively, and the memory stores program data, and the processor executes the program data in the memory to implement the steps of the method according to any one of claims 1-8.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which is executable by a processor to implement the steps in the method according to any one of claims 1-8.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211091774.0A CN115167846B (en) | 2022-09-07 | 2022-09-07 | Recommendation method of downstream operator, electronic device and computer-readable storage medium |
| PCT/CN2023/082782 WO2024051146A1 (en) | 2022-09-07 | 2023-03-21 | Methods, systems, and computer-readable media for recommending downstream operator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211091774.0A CN115167846B (en) | 2022-09-07 | 2022-09-07 | Recommendation method of downstream operator, electronic device and computer-readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115167846A true CN115167846A (en) | 2022-10-11 |
| CN115167846B CN115167846B (en) | 2022-12-20 |
Family
ID=83482071
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211091774.0A Active CN115167846B (en) | 2022-09-07 | 2022-09-07 | Recommendation method of downstream operator, electronic device and computer-readable storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115167846B (en) |
| WO (1) | WO2024051146A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024051146A1 (en) * | 2022-09-07 | 2024-03-14 | Zhejiang Dahua Technology Co., Ltd. | Methods, systems, and computer-readable media for recommending downstream operator |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180082002A1 (en) * | 2016-09-19 | 2018-03-22 | Exxonmobil Research And Engineering Company | Systems and methods for online model validation |
| US10025827B1 (en) * | 2017-01-17 | 2018-07-17 | International Business Machines Corporation | Operator fusion management in a stream computing environment |
| CN112330080A (en) * | 2021-01-07 | 2021-02-05 | 平安科技（深圳）有限公司 | Factor screening method, device, equipment and medium based on connectivity graph |
| CN112364290A (en) * | 2020-11-18 | 2021-02-12 | 中睿信数字技术有限公司 | Method and system for constructing visual calculation model based on stream-oriented calculation |
| CN113591934A (en) * | 2021-07-07 | 2021-11-02 | 浙江大华技术股份有限公司 | Method, device and equipment for arranging business analysis model and storage medium |
| CN114117206A (en) * | 2021-11-09 | 2022-03-01 | 北京达佳互联信息技术有限公司 | Recommendation model processing method and device, electronic equipment and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9582250B1 (en) * | 2015-08-28 | 2017-02-28 | International Business Machines Corporation | Fusion recommendation for performance management in streams |
| CN113204702A (en) * | 2021-04-23 | 2021-08-03 | Oppo广东移动通信有限公司 | Recommendation method, recommendation system and storage medium |
| CN114547428A (en) * | 2022-03-07 | 2022-05-27 | 北京达佳互联信息技术有限公司 | Recommendation model processing method and device, electronic equipment and storage medium |
| CN115167846B (en) * | 2022-09-07 | 2022-12-20 | 浙江大华技术股份有限公司 | Recommendation method of downstream operator, electronic device and computer-readable storage medium |
-
2022
- 2022-09-07 CN CN202211091774.0A patent/CN115167846B/en active Active
-
2023
- 2023-03-21 WO PCT/CN2023/082782 patent/WO2024051146A1/en unknown
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180082002A1 (en) * | 2016-09-19 | 2018-03-22 | Exxonmobil Research And Engineering Company | Systems and methods for online model validation |
| US10025827B1 (en) * | 2017-01-17 | 2018-07-17 | International Business Machines Corporation | Operator fusion management in a stream computing environment |
| CN112364290A (en) * | 2020-11-18 | 2021-02-12 | 中睿信数字技术有限公司 | Method and system for constructing visual calculation model based on stream-oriented calculation |
| CN112330080A (en) * | 2021-01-07 | 2021-02-05 | 平安科技（深圳）有限公司 | Factor screening method, device, equipment and medium based on connectivity graph |
| CN113591934A (en) * | 2021-07-07 | 2021-11-02 | 浙江大华技术股份有限公司 | Method, device and equipment for arranging business analysis model and storage medium |
| CN114117206A (en) * | 2021-11-09 | 2022-03-01 | 北京达佳互联信息技术有限公司 | Recommendation model processing method and device, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 成英超等: "统计推荐模型中的异构数据源资源配置", 《计算机应用研究》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024051146A1 (en) * | 2022-09-07 | 2024-03-14 | Zhejiang Dahua Technology Co., Ltd. | Methods, systems, and computer-readable media for recommending downstream operator |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2024051146A1 (en) | 2024-03-14 |
| CN115167846B (en) | 2022-12-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Temple et al. | Using machine learning to infer constraints for product lines | |
| CN106503656A (en) | A kind of image classification method, device and computing device | |
| CN110457175B (en) | Service data processing method and device, electronic equipment and medium | |
| CN115167846B (en) | Recommendation method of downstream operator, electronic device and computer-readable storage medium | |
| EP3872818A2 (en) | Method for providing diagnostic system using semi-supervised learning, and diagnostic system using same | |
| CN115048370B (en) | Artificial intelligence processing method for big data cleaning and big data cleaning system | |
| CN108875317A (en) | Software clone detection method and device, detection device and storage medium | |
| CN109285024B (en) | Online feature determination method and device, electronic equipment and storage medium | |
| CN108717519B (en) | Text classification method and device | |
| CN112633501A (en) | Development method and system of machine learning model framework based on containerization technology | |
| CN112115996B (en) | Image data processing method, device, equipment and storage medium | |
| CN111833115B (en) | Operation identification method and device, storage medium and server | |
| CN110135428A (en) | Image segmentation processing method and device | |
| CN111950579A (en) | Training method and training device for classification model | |
| CN116629330B (en) | Operator detection method and device and computer equipment | |
| CN110880117A (en) | False service identification method, device, equipment and storage medium | |
| CN110457896A (en) | The detection method and detection device of online access | |
| CN115423780A (en) | Image quality-based key frame extraction method and device, electronic equipment and medium | |
| CN112699435A (en) | Building safety detection method and building safety detection system | |
| CN114978765A (en) | Big data processing method serving information attack defense and AI attack defense system | |
| CN110177006B (en) | Node testing method and device based on interface prediction model | |
| CN114743090A (en) | Open type building block splicing prompting method and device, electronic equipment and storage medium | |
| CN112529738A (en) | Overall detection method and system for building engineering | |
| CN113282781A (en) | Image retrieval method and device | |
| CN110750727A (en) | Data processing method, device, system and computer readable storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |