CN113312854A - Type selection recommendation method and device, electronic equipment and readable storage medium - Google Patents
Type selection recommendation method and device, electronic equipment and readable storage medium Download PDFInfo
- Publication number
- CN113312854A CN113312854A CN202110810868.8A CN202110810868A CN113312854A CN 113312854 A CN113312854 A CN 113312854A CN 202110810868 A CN202110810868 A CN 202110810868A CN 113312854 A CN113312854 A CN 113312854A
- Authority
- CN
- China
- Prior art keywords
- components
- target
- component
- knowledge graph
- type selection
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application provides a type selection recommendation method, a type selection recommendation device, electronic equipment and a readable storage medium. According to the scheme, relevant components with preset substitution relations with the target components can be found based on the knowledge graph and under the guidance of the type selection target, and then recommendation is carried out together.
Description
Technical Field
The application relates to the technical field of equipment information management, in particular to a type selection recommendation method and device, an electronic device and a readable storage medium.
Background
Conventionally, the type of a component is often selected based on the needs of a designer to match a target component. However, since the designer may not take into account the various aspects of the factors and may not list the requirements from various different angles, the target component that is matched based on the designer's requirement information may be only the appropriate component that is matched from one or more of the angles.
However, these matched components are only displayed for the designer to select, so that it is difficult to effectively and comprehensively recommend the selection of the components, and it is difficult to provide effective component selection support for the designer.
Disclosure of Invention
The application aims to provide a type selection recommendation method, a device, an electronic device and a readable storage medium, which can comprehensively and effectively provide support for user type selection.
The embodiment of the application can be realized as follows:
in a first aspect, the present application provides a type selection recommendation method, including:
acquiring performance requirements, and searching matched target components from a knowledge base according to the performance requirements;
according to the set type selection target and a knowledge graph obtained by optimization in advance, searching relevant components which meet a preset substitution relation between the knowledge graph and the target component;
obtaining scoring information of the target component and the related components;
and generating a recommendation list for recommendation according to the grading information of the target component and the related components.
Therefore, related components with preset substitution relations with the target components can be searched based on the knowledge graph and under the guidance of the type selection target, and then recommendation is carried out together.
In an optional embodiment, the knowledge graph comprises a mapping relationship between each two of the plurality of components;
the step of searching relevant components which meet a preset substitution relation between the target components and the knowledge graph according to the set type selection target and the knowledge graph obtained by pre-optimization comprises the following steps:
searching a related component forming a triple with the target component according to a knowledge graph obtained by optimization in advance, wherein the triple comprises two components with a mapping relation;
calculating the similarity between the target component and the associated component based on the set type selection target;
and determining the related components with the similarity exceeding a preset threshold as the related components meeting a preset substitution relation with the target components.
And determining the related components by calculating the similarity between the target components and the related components forming the triplets with the target components, and accurately determining the alternative components of the target components based on the similarity.
In an alternative embodiment, each of the components has a plurality of pieces of device information;
the step of calculating the similarity between the target component and the associated component based on the set type selection target includes:
setting corresponding weight values for various device information of the target component and the associated component based on the set type selection target;
and calculating the similarity between the target component and the associated component according to the component information carrying the weight.
In the application, the weight is set based on the type selection target, so that the mechanism of similarity calculation conforms to the guidance of the type selection target, and the similarity calculation is inclined to the side key point of the type selection target.
In an optional embodiment, the method further comprises optimizing a knowledge graph in advance, and the step comprises:
constructing an initial knowledge graph based on the device information of the components in the knowledge base, wherein the initial knowledge graph comprises a plurality of groups of triples formed by the device information of every two components and the mapping relation;
taking the triples in the initial knowledge graph as positive samples, and randomly replacing device information of any component contained in the triples in the initial knowledge graph to serve as negative samples;
and optimizing the initial knowledge graph according to the obtained positive sample, the negative sample and the constructed loss function until the optimized knowledge graph is obtained when a preset condition is met.
By optimizing the training samples in advance to obtain the optimized knowledge graph, the knowledge graph which accurately reflects the mapping relation between the components can be obtained, and the accuracy of searching the subsequent replaceable related components is improved.
In an alternative embodiment, the step of optimizing the initial knowledge-graph according to the obtained positive samples, negative samples and the constructed loss function includes:
obtaining a score function of the positive sample according to the obtained device information and mapping relation of the components contained in the triples in the positive sample;
obtaining a score function of the negative sample according to the device information and the mapping relation of the components contained in the triples in the obtained negative sample;
and constructing a loss function based on the score function of the positive sample and the score function of the negative sample, and performing minimization processing on the loss function to optimize the initial knowledge graph.
By adopting the mode of constructing the loss function, the knowledge graph can be optimized based on the loss function, so that the optimization process is convenient and has basis.
In an alternative embodiment, the target component and the associated component are from a plurality of different databases;
the step of generating a recommendation list for recommendation according to the scoring information of the target component and the related component comprises the following steps:
obtaining a database to which each target component and related components belong;
identifying target components and related components belonging to different databases by adopting different identification information;
and generating a recommendation list for recommendation according to the grading information, the identified target component and the identified related component.
By acquiring the database source of each component and adopting different identification marking modes, the source of each component can be obviously identified, and a user can conveniently know the source information of each component.
In an alternative embodiment, the method further comprises:
acquiring a selection catalog of the user after selecting the components based on the recommendation list;
obtaining modification information of the selected directory by the reviewers;
and adjusting the optimized knowledge graph based on the modification information.
The knowledge graph is adjusted based on the modification information of the reviewers, and the knowledge graph can be further optimized under the guidance of expert experience.
In a second aspect, the present application provides a type selection recommendation apparatus, the apparatus comprising:
the acquisition module is used for acquiring performance requirements and searching matched target components from a knowledge base according to the performance requirements;
the searching module is used for searching relevant components which meet a preset substitution relation between the relevant components and the target components in the knowledge map according to a set type selection target and the knowledge map obtained through optimization in advance;
the obtaining module is used for obtaining grading information of the target component and the related components;
and the recommendation module is used for generating a recommendation list according to the grading information of the target component and the related components so as to recommend the target component and the related components.
The type selection recommendation device provided by the application can search related components having preset substitution relations with the target components based on the knowledge map and under the guidance of the type selection target, and further recommend the components together, the type selection recommendation is not limited to the matched target components, but also comprises replaceable related components, and support can be comprehensively and effectively provided for user type selection.
In a third aspect, the present application provides an electronic device comprising one or more storage media and one or more processors in communication with the storage media, the one or more storage media storing processor-executable machine-executable instructions that, when executed by the electronic device, are executed by the processors to perform the method steps of any one of the preceding embodiments.
In a fourth aspect, the present application provides a computer-readable storage medium having stored thereon machine-executable instructions that, when executed, implement the steps of any one of the preceding embodiments.
The beneficial effects of the embodiment of the application include, for example:
the application provides a type selection recommendation method, a type selection recommendation device, electronic equipment and a readable storage medium. According to the scheme, relevant components with preset substitution relations with the target components can be found based on the knowledge graph and under the guidance of the type selection target, and then recommendation is carried out together.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.
FIG. 1 is a flowchart of a type selection recommendation method provided in an embodiment of the present application;
FIG. 2 is a schematic illustration of a knowledge-graph provided by an embodiment of the present application;
FIG. 3 is a flow chart of an optimization method provided by an embodiment of the present application;
FIG. 4 is a flowchart of sub-steps included in step S120 of FIG. 1;
FIG. 5 is a flowchart of sub-steps included in step S122 of FIG. 4;
FIG. 6 is a flowchart of sub-steps included in step S140 of FIG. 1;
fig. 7 is a block diagram of an electronic device according to an embodiment of the present disclosure;
fig. 8 is a functional block diagram of a type-selection recommendation device according to an embodiment of the present application.
Icon: 110-a storage medium; 120-a processor; 130-type selection recommendation means; 131-an acquisition module; 132-a lookup module; 133-an obtaining module; 134-recommendation module; 140-communication interface.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, 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 some embodiments of the present application, but not all embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.
In the description of the present application, it should be noted that the features in the embodiments of the present application may be combined with each other without conflict.
Please refer to fig. 1, which is a flowchart illustrating a type selection recommendation method according to an embodiment of the present application. It should be understood that, in other embodiments, the order of some steps in the alternative recommendation method described in this embodiment may be interchanged according to actual needs, or some steps may be omitted or deleted. The detailed steps of the model selection recommendation method are described as follows.
Step S110, acquiring performance requirements, and searching matched target components from a knowledge base according to the performance requirements.
And S120, searching relevant components which meet a preset substitution relation between the relevant components and the target components in the knowledge map according to the set type selection target and the knowledge map obtained through optimization in advance.
And step S130, obtaining the grading information of the target component and the related components.
And step S140, generating a recommendation list for recommendation according to the grading information of the target component and the related components.
In the design process of industrial equipment, designers can select potential components which may be used. The designer can determine the required performance index according to the design requirement, and inputs the corresponding performance requirement into the system so as to trigger the system to search the corresponding component. Alternatively, the required performance requirements may be input one by one in the system, or may be entered in bulk in the form of a BOM manifest.
The performance requirement includes a performance index for a target component, and may include any one or more of a size, a function, a position, a material code, a name, a specification, a usage amount, a unit, a process level, a production attribute, and the like of the desired component, for example.
In this embodiment, the information of the components stored in the database is obtained from a plurality of different databases in advance, and is stored in the knowledge base in a unified manner. Optionally, the components and their multiple items of device information in each database may be obtained and stored in a knowledge base, such as component basic information, manufacturer data, qualified supplier data, standard specification data, selection catalog data, forbidden operation risk information, domestic substitute data, pseudo-localization information, failure analysis data, verification data, detection quality data, natural delivery data, life prolonging test data, life evaluation information, installed quality data, maintenance data, quality zeroing data, and the like. The basic information of the components may include information such as the size, name, code, and production attribute of the components.
In this embodiment, the device information of each component may be stored in the knowledge base in the form of RDF or embed.
Therefore, the target component which can be directly matched with the performance requirement of the designer can be searched from the knowledge base based on the obtained performance requirement input by the designer.
In this embodiment, the device information of the components in the knowledge base is converted into a knowledge map and stored. The knowledge graph includes a plurality of nodes, each node represents a unitary device (device information of a device), the nodes are connectable, and a mapping relationship between the devices is represented, as shown in fig. 2. In addition, the knowledge graph further includes a plurality of index nodes of related performance indexes, for example, the index node identified with high capacity and high temperature resistance in fig. 2. Nodes of components surrounding the corresponding index node indicate that the components have corresponding performance characteristics. For example, the components surrounding the index node with large capacity have performance characteristics of large capacity, and the components surrounding the index node with high temperature resistance have performance characteristics of high temperature resistance, but the nodes with large capacity and high temperature resistance identified in the drawing are not represented, and only the large capacity and high temperature resistance are emphasized, and the other nodes in the drawing, such as A, B, C and node D, have several properties, including but not limited to: size, manufacturer, historical quality data. In the English slogan in the figure, HR represents Func replacable, HP represents Has property, and LR represents Loc replacable, and the specific expressions are translated into: functional substitutions, inclusion of attributes and in situ substitutions are all words of art and are understood by those of skill in the art.
Based on the obtained performance requirements, the corresponding target components can be automatically matched. And further, all device information of the target device, such as device specification, manufacturer, historical quality data of the device and the like, can be obtained. The matched target component and the device information thereof can be displayed for the designer to know.
In this embodiment, considering that target components matched based on performance requirements input by a designer are often fewer, and in an actual application process, other replaceable components may exist, and if such replaceable components can be found and recommended to the designer, type selection support can be provided to the designer more comprehensively.
In the industrial design process, the type selection of the designer is usually conducted under the guidance of a certain type selection target, wherein the type selection target can be set correspondingly according to the actual requirement, for example, the type selection target can be the most cost type, the most quality type, the most reliability type or the most stock type, and the like. And, there may be a plurality of or overlapping relationships between different type-selecting targets, that is, there may be a plurality of type-selecting targets. For example, it may be that the objective of cost optimization is met first, and inventory optimization is met second.
In this embodiment, the knowledge graph pre-constructed in the knowledge base may reflect a certain relationship between each component, and thus, the relevant component satisfying the preset substitution relationship with the target component may be searched based on the type selection target and the knowledge graph set by the designer.
After the target component and the related component are found, the grading information of the corresponding component can be obtained based on the operation behavior, browsing behavior and the like of the target component and the related component on the basis of the history user. And finally, generating a recommendation list based on the grading information of each component, wherein the recommendation list can comprise each target component and related components which are sorted based on the grading information.
In this embodiment, the manner of determining the score information of each target component and the related component based on the historical operation behavior and browsing behavior of the component by the user may be, for example, when the historical target component and the historical related component are pushed to the user as recommended components, if the user finally selects the recommended components, the score information may be set to be a higher score value, and the score value of the target component or the related component that the user does not finally select may be set to be a lower score value. In addition, if the user does not select the recommendation, but the browsing times and the browsing duration of the user are longer after the recommendation is made, the score value can be correspondingly increased. If the user is not selected and is not browsing, then its score value may be set to be lowest accordingly.
According to the model selection recommendation scheme provided by the embodiment, after the target component matched with the performance requirement is found, the relevant component having the preset substitution relation with the target component is found based on the model selection target and the knowledge graph obtained through optimization in advance, and the target component and the relevant component are recommended to the user together. According to the scheme, the type selection recommendation is not only limited to the matched target component, but also comprises the replaceable related components, so that support can be comprehensively and effectively provided for the type selection of the user.
In this embodiment, the replaceable component is searched based on the knowledge graph, and the constructed knowledge graph can reflect the relationship information between the components, so that the premise of accurately searching the replaceable component is whether a reasonable knowledge graph is constructed or not.
In this embodiment, before actually performing the search for the replaceable component, the optimization process of the knowledge graph may be performed in advance. Referring to FIG. 3, the optimization of the knowledge-graph can be achieved in the following manner.
And S101, constructing an initial knowledge graph based on the device information of the components in the knowledge base, wherein the initial knowledge graph comprises a plurality of groups of triples formed by the device information and the mapping relation of every two components.
And step S102, taking the triples in the initial knowledge graph as positive samples, and randomly replacing the device information of any component contained in the triples in the initial knowledge graph to be taken as negative samples.
And S103, optimizing the initial knowledge graph according to the obtained positive sample, the negative sample and the constructed loss function until the optimized knowledge graph is obtained when a preset condition is met.
In this embodiment, an initial knowledge graph including device information of a plurality of components may be constructed in advance based on expert experience knowledge. The initial knowledge graph includes a plurality of nodes, each node represents each component (component information of the component), and each component may be an entity. And the two nodes can be connected, so that the mapping relation between the two nodes is represented.
In this embodiment, the representation in the vector is learned for each element by mapping the elements of the entities and relationships included in the knowledge-graph into a continuous vector space based on the inference representing learning. The representation in vector space may be one or more vectors or matrices. Representation learning lets algorithms automatically capture features needed for reasoning in the process of learning vector representations. The information represented by discrete symbols in the knowledge graph can be coded in different vector space representations through training and learning, so that the inference of the knowledge graph can be automatically realized through calculation between preset vector spaces, and an explicit inference step is not needed.
In one possible implementation, representation learning based reasoning can be implemented using a TranE model. Every two nodes connected in the knowledge graph may form a group of triples, and the triples include two components with mapping relationships. The triplet may be represented as (h, r, t), where h represents the head entity, r represents the relationship, and t represents the tail entity.
Each entity and relationship in the knowledge-graph can be represented as a vector, and according to the enlightenment of the word vector, the relationship in the triple can be regarded as a translation from a head entity vector to a tail entity vector. Ideally, the triples in the knowledge-graph should satisfy the following relationship:
in this embodiment, the knowledge graph may be optimized by constructing the loss function. The adopted positive sample can be a triple in the initial knowledge graph, namely, a plurality of groups of sets of two components respectively containing mapping relations. The adopted negative sample can be obtained by randomly replacing the device information of any component contained in the triplet in the initial knowledge graph, for example, the head entity in the triplet can be randomly replaced, and the tail entity in the triplet can also be randomly replaced.
Based on the above assumed ideal relationship in the triplet, each positive sample obtained should satisfy the following relationship:
and each constructed negative sample satisfies the following relation:
on the basis of obtaining the positive samples and the negative samples, the initial knowledge graph can be optimized based on the information of the positive samples and the negative samples and the constructed loss function.
In the embodiment, the knowledge graph which accurately reflects the mapping relation between the components can be obtained by optimizing the training sample to obtain the optimized knowledge graph, so that the accuracy of searching the subsequent replaceable related components is improved.
In this embodiment, the similarity between h + r and t in the triplet may be measured by using vector similarity, and in a possible implementation manner, the similarity between the h + r and t may be calculated by using an euclidean distance calculation manner. The score function of the positive sample can be obtained according to the device information and the mapping relation of the components and parts contained in the triple in the positive sample, and meanwhile, the score function of the negative sample can be obtained according to the device information and the mapping relation contained in the triple in the negative sample.
The score function of each triplet in the positive sample and the negative sample can be calculated by adopting the following formula:
where f denotes the score function of positive or negative samples, for positive samples the score function f should be as small as possible, and for negative samples the score function f should be as large as possible.
On the basis, a loss function is constructed based on the score function of the positive samples and the score function of the negative samples, and the loss function is subjected to minimization processing to optimize the initial knowledge graph.
In this embodiment, the obtained positive samples and negative samples respectively include a plurality of samples, and a positive sample set and a negative sample set can be constructed. Based on the score functions of the positive and negative examples, a loss function can be constructed in the following manner:
wherein L represents a loss function, S represents a positive sample set in the initial knowledge-graph, S' represents a constructed negative sample set, [ x ]]+ represents max (0, x);
the interval in the loss function is represented as a hyper parameter larger than zero to be set.
And performing multiple iterations on the loss function to perform minimization processing, and after the connection relation, the connection strength and the like between the nodes in the initial knowledge graph can be adjusted after each iteration, performing the next iteration until the iteration is stopped when preset conditions are met. The preset condition may be that the set number of iterations is reached, or that the loss function reaches convergence, and the like. And when the training reaches the preset condition, the acquired knowledge graph is the optimized knowledge graph.
In the embodiment, by adopting the mode of constructing the loss function, the knowledge graph can be optimized based on the loss function, so that the optimization process is convenient and has basis.
The TranE model is a simple and efficient knowledge graph representation learning method due to effective and reasonable vector space assumption, and can complete link prediction tasks of various relations. The simple and efficient TranE model can automatically and well capture inference characteristics, does not need manual design, is very suitable for popularization on large-scale complex knowledge graphs, and is an effective knowledge graph inference means.
In other possible implementation manners, the inference based on the representation learning can also be realized by adopting a TranH model or a TranR model, and the TranH model or the TranR model can be applied to the scenes of one-to-many, many-to-one, many-to-many and the like relations.
In addition, since many complex relationships may exist between the reasoning capabilities of the knowledge graph representation learning and the vector space assumptions employed, linear mapping assumptions may be employed between entities in addition to translation assumptions. The entity can be represented as a vector, the relationship can be represented as a matrix, and the relationship can be regarded as a linear transformation of a vector space. In this case, for a triplet, ideally, the following formula can be satisfied:
wherein h and t represent a head entity and a tail entity respectively, and Mr is a matrix representation of the relation r. The expression of the above expression means that the head entity can be converted into the tail entity vector after being multiplied by the relation matrix and subjected to linear transformation in space. In this case, the goal of the training is to make hMr in the positive samples as close as possible to t, and hMr in the negative samples as far as possible away from t. Accordingly, in this case, the score function of the positive or negative sample may be expressed as follows:
in this embodiment, the present invention can be applied to calculation of a real number, and also to calculation in the case of a complex number. When applied to complex computations, then the score function for positive or negative samples can be expressed as follows:
in addition, graph neural networks, in which dependency relationships between nodes in a graph are captured as information propagates between the nodes, are also an efficient method for processing data of graph structures. The graph structure is represented in such a way that the model can reason based on the graph.
In the actual application process, any one of the models can be adaptively adopted for calculation according to actual services and data characteristics, and then the knowledge graph is optimized.
In this embodiment, the knowledge graph can be optimized in advance through the above manner, and based on the obtained optimized knowledge graph, in the actual application process, the relevant component having the preset substitution relationship with the target component can be found based on the knowledge graph.
Referring to fig. 4, in this embodiment, the related component having the preset substitution relationship with the target component can be found in the following manner:
and step S121, searching for a related component forming a triple with the target component according to a knowledge graph obtained through optimization in advance, wherein the triple comprises two components with a mapping relation.
And step S122, calculating the similarity between the target component and the associated component based on the set type selection target.
And S123, determining the related component with the similarity exceeding the preset threshold as the related component meeting the preset substitution relation with the target component.
From the above, two components in the knowledge graph may form a group of triples, and a certain mapping relationship exists between two components in the triples. Thus, based on the target component, the associated component in which the triplet is formed with the target component can be found. For example, in the triplet including the target component, the target component is a head entity and the associated component is a tail entity, and a relationship exists between the target component and the associated component, or the target component is the tail entity and the associated component is the head entity, and a relationship exists between the target component and the associated component.
The two components constructing the triplet may not necessarily satisfy the preset substitution relationship, and in this embodiment, the similarity between the target component and the associated component may be calculated under the guidance of the type-selection target.
The similarity between the target component and the associated component can be calculated by adopting an Euclidean distance calculation method, and optionally, the Euclidean distance between the target component and the associated component can be calculated by adopting the following method:
the smaller the Euclidean distance between the target component and the related component is, the higher the similarity of the target component is, and the larger the Euclidean distance is, the lower the similarity of the target component and the related component is. In this embodiment, when the similarity between the target component and the associated component exceeds the preset threshold, the associated component may be determined as a related component that satisfies the preset alternative relationship with the target component.
In this embodiment, the similarity between the target component and the associated component constituting the triplet with the target component is calculated to determine the associated component, and the alternative component of the target component can be accurately determined based on the similarity.
In this embodiment, the device information of each component may include a plurality of items, and the above expression performs calculation based on the plurality of items of device information included in the two components when calculating the euclidean distance between the two components. While the euclidean distance between the components is calculated under the guidance of the set type selection target, as a possible implementation manner, in detail, referring to fig. 5, the following manner may be adopted:
and step S1221, setting corresponding weight values for various device information contained in the target component and the associated component based on the set type selection target.
Step S1222, calculating the similarity between the target component and the associated component according to the component information carrying the weight.
In this embodiment, the type selection target is to determine relevant components and devices to guide. When the type selection target is one, for example, when the type selection target is cost-optimized, the cost data in the device information may be set to the maximum weight when the corresponding weights are set for each item of device information of the target device and the associated device. Thus, if the cost data occupies a larger weight, the similarity can be inclined to the cost data.
Further, if there are multiple typing targets, for example, the typing targets are first cost-optimized and second inventory-optimized. When setting the corresponding weight for each item of device information of the target device and the associated device, the cost data in the device information may be set as the maximum weight, and the weight of the inventory data in the device information may be set as the second largest value. Therefore, under the condition that the cost data and the inventory data occupy larger weights, the similarity of the target component and the related components can be calculated, and the requirements of the model selection target with optimal cost and optimal inventory can be met.
In this embodiment, through the above manner, on the basis of the optimized knowledge graph, the related components that satisfy the preset substitution relationship with the target component can be found out by calculating the similarity between the components. On the basis, recommendation of the target component and the related components is achieved.
In the process, the weight is set based on the type selection target, so that the mechanism of similarity calculation conforms to the guidance of the type selection target, and the similarity calculation is inclined to the side point of the type selection target. As can be seen from the above, the components in the knowledge base may be from different databases, and in the actual process of industrial design, the components from some databases are often given priority, for example, the preferred directory and the domestic substitute directory, and in addition to the two directories, the components may also be from other databases such as the pseudo-localization database and the forbidden operation risk database.
In order to make designers know the source of each component at a glance when recommending components, in this embodiment, please refer to fig. 6, the recommendation list can be generated in the following manner:
step S141, a database to which each target component and related components belong is obtained.
And step S142, identifying the target component and the related component belonging to different databases by adopting different identification information.
And S143, generating a recommendation list for recommendation according to the grading information, the identified target component and the identified related component.
In this embodiment, after the databases to which the target components and the related components belong are obtained, the target components and the related components belonging to different databases may be identified by, for example, color separation labeling or remarking.
Therefore, the database sources of the components are acquired, the different identification marking modes are adopted, the sources of the components can be obviously identified, and the recommendation list presented to a designer is prompted, so that the designer can distinguish and know the source information of the components conveniently. In this embodiment, in the finally generated recommendation list, for the relevant components, in addition to recommendation based on ranking of the score information, ranking may also be performed based on the similarity between each relevant component and the target component. For example, the recommendation list may be generated by sequentially sorting the target components in descending order of similarity.
In addition, in one possible implementation, the target components and related components in the setting database may also be arranged in front when generating the recommendation list, and the setting database may be, for example, a preferred directory and a domestic substitute directory.
In this embodiment, the recommendation list finally presented to the designer includes detailed device information, historical quality information, and the like of each recommended target device and related devices. In addition, in this embodiment, risk evaluation indexes set by relevant experts and system engineers may also be obtained, the recommended components may also be sorted in the recommendation list based on the obtained risk evaluation indexes, and the recommended components may be presented in the order from high to low of the use risk.
In this embodiment, after presenting to the designer in the form of the recommendation list, the designer can perform final component selection based on the requirement, and the selected components need to be sent to the reviewer for review. Therefore, the type selection recommendation method provided by the embodiment may further include the following steps:
and acquiring a selection catalogue of the user after selecting the components based on the recommendation list, acquiring modification information of the selection catalogue by a reviewer, and adjusting the optimized knowledge graph based on the modification information.
In this embodiment, the system may record the review and selection of the components listed in the selection list by the reviewer, such as replacing the components in the selection list, controlling the localization rate, and so on. The knowledge-graph may be adjusted based on reviewers' modifications to the selected categories, such as adjusting relationships between nodes in the knowledge-graph.
In the case of adjusting the relationship between nodes in the knowledge graph, the recommendation condition of the component based on the knowledge graph is changed accordingly. For example, component recommendations based on a changed knowledge map will tend to favor components recommended in the candidate list as modified by the reviewer.
Therefore, the knowledge graph can be adjusted by combining the expert review information on the basis of the optimization of the knowledge graph based on the loss function, and the knowledge graph can be continuously adjusted and optimized, so that the recommendation of components is more scientific and reasonable.
Referring to fig. 7, an electronic device for performing the type selection recommendation method is further provided in an embodiment of the present application. The electronic device may include a storage medium 110, a processor 120, a type selection recommender 130, and a communication interface 140. In this embodiment, the storage medium 110 and the processor 120 are both located in the electronic device and are separately disposed. However, it should be understood that the storage medium 110 may be separate from the electronic device and may be accessed by the processor 120 through a bus interface. Alternatively, the storage medium 110 may be integrated into the processor 120, for example, may be a cache and/or general purpose registers.
The type selection recommending device 130 may be understood as the electronic device or the processor 120 of the electronic device, or may be understood as a software functional module that is independent of the electronic device or the processor 120 and implements the type selection recommending method under the control of the electronic device.
As shown in fig. 8, the type-selecting recommendation apparatus 130 may include an obtaining module 131, a searching module 132, an obtaining module 133, and a recommending module 134. The functions of the functional modules of the type selection recommending device 130 are described in detail below.
The obtaining module 131 is configured to obtain a performance requirement, and search the matched target component from the knowledge base according to the performance information.
It is understood that the obtaining module 131 can be used to execute the step S110, and for the detailed implementation of the obtaining module 131, reference can be made to the content related to the step S110.
The searching module 132 is configured to search, according to the set type selection target and the pre-optimized knowledge graph, a relevant component that satisfies a preset substitution relationship with the target component in the knowledge graph.
It is understood that the search module 132 can be used to perform the step S120, and for the detailed implementation of the search module 132, reference can be made to the above description related to the step S120.
An obtaining module 133, configured to obtain scoring information of the target component and the related component.
It is understood that the obtaining module 133 may be configured to perform the step S130, and for a detailed implementation of the obtaining module 133, reference may be made to the content related to the step S130.
And the recommending module 134 is configured to generate a recommending list for recommending according to the scoring information of the target component and the related component.
It is understood that the recommending module 134 can be used to execute the step S140, and the detailed implementation of the recommending module 134 can refer to the content related to the step S140.
In a possible implementation manner, the knowledge graph includes a mapping relationship between each two of the plurality of components, and the search module 132 may be specifically configured to:
searching a related component forming a triple with the target component according to a knowledge graph obtained by optimization in advance, wherein the triple comprises two components with a mapping relation;
calculating the similarity between the target component and the associated component based on the set type selection target;
and determining the related components with the similarity exceeding a preset threshold as the related components meeting a preset substitution relation with the target components.
In a possible implementation manner, each component has multiple pieces of device information, and the search module 132 may be specifically configured to:
setting corresponding weight values for various device information of the target component and the associated component based on the set type selection target;
and calculating the similarity between the target component and the associated component according to the component information carrying the weight.
In a possible implementation manner, the type-selection recommending apparatus 130 may further include an optimizing module for optimizing the knowledge graph in advance, and the optimizing module may be configured to:
constructing an initial knowledge graph based on the device information of the components in the knowledge base, wherein the initial knowledge graph comprises a plurality of groups of triples formed by the device information of every two components and the mapping relation;
taking the triples in the initial knowledge graph as positive samples, and randomly replacing device information of any component contained in the triples in the initial knowledge graph to serve as negative samples;
and optimizing the initial knowledge graph according to the obtained positive sample, the negative sample and the constructed loss function until the optimized knowledge graph is obtained when a preset condition is met.
In a possible implementation manner, the optimization module may be specifically configured to:
obtaining a score function of the positive sample according to the obtained device information and mapping relation of the components contained in the triples in the positive sample;
obtaining a score function of the negative sample according to the device information and the mapping relation of the components contained in the triples in the obtained negative sample;
and constructing a loss function based on the score function of the positive sample and the score function of the negative sample, and performing minimization processing on the loss function to optimize the initial knowledge graph.
In a possible implementation manner, the target component and the related component are from a plurality of different databases, and the recommendation module 134 may be specifically configured to:
obtaining a database to which each target component and related components belong;
identifying target components and related components belonging to different databases by adopting different identification information;
and generating a recommendation list for recommendation according to the grading information, the identified target component and the identified related component.
In a possible implementation manner, the type selection recommending apparatus 130 may further include an adjusting module, and the adjusting module may be configured to:
acquiring a selection catalog of the user after selecting the components based on the recommendation list;
obtaining modification information of the selected directory by the reviewers;
and adjusting the optimized knowledge graph based on the modification information.
The description of the processing flow of each module in the device and the interaction flow between the modules may refer to the related description in the above method embodiments, and will not be described in detail here.
Further, an embodiment of the present application also provides a computer-readable storage medium, where machine-executable instructions are stored in the computer-readable storage medium, and when the machine-executable instructions are executed, the method for type selection recommendation provided in the foregoing embodiment is implemented.
Specifically, the computer readable storage medium can be a general storage medium, such as a removable disk, a hard disk, and the like, and when executed, the computer program on the computer readable storage medium can execute the above type selection recommendation method. With regard to the processes involved when the executable instructions in the computer-readable storage medium are executed, reference may be made to the related descriptions in the above method embodiments, which are not described in detail herein.
In summary, the embodiment of the present application provides a type selection recommendation method, a device, an electronic device, and a readable storage medium, which first find a matched target component according to an acquired performance requirement, find a relevant component satisfying a preset substitution relationship with the target component according to a set type selection target and a knowledge graph obtained through pre-optimization, then obtain scoring information of the target component and the relevant component, and finally generate a recommendation list according to the scoring information of the target component and the relevant component for recommendation. According to the scheme, relevant components with preset substitution relations with the target components can be found based on the knowledge graph and under the guidance of the type selection target, and then recommendation is carried out together.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present application should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A type selection recommendation method is characterized by comprising the following steps:
acquiring performance requirements, and searching matched target components from a knowledge base according to the performance requirements;
according to the set type selection target and a knowledge graph obtained by optimization in advance, searching relevant components which meet a preset substitution relation between the knowledge graph and the target component;
obtaining scoring information of the target component and the related components;
and generating a recommendation list for recommendation according to the grading information of the target component and the related components.
2. The type selection recommendation method according to claim 1, wherein the knowledge graph comprises a mapping relationship between each two of the plurality of components;
the step of searching relevant components which meet a preset substitution relation between the target components and the knowledge graph according to the set type selection target and the knowledge graph obtained by pre-optimization comprises the following steps:
searching a related component forming a triple with the target component according to a knowledge graph obtained by optimization in advance, wherein the triple comprises two components with a mapping relation;
calculating the similarity between the target component and the associated component based on the set type selection target;
and determining the related components with the similarity exceeding a preset threshold as the related components meeting a preset substitution relation with the target components.
3. The type selection recommendation method according to claim 2, wherein each of the components has a plurality of pieces of device information;
the step of calculating the similarity between the target component and the associated component based on the set type selection target includes:
setting corresponding weight values for various device information of the target component and the associated component based on the set type selection target;
and calculating the similarity between the target component and the associated component according to the component information carrying the weight.
4. The type-selection recommendation method according to claim 1, further comprising optimizing a knowledge graph in advance, the method comprising:
constructing an initial knowledge graph based on the device information of the components in the knowledge base, wherein the initial knowledge graph comprises a plurality of groups of triples formed by the device information of every two components and the mapping relation;
taking the triples in the initial knowledge graph as positive samples, and randomly replacing device information of any component contained in the triples in the initial knowledge graph to serve as negative samples;
and optimizing the initial knowledge graph according to the obtained positive sample, the negative sample and the constructed loss function until the optimized knowledge graph is obtained when a preset condition is met.
5. The type-selection recommendation method according to claim 4, wherein the step of optimizing the initial knowledge-graph according to the obtained positive examples, negative examples and the constructed loss function comprises:
obtaining a score function of the positive sample according to the obtained device information and mapping relation of the components contained in the triples in the positive sample;
obtaining a score function of the negative sample according to the device information and the mapping relation of the components contained in the triples in the obtained negative sample;
and constructing a loss function based on the score function of the positive sample and the score function of the negative sample, and performing minimization processing on the loss function to optimize the initial knowledge graph.
6. The type selection recommendation method according to claim 1, wherein the target component and the related component are from a plurality of different databases;
the step of generating a recommendation list for recommendation according to the scoring information of the target component and the related component comprises the following steps:
obtaining a database to which each target component and related components belong;
identifying target components and related components belonging to different databases by adopting different identification information;
and generating a recommendation list for recommendation according to the grading information, the identified target component and the identified related component.
7. The type-selection recommendation method according to claim 1, further comprising:
acquiring a selection catalog of the user after selecting the components based on the recommendation list;
obtaining modification information of the selected directory by the reviewers;
and adjusting the optimized knowledge graph based on the modification information.
8. A type selection recommendation device, characterized in that the device comprises:
the acquisition module is used for acquiring performance requirements and searching matched target components from a knowledge base according to the performance requirements;
the searching module is used for searching relevant components which meet a preset substitution relation between the relevant components and the target components in the knowledge map according to a set type selection target and the knowledge map obtained through optimization in advance;
the obtaining module is used for obtaining grading information of the target component and the related components;
and the recommendation module is used for generating a recommendation list according to the grading information of the target component and the related components so as to recommend the target component and the related components.
9. An electronic device comprising one or more storage media and one or more processors in communication with the storage media, the one or more storage media storing processor-executable machine-executable instructions that, when executed by the electronic device, are executed by the processors to perform the method steps of any of claims 1-7.
10. A computer-readable storage medium, characterized in that it stores machine-executable instructions which, when executed, implement the method steps of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110810868.8A CN113312854B (en) | 2021-07-19 | 2021-07-19 | Type selection recommendation method and device, electronic equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110810868.8A CN113312854B (en) | 2021-07-19 | 2021-07-19 | Type selection recommendation method and device, electronic equipment and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113312854A true CN113312854A (en) | 2021-08-27 |
| CN113312854B CN113312854B (en) | 2021-11-02 |
Family
ID=77381687
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110810868.8A Active CN113312854B (en) | 2021-07-19 | 2021-07-19 | Type selection recommendation method and device, electronic equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113312854B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114117244A (en) * | 2022-01-24 | 2022-03-01 | 深圳市云采网络科技有限公司 | Component relation identification method and electronic equipment |
| CN114692486A (en) * | 2022-03-10 | 2022-07-01 | 中国科学技术大学 | Product optimization design method based on user feedback knowledge graph |
| CN115080587A (en) * | 2022-05-19 | 2022-09-20 | 华南理工大学 | Electronic component replacing method, device and medium based on knowledge graph |
| CN117421941A (en) * | 2023-12-19 | 2024-01-19 | 中冶南方工程技术有限公司 | Element selection method, device, equipment and storage medium applied to metallurgical field |
| CN117473910A (en) * | 2023-11-28 | 2024-01-30 | 南方电网调峰调频发电有限公司检修试验分公司 | Method and device for selecting low-voltage electrical apparatus components, computer equipment and storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160314158A1 (en) * | 2015-04-27 | 2016-10-27 | Rovi Guides, Inc. | Systems and methods for updating a knowledge graph through user input |
| CN107729444A (en) * | 2017-09-30 | 2018-02-23 | 桂林电子科技大学 | Recommend method in a kind of personalized tourist attractions of knowledge based collection of illustrative plates |
| CN110472067A (en) * | 2019-08-19 | 2019-11-19 | 腾讯科技（深圳）有限公司 | Knowledge mapping indicates learning method, device, computer equipment and storage medium |
| CN110472155A (en) * | 2019-07-03 | 2019-11-19 | 五邑大学 | Collaborative recommendation method, device, equipment and the storage medium of knowledge based map |
| CN110502621A (en) * | 2019-07-03 | 2019-11-26 | 平安科技（深圳）有限公司 | Answering method, question and answer system, computer equipment and storage medium |
| CN111881934A (en) * | 2020-06-18 | 2020-11-03 | 江苏大学 | Method for discovering spatial relationship between attributes and categories of electronic components |
| CN112084383A (en) * | 2020-09-07 | 2020-12-15 | 中国平安财产保险股份有限公司 | Information recommendation method, device and equipment based on knowledge graph and storage medium |
| CN112612973A (en) * | 2020-12-31 | 2021-04-06 | 重庆邮电大学 | Personalized intelligent clothing matching recommendation method combining knowledge graph |
| CN112667877A (en) * | 2020-12-25 | 2021-04-16 | 陕西师范大学 | Scenic spot recommendation method and equipment based on tourist knowledge map |
| CN112765369A (en) * | 2021-01-31 | 2021-05-07 | 西安电子科技大学 | Knowledge graph information representation learning method, system, equipment and terminal |
-
2021
- 2021-07-19 CN CN202110810868.8A patent/CN113312854B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160314158A1 (en) * | 2015-04-27 | 2016-10-27 | Rovi Guides, Inc. | Systems and methods for updating a knowledge graph through user input |
| CN107729444A (en) * | 2017-09-30 | 2018-02-23 | 桂林电子科技大学 | Recommend method in a kind of personalized tourist attractions of knowledge based collection of illustrative plates |
| CN110472155A (en) * | 2019-07-03 | 2019-11-19 | 五邑大学 | Collaborative recommendation method, device, equipment and the storage medium of knowledge based map |
| CN110502621A (en) * | 2019-07-03 | 2019-11-26 | 平安科技（深圳）有限公司 | Answering method, question and answer system, computer equipment and storage medium |
| CN110472067A (en) * | 2019-08-19 | 2019-11-19 | 腾讯科技（深圳）有限公司 | Knowledge mapping indicates learning method, device, computer equipment and storage medium |
| CN111881934A (en) * | 2020-06-18 | 2020-11-03 | 江苏大学 | Method for discovering spatial relationship between attributes and categories of electronic components |
| CN112084383A (en) * | 2020-09-07 | 2020-12-15 | 中国平安财产保险股份有限公司 | Information recommendation method, device and equipment based on knowledge graph and storage medium |
| CN112667877A (en) * | 2020-12-25 | 2021-04-16 | 陕西师范大学 | Scenic spot recommendation method and equipment based on tourist knowledge map |
| CN112612973A (en) * | 2020-12-31 | 2021-04-06 | 重庆邮电大学 | Personalized intelligent clothing matching recommendation method combining knowledge graph |
| CN112765369A (en) * | 2021-01-31 | 2021-05-07 | 西安电子科技大学 | Knowledge graph information representation learning method, system, equipment and terminal |
Non-Patent Citations (2)
| Title |
|---|
| KUN ZHOU等: "Improving Conversational Recommender Systems via Knowledge Graph based Semantic Fusion", 《PROCEEDINGS OF THE 26TH ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY & DATA MINING》 * |
| 丁春光等: "基于大数据与人工智能的元器件数据智慧管理", 《计算机科学与技术》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114117244A (en) * | 2022-01-24 | 2022-03-01 | 深圳市云采网络科技有限公司 | Component relation identification method and electronic equipment |
| CN114692486A (en) * | 2022-03-10 | 2022-07-01 | 中国科学技术大学 | Product optimization design method based on user feedback knowledge graph |
| CN114692486B (en) * | 2022-03-10 | 2024-03-15 | 中国科学技术大学 | Product optimization design method based on user feedback knowledge graph |
| CN115080587A (en) * | 2022-05-19 | 2022-09-20 | 华南理工大学 | Electronic component replacing method, device and medium based on knowledge graph |
| CN115080587B (en) * | 2022-05-19 | 2024-04-16 | 华南理工大学 | Electronic component replacement method, device and medium based on knowledge graph |
| CN117473910A (en) * | 2023-11-28 | 2024-01-30 | 南方电网调峰调频发电有限公司检修试验分公司 | Method and device for selecting low-voltage electrical apparatus components, computer equipment and storage medium |
| CN117421941A (en) * | 2023-12-19 | 2024-01-19 | 中冶南方工程技术有限公司 | Element selection method, device, equipment and storage medium applied to metallurgical field |
| CN117421941B (en) * | 2023-12-19 | 2024-04-26 | 中冶南方工程技术有限公司 | Element selection method, device, equipment and storage medium applied to metallurgical field |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113312854B (en) | 2021-11-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113312854B (en) | Type selection recommendation method and device, electronic equipment and readable storage medium | |
| US10380498B1 (en) | Platform services to enable one-click execution of the end-to-end sequence of modeling steps | |
| CN108320171B (en) | Hot-sold commodity prediction method, system and device | |
| US9390142B2 (en) | Guided predictive analysis with the use of templates | |
| Song et al. | An integrated rough number-based approach to design concept evaluation under subjective environments | |
| US8280894B2 (en) | Method and system for maintaining item authority | |
| US8190556B2 (en) | Intellegent data search engine | |
| US20150019530A1 (en) | Query language for unstructed data | |
| CN109033101B (en) | Label recommendation method and device | |
| US20070118546A1 (en) | User's preference prediction from collective rating data | |
| CN106997390B (en) | Commodity transaction information searching method for equipment accessories or parts | |
| US8577849B2 (en) | Guided data repair | |
| Torkestani | An adaptive learning to rank algorithm: Learning automata approach | |
| CN114638234B (en) | Big data mining method and system applied to online business handling | |
| CN103020289B (en) | A kind of search engine user individual demand supplying method based on Web log mining | |
| CN116187524B (en) | Supply chain analysis model comparison method and device based on machine learning | |
| EP2674880A1 (en) | Creating optimal comparison criterion within associative memories | |
| CN115456707A (en) | Method and device for providing commodity recommendation information and electronic equipment | |
| US20210090105A1 (en) | Technology opportunity mapping | |
| US8577814B1 (en) | System and method for genetic creation of a rule set for duplicate detection | |
| JP4891638B2 (en) | How to classify target data into categories | |
| Guan et al. | Enhanced SVD for collaborative filtering | |
| US8671120B1 (en) | Method and system for manually maintaining item authority | |
| CN115080587B (en) | Electronic component replacement method, device and medium based on knowledge graph | |
| CN115619503A (en) | Article recommendation method and device, storage medium and computer equipment |
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 | ||
| CP03 | Change of name, title or address |
Address after: 610000 No. 270, floor 2, No. 8, Jinxiu street, Wuhou District, Chengdu, Sichuan Patentee after: Chengdu shuzhilian Technology Co.,Ltd. Address before: 610000 No.2, 4th floor, building 1, Jule Road intersection, West 1st section of 1st ring road, Wuhou District, Chengdu City, Sichuan Province Patentee before: CHENGDU SHUZHILIAN TECHNOLOGY Co.,Ltd. |
|
| CP03 | Change of name, title or address |