A multi-dimensional coordination-based intelligent screening design method and device

Abstract

The application discloses a kind of intelligent screening design method and device based on multidimensional cooperation, its method includes: determining target supplier name in supplier database, and at least two kinds of corresponding product information are inquired out;At least two target keywords are extracted from product information, and based on the total number of target keyword and preset keyword level, the weight index of each target keyword is calculated;Product information is input to first convolutional neural network, and the classification feature of each product information is obtained;According to the weight index of each target keyword and classification feature, the reference index of target supplier is calculated, and target supplier is screened.Processing corresponding product information of target supplier name efficiently by combining keyword extraction algorithm and convolutional neural network and other intelligent analysis means, to realize the preliminary screening of the target supplier, not only improve the overall screening efficiency, but also guarantee the accuracy of result.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of online bidding, and particularly relates to an intelligent screening design method and device based on multi-dimensional cooperation. BACKGROUND

[0002] Bidding is a professional term of bidding and tendering, generally refers to the behavior that the bidder submits a bid to the tenderer within the specified period according to the conditions specified in the tender announcement or the invitation to bid. Most government agencies and public utility units can purchase equipment, materials and daily necessities through bidding, and often use bidding when exploring resources, developing mineral resources or building projects.

[0003] The bidding mode gradually changes from offline to online. After receiving a large amount of bidding information from suppliers or bidders, the large amount of bidding information can only be screened by manual work, which not only greatly increases the time cost of bidding, but also easily affects the overall review efficiency of bidding. SUMMARY

[0004] To solve the technical problems that manual screening of a large amount of bidding information greatly increases the time cost of bidding and easily affects the overall review efficiency of bidding, the application provides an intelligent screening design method and device based on multi-dimensional cooperation, and the technical scheme is as follows:

[0005] In the first aspect, the application embodiment provides an intelligent screening design method based on multi-dimensional cooperation, which comprises:

[0006] The target supplier name is determined in the supplier database, and at least two kinds of product information corresponding to the target supplier name are queried out. The supplier database includes at least two supplier names, and the number of supplier names in the supplier database is updated according to a preset time interval;

[0007] At least two target keywords are extracted from the at least two kinds of product information, and the weight index of each target keyword is calculated based on the preset keyword level and the total number of the at least two target keywords;

[0008] The at least two kinds of product information are input into the trained first convolutional neural network, and the classification features of each kind of product information are obtained according to the prediction result of the first convolutional neural network. The first convolutional neural network is trained by a plurality of sample product information with known classification features and a second convolutional neural network;

[0009] The reference index of the target supplier is calculated according to the weight index of each target keyword and the classification features of each kind of product information, and the target supplier is screened based on the reference index of the target supplier.

[0010] In an optional implementation of the first aspect, the weight index of each target keyword is calculated based on the preset keyword level and the total number of the at least two target keywords, including:

[0011] The proportion parameter of each target keyword is obtained according to the total number of the at least two target keywords and the number of each target keyword.

[0012] The level index corresponding to each target keyword is determined based on the preset keyword level, and the weight index of each target keyword is calculated according to the proportion parameter of each target keyword and the level index corresponding to each target keyword.

[0013] In another optional implementation of the first aspect, the classification feature of each product information is obtained according to the prediction result of the first convolutional neural network, including:

[0014] The prediction value corresponding to each product information is divided from the prediction result of the first convolutional neural network.

[0015] When the prediction value corresponding to the product information is in a preset first interval, the classification feature of the product information is determined as a first category.

[0016] When the prediction value corresponding to the product information is in a preset second interval, the classification feature of the product information is determined as a second category.

[0017] When the prediction value corresponding to the product information is in a preset third interval, the associated information corresponding to the product information is queried.

[0018] The product information and the associated information corresponding to the product information are superimposed, and the product information after the superposition is normalized.

[0019] The product information after the normalization is input to the first convolutional neural network again, and the classification feature of the product information is obtained according to the prediction result of the first convolutional neural network.

[0020] In another optional implementation of the first aspect, the reference index of the target supplier is calculated according to the weight index of each target keyword and the classification feature of each product information, including:

[0021] All product information with the first category of classification features is summarized to obtain a first set of target keywords.

[0022] The first reference index of the target supplier is calculated according to the weight index of each target keyword and the first set of target keywords.

[0023] The product information of the second category is classified and the target keyword is obtained.

[0024] The second reference index of the target supplier is calculated according to the weight index of each target keyword and the second set of target keywords.

[0025] The reference index of the target supplier is calculated based on the preset first parameter corresponding to the first category, the first reference index of the target supplier, the preset second parameter corresponding to the second category, and the second reference index of the target supplier.

[0026] In another optional implementation of the first aspect, after the reference index of the target supplier is calculated, before the target supplier is screened based on the reference index of the target supplier, the method further includes:

[0027] The historical shipment data corresponding to the at least two product information is obtained.

[0028] The theoretical shipment data of the target supplier is predicted according to the historical shipment data, and the third reference index of the target supplier is obtained based on the theoretical shipment data of the target supplier and the preset shipment data.

[0029] The target supplier is screened based on the reference index of the target supplier, including:

[0030] The target reference index of the target supplier is calculated according to the reference index of the target supplier and the third reference index of the target supplier.

[0031] The target supplier is screened based on the target reference index of the target supplier.

[0032] In another optional implementation of the first aspect, the target supplier is screened based on the target reference index of the target supplier, including:

[0033] When the target reference index of the target supplier is lower than the preset index threshold, the name of the target supplier is removed in the supplier database.

[0034] When the target reference index of the target supplier is higher than the preset index threshold, the name of the target supplier is marked in the supplier database.

[0035] In another optional implementation of the first aspect, the first convolutional neural network includes one hourglass structure, and the second convolutional neural network includes four hourglass structures; the loss function of the first convolutional neural network includes the loss parameter obtained after the second convolutional neural network is trained, and the second convolutional neural network is trained by the sample product information of the known classification features.

[0036] Secondly, embodiments of this application provide an intelligent screening design device based on multi-dimensional collaboration, comprising:

[0037] The data query module is used to determine the target supplier name in the supplier database and query at least two types of product information corresponding to the target supplier name; wherein, the supplier database includes at least two supplier names, and the number of supplier names in the supplier database is updated at preset time intervals;

[0038] The first calculation module is used to extract at least two target keywords from at least two types of product information, and calculate the weight index of each target keyword based on the preset keyword level and the total number of at least two target keywords.

[0039] The second calculation module is used to input at least two types of product information into a trained first convolutional neural network, and obtain the classification features of each type of product information based on the prediction results of the first convolutional neural network; wherein, the first convolutional neural network is trained from sample product information with multiple known classification features and the second convolutional neural network;

[0040] The data analysis module is used to calculate the reference indicators for target suppliers based on the weight index of each target keyword and the classification characteristics of each product information, and to screen target suppliers based on the reference indicators.

[0041] In one alternative embodiment of the second aspect, the first computation module includes:

[0042] The first calculation unit is used to calculate the proportion parameter of each target keyword based on the total number of at least two target keywords and the number of each target keyword.

[0043] The second calculation unit is used to determine the level index corresponding to each target keyword based on the preset keyword level, and to calculate the weight index of each target keyword according to the proportion parameter of each target keyword and the level index corresponding to each target keyword.

[0044] In another alternative embodiment of the second aspect, the second computation module includes:

[0045] A segmentation unit is used to segment the predicted value corresponding to each type of product information from the prediction results of the first convolutional neural network.

[0046] The first determining unit is used to determine the classification feature of the product information as the first category when the predicted value corresponding to the product information is in a preset first interval.

[0047] The second determining unit is configured to determine that the classification feature of the product information is a second category when the predicted value corresponding to the product information is in a preset second interval.

[0048] The third determining unit is configured to query the associated information corresponding to the product information when the predicted value corresponding to the product information is in a preset third interval.

[0049] The processing unit is configured to perform superposition processing on the product information and the associated information corresponding to the product information, and perform normalization processing on the product information after the superposition processing.

[0050] The prediction unit is configured to input the product information after the normalization processing into the first convolutional neural network again, and obtain the classification feature of the product information according to a prediction result of the first convolutional neural network.

[0051] In another optional implementation of the second aspect, the data analysis module comprises:

[0052] The first summarizing unit is configured to perform summarizing processing on all product information with the classification feature being the first category, to obtain a first set of target keywords.

[0053] The third calculating unit is configured to calculate a first reference index of the target supplier according to the weight index of each target keyword and the first set of target keywords.

[0054] The second summarizing unit is configured to perform summarizing processing on all product information with the classification feature being the second category, to obtain a second set of target keywords.

[0055] The fourth calculating unit is configured to calculate a second reference index of the target supplier according to the weight index of each target keyword and the second set of target keywords.

[0056] The fifth calculating unit is configured to calculate the reference index of the target supplier based on a preset first parameter corresponding to the first category, the first reference index of the target supplier, a preset second parameter corresponding to the second category, and the second reference index of the target supplier.

[0057] In another optional implementation of the second aspect, the data analysis module further comprises:

[0058] The obtaining unit is configured to obtain historical shipment data corresponding to at least two product information before performing screening processing on the target supplier based on the reference index of the target supplier after calculating the reference index of the target supplier.

[0059] The sixth calculating unit is configured to predict theoretical shipment data of the target supplier according to the historical shipment data, and obtain a third reference index of the target supplier based on the theoretical shipment data of the target supplier and preset shipment data.

[0060] The target supplier is screened based on the reference index of the target supplier, and the screening is specifically used for:

[0061] The target reference index of the target supplier is calculated according to the reference index of the target supplier and the third reference index of the target supplier.

[0062] The target supplier is screened based on the target reference index of the target supplier.

[0063] In another optional implementation of the second aspect, the target supplier is screened based on the target reference index of the target supplier, and the screening is specifically used for:

[0064] When the target reference index of the target supplier is lower than the preset index threshold, the name of the target supplier is removed in the supplier database.

[0065] When the target reference index of the target supplier is higher than the preset index threshold, the name of the target supplier is marked in the supplier database.

[0066] In another optional implementation of the second aspect, the first convolutional neural network includes one hourglass structure, and the second convolutional neural network includes four hourglass structures; the loss function of the first convolutional neural network includes a loss parameter obtained after the second convolutional neural network is trained, and the second convolutional neural network is trained by sample product information of a plurality of known classification features.

[0067] In a third aspect, the embodiments of the present application further provide an intelligent screening design device based on multi-dimensional cooperation, comprising a processor and a memory;

[0068] The processor is connected with the memory;

[0069] The memory is used for storing executable program codes;

[0070] The processor runs a program corresponding to the executable program codes by reading the executable program codes stored in the memory, so as to realize the intelligent screening design method based on multi-dimensional cooperation provided by the first aspect or any one of the implementation manners of the first aspect of the embodiments of the present application.

[0071] In a fourth aspect, the embodiments of the present application provide a computer storage medium, the computer storage medium stores a computer program, the computer program includes program instructions, and the program instructions can realize the intelligent screening design method based on multi-dimensional cooperation provided by the first aspect or any one of the implementation manners of the first aspect of the embodiments of the present application when executed by a processor.

[0072] In the embodiment of the present application, when preliminarily screening the suppliers, the target supplier name can be determined in the supplier database, and at least two product information corresponding to the target supplier name can be inquired out; at least two target keywords can be extracted from the at least two product information, and the weight index of each target keyword can be calculated based on the preset keyword level and the total number of the at least two target keywords; the at least two product information can be input into the trained first convolutional neural network, and the classification features of each product information can be obtained according to the prediction result of the first convolutional neural network; the reference index of the target supplier can be calculated according to the weight index of each target keyword and the classification features of each product information, and the target supplier can be screened based on the reference index of the target supplier. By combining the keyword extraction algorithm and the convolutional neural network and other intelligent analysis means, the product information corresponding to the target supplier name is efficiently processed to realize the preliminary screening of the target supplier, which not only improves the overall screening efficiency, but also guarantees the accuracy of the result. BRIEF DESCRIPTION OF DRAWINGS

[0073] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required to be used in the embodiments will be briefly introduced. Obviously, 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 creative labor.

[0074] Figure 1 The overall flowchart of the intelligent screening design method based on multi-dimensional cooperation provided in the embodiment of the present application;

[0075] Figure 2 The structure diagram of the convolutional neural network provided in the embodiment of the present application;

[0076] Figure 3 The diagram of the category-preset parameter type list provided in the embodiment of the present application;

[0077] Figure 4 The structure diagram of the intelligent screening design device based on multi-dimensional cooperation provided in the embodiment of the present application;

[0078] Figure 5 The structure diagram of another intelligent screening design device based on multi-dimensional cooperation provided in the embodiment of the present application. DETAILED DESCRIPTION

[0079] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application.

[0080] In the following description, the terms "first", "second", etc. are used only for the purpose of description, and should not be understood as indicating or implying relative importance. The following description provides a plurality of embodiments of the present application, and different embodiments can be replaced or combined, so the present application can also be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, C, and another embodiment includes features B, D, the present application should also be considered to include embodiments containing one or more of all other possible combinations of A, B, C, and D, even if the embodiment is not explicitly described in the following.

[0081] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes can be made in the function and arrangement of elements described without departing from the scope of the present application. Various examples can omit, substitute, or add various procedures or components as appropriate. For instance, the methods described can be performed in an order different than described, and various steps can be added, omitted, or combined. Also, features described with respect to some examples can be combined in other examples.

[0082] Please refer to Figure 1 , Figure 1 The overall flowchart of the intelligent screening design method based on multi-dimensional collaboration provided by the embodiments of the present application is shown.

[0083] As Figure 1 shown, the intelligent screening design method based on multi-dimensional collaboration can at least include the following steps:

[0084] Step 102, determining the target supplier name in the supplier database, and querying at least two product information corresponding to the target supplier name.

[0085] In the embodiments of the present application, the intelligent screening design method based on multi-dimensional collaboration can be but not limited to applied in the online bidding system, which can obtain the bidding information sent by each participating supplier, and complete the preliminary intelligent screening of the participating suppliers according to the bidding information.

[0086] Specifically, when the participating suppliers are preliminarily screened, the supplier database can be established in combination with the bidding information sent by each participating supplier, and at least two participating supplier names can be included in the supplier database. It can be understood that the supplier database can be updated at a predetermined time interval, so that the bidding information sent by all participating suppliers within the specified bidding time can be summarized in real time.

[0087] Further, after the supplier database is established, a target supplier name to be screened can be determined in the supplier database, and the target supplier name can be a supplier name of any record in the supplier database. It can be understood that the manner of determining the target supplier name to be screened can be realized manually, for example, an operator can manually input the target supplier name to be screened in a display interface of the online bidding system, or can select any supplier name in a selection box corresponding to the supplier database as the target supplier name to be screened, and the subsequent steps can be automatically executed by selecting a query control, but are not limited thereto. Possibly, the manner of determining the target supplier name to be screened can also be automatically executed by the online bidding system, for example, the online bidding system can determine the target supplier name to be screened according to the record time of each supplier name in the supplier database in the order from the front to the back.

[0088] It should be noted that each supplier name in the supplier database can be the target supplier name to be screened, so that each supplier in the supplier database can be preliminarily screened, thereby improving the overall bidding efficiency.

[0089] Further, after the target supplier name is determined, at least two kinds of product information corresponding to the target supplier name can be queried according to the target supplier name, and the at least two kinds of product information can include, but are not limited to, the name information and the introduction information corresponding to each of the at least two kinds of products. The introduction information can be understood as the description information of the product function and the description information of the product usage. The at least two kinds of product information can be obtained according to the bidding information sent by the supplier, or can be obtained by querying the Internet. It can be understood that the at least two kinds of product information can also be understood as at least two kinds of business information, and the business information can include, but is not limited to, the business capabilities provided by the target supplier and the related introduction information.

[0090] In the embodiment of the present application, the preliminary screening of each supplier can be realized by intelligently analyzing the multiple product information provided by each supplier, for example, the completely unsuitable suppliers are eliminated, and the suppliers that can continue to be screened are retained.

[0091] Step 104, at least two target keywords are extracted from the at least two kinds of product information, and a weight index of each target keyword is calculated based on a preset keyword level and a total number of the at least two target keywords.

[0092] Specifically, after the at least two product information corresponding to the target supplier name are queried, at least two target keywords can be extracted from the at least two products. The extraction of the target keywords can be, but is not limited to, using natural language processing technology. For example, the at least two product information can be input into a trained Bert model to determine the at least two target keywords according to the prediction result of the Bert model.

[0093] It can be understood that the total number of characters or the same number of characters of each of the at least two target keywords can be inconsistent. For example, two target keywords A and B can be extracted from the at least two products, the total number of characters of the target keyword A can be a, the total number of characters of the target keyword B can be b, and the same number of characters of the target keyword A and the target keyword B can be c. Here, a can be greater than b, and b can be greater than c, but this is not limited.

[0094] It should be noted that the number of each of the at least two target keywords can be one or more, for example, three target keywords A, B and C can be extracted from the first product information, and three target keywords B, C and D can be extracted from the second product information. It can be concluded that the number of the target keyword A and the target keyword D is 1, the number of the target keyword B and the target keyword C is 2, and this is not limited in the embodiments of the present application.

[0095] Further, after the at least two target keywords are extracted, the total number of all target keywords and the number of each target keyword can be counted, and the proportion parameter of each target keyword can be calculated, for example, the number of the target keyword A is a, and the total number of all target keywords is n. The proportion parameter of the target keyword A can be represented as: a / n.

[0096] Further, after extracting the at least two target keywords, a grade index corresponding to each target keyword can be determined based on a preset keyword grade. The preset keyword grade can be established by the online bidding system in advance according to the bidding demand, and includes a plurality of preset keywords generated according to the bidding demand and a grade corresponding to each preset keyword. Different types of grades can be used to represent the importance of the corresponding preset keyword, and each type of grade can correspond to a different grade index, which can also reflect the importance of the corresponding preset keyword. Here, the preset keyword grade is divided into grade S, grade M and grade T. Each grade can correspond to one or more target keywords, and the target keywords corresponding to each grade are different. Grade S can represent that the importance of the corresponding target keyword is very important, and the corresponding grade index can be represented as s. Grade M can represent that the importance of the corresponding target keyword is general, and the corresponding grade index can be represented as m. Grade T can represent that the importance of the corresponding target keyword is less important, and the corresponding grade index can be represented as t. It can be understood that s is greater than m, and m is greater than t, but this is not limited in the embodiments of the present application.

[0097] Further, after obtaining the proportion parameter of each target keyword and the grade index corresponding to each target keyword, the weight index of each target keyword can be obtained by multiplying the proportion parameter of each target keyword and the corresponding grade index.

[0098] Step 106, inputting the at least two product information to the trained first convolutional neural network, and obtaining the classification features of each product information according to the prediction result of the first convolutional neural network.

[0099] Specifically, after obtaining the at least two product information, the at least two product information can be input to the trained first convolutional neural network to obtain the prediction result of the first convolutional neural network.

[0100] It can be understood that the first convolutional neural network mentioned in the embodiments of the present application can be trained by a plurality of sample product information with known classification features and a second convolutional neural network. The first convolutional neural network includes a sandglass structure, and the second convolutional neural network includes four sandglass structures. In the training process of the first convolutional neural network, the second convolutional neural network is first trained based on the plurality of sample product information with known classification features, and then the first convolutional neural network is trained based on the plurality of sample product information with known classification features and the loss parameters in the trained second convolutional neural network, thereby obtaining the trained first convolutional neural network. For details, please refer to Figure 2The structure diagram of the convolutional neural network provided by the embodiment of the application is shown. The upper half of the diagram can represent the training process of the second convolutional neural network, and the lower half of the diagram can represent the training process of the first convolutional neural network.

[0101] It should be noted that, compared with other neural networks, the first convolutional neural network mentioned in the embodiment of the application not only has a simple overall design structure, but also reduces the occupied space in the control terminal by using basic operators and a lightweight model architecture, and can guarantee the accuracy and efficiency of the output results.

[0102] In the process of training the second convolutional neural network, a second convolutional neural network including four identical four-hourglass structures can be constructed first. The four identical four-hourglass structures are connected in series, and each hourglass structure includes 4 encoding layers, 4 distillation layers, and 4 decoding layers. In the hourglass structure, data can be directly transmitted to a deeper layer through skip connection, or can be transmitted through a downsampling layer and an upsampling layer.

[0103] It can be understood that the features output by the previous hourglass structure in the second convolutional neural network can be used as the input of the subsequent hourglass structure. The second convolutional neural network is only used to improve the performance of the first convolutional neural network, so that the first convolutional neural network can use 1 / 4 of the parameters to approach the performance of the second convolutional neural network.

[0104] Further, after the structure of the second convolutional neural network is constructed, sample product information with known classification features can be input into the second convolutional neural network, and a corresponding loss function can be calculated according to the prediction result. Among them, the four distillation layers of the last hourglass structure in the second convolutional neural network will be used to calculate the loss function with the four distillation layers of the previous three hourglass structures. After the training of the second convolutional neural network is completed, the first convolutional neural network can be trained based on the above-mentioned multiple sample product information with known classification features. In the process of calculating the loss function of the first convolutional neural network, the four distillation layers of the hourglass structure of the first convolutional neural network are first used to calculate the loss with the four distillation layers of the last hourglass structure of the second convolutional neural network and added to the loss function of the first convolutional neural network. Then, the prediction result of the first convolutional neural network is used to calculate the loss with the prediction result of the second convolutional neural network and added to the loss function of the first convolutional neural network.

[0105] After the training of the first convolutional neural network and the second convolutional neural network is completed, all parameters can be compressed from 32-bit to 8-bit using the quantization function of pytorch, which greatly reduces the size of the model without losing performance.

[0106] Further, after obtaining the prediction result of the first convolutional neural network, a prediction value corresponding to each product information can be divided from the prediction result. For example, taking at least two product information including Q and W as an example, product information Q and product information W can be sequentially input into the first convolutional neural network, and the obtained prediction result can be represented as (q, w). The prediction result can be divided according to the character “,”, but is not limited to this. The prediction value corresponding to the product information Q can be q, and the prediction value corresponding to the product information W can be w, but the present application is not limited to this.

[0107] Possibly, when it is detected that the prediction value corresponding to a certain product information is in a preset first interval, it can be determined that the classification feature of the product information is a first category. Here, the first category can represent that the product information belongs to an important category, but is not limited to this.

[0108] Possibly, when it is detected that the prediction value corresponding to a certain product information is in a preset second interval, it can be determined that the classification feature of the product information is a second category. Here, the second category can represent that the product information belongs to a non-important category, but is not limited to this. It can be understood that the minimum value of the preset first interval is greater than the maximum value of the preset second interval.

[0109] Possibly, when it is detected that the prediction value corresponding to a certain product information is in a preset third interval, it can be indicated that the classification feature of the product information cannot be effectively determined, and then the associated information corresponding to the product information can be queried. The associated information corresponding to the product information can include other product information similar to the product information, or other supplier names associated with the target supplier and product information corresponding to the other supplier names, but is not limited to this. The associated information corresponding to the product information can be obtained through Internet query, but the present application is not limited to this. It can be understood that the maximum value of the preset third interval is less than the minimum value of the preset first interval, and the minimum value of the preset third interval is greater than the maximum value of the preset second interval.

[0110] Then, after obtaining the associated information corresponding to the product information, the product information and the associated information can be superimposed and processed, and the product information after the superposition processing can be normalized to ensure that the product information after the normalization processing has reliability and effectiveness.

[0111] Then, the normalized product information can be re-input into the first convolutional neural network, and the classification feature of the normalized product information is judged again according to the prediction result of the first convolutional neural network. Possibly, when the prediction value corresponding to the product information is in a preset first interval, it can be determined that the classification feature of the product information is the first category. Possibly, when it is detected that the prediction value corresponding to a certain product information is in a preset second interval, it can be determined that the classification feature of the product information is the second category.

[0112] It should be noted that when the prediction result corresponding to the normalized product information still cannot determine the classification feature, the product information can be but not limited to be excluded to ensure the overall screening efficiency.

[0113] Step 108, according to the weight index of each target keyword and the classification feature of each product information, the reference index of the target supplier is calculated, and the target supplier is screened based on the reference index of the target supplier.

[0114] Specifically, after obtaining the classification feature of each product information, all product information corresponding to the first category of classification features can be aggregated, and the set of all product information can be used as the first set of target keywords.

[0115] Further, after obtaining the first set of target keywords, all product information corresponding to the target keywords in the first set of target keywords can be aggregated, for example, but not limited to, the total number of all target keywords and the number of each target keyword can be counted. Then, the proportion of each target keyword corresponding to the first set of target keywords can be obtained by calculating the number of each target keyword and the total number of all target keywords, and the first reference index of the target supplier can be calculated by combining the weight index of each target keyword mentioned above.

[0116] Here, the first set of target keywords includes target keywords A, B and C, the number of target keyword A is M, the number of target keyword B is N, the number of target keyword C is T, and the weight index of target keyword A is a, the weight index of target keyword B is b, and the weight index of target keyword C is c. For example, but not limited to, the first reference index of the target supplier can be obtained by the following formula:

[0117] S1=a*M / (M+N+T)+b*N / (M+N+T)+c*T / (M+N+T)

[0118] Similarly, the classified features corresponding to all product information of the second category can be summarized, and the set of all product information can be taken as the second set of target keywords.

[0119] Further, after obtaining the second set of target keywords, the target keywords corresponding to all product information in the second set of target keywords can be summarized, for example, but not limited to, the total number of all target keywords and the number of each target keyword can be counted. Then, the proportion of each target keyword corresponding to the second set of target keywords can be obtained by calculating the proportion of the number of each target keyword to the total number of all target keywords, and the second reference index of the target supplier can be calculated in combination with the weight index of each target keyword mentioned above.

[0120] Here, the second set of target keywords includes target keywords D, E and F, where the number of target keyword D is m, the number of target keyword E is n, the number of target keyword F is t, and the weight index corresponding to target keyword D is d, the weight index corresponding to target keyword E is e, and the weight index corresponding to target keyword F is f. For example, but not limited to, the second reference index of the target supplier can be obtained by the following formula:

[0121] S2=d*m / (m+n+t) +e*n / (m+n+t) +f*t / (m+n+t)

[0122] Further, after obtaining the first reference index and the second reference index of the target supplier respectively, the reference index of the target supplier can be calculated based on the first preset parameter corresponding to the first category, the first reference index of the target supplier, the second preset parameter corresponding to the second category, and the second reference index of the target supplier. Here, please refer to Figure 3 The embodiment of the application provides a schematic diagram of a category-preset parameter type list. As shown in Figure 3 The category-preset parameter type list can be pre-established by the online bidding system, and the preset parameter type corresponding to the first category can be the first parameter, and the preset parameter type corresponding to the second category can be the second parameter.

[0123] It can be understood that, taking the first preset parameter corresponding to the first category as P1, the second preset parameter corresponding to the second category as P2, the first reference index of the target supplier as S1, and the second reference index of the target supplier as S2, for example, but not limited to, the reference index of the target supplier can be obtained by the following formula:

[0124] S= P1*S1+P2*S2

[0125] As an option of the embodiment of the present application, after the reference index of the target supplier is calculated, before the target supplier is screened based on the reference index of the target supplier, the method further includes:

[0126] obtaining historical shipment data corresponding to the at least two product information;

[0127] predicting theoretical shipment data of the target supplier according to the historical shipment data, and obtaining a third reference index of the target supplier based on the theoretical shipment data of the target supplier and preset shipment data;

[0128] screening the target supplier based on the reference index of the target supplier, including:

[0129] calculating a target reference index of the target supplier according to the reference index of the target supplier and the third reference index of the target supplier;

[0130] screening the target supplier based on the target reference index of the target supplier.

[0131] In the embodiment of the present application, the historical shipment data of the target supplier and the product information can also be combined for analysis to further improve the accuracy and reliability of the preliminary screening of the target supplier.

[0132] Specifically, after the reference index of the target supplier is calculated, the historical shipment data corresponding to the at least two products of the target supplier can be obtained through the Internet, or can be obtained according to the bidding information of the target supplier, but is not limited to this. Then, the historical shipment data can be input into the trained deep learning neural network to obtain the predicted theoretical shipment data. The historical shipment data can be a plurality of historical shipment data within a preset time interval, for example, the historical shipment data can include shipment data corresponding to each of the previous three months from today. It can be understood that the deep learning neural network structure mentioned in the embodiment of the present application can be consistent with the neural network commonly used in the art (such as RNN), which can predict the theoretical shipment data corresponding to the current month based on the input shipment data corresponding to each of the previous three months, but is not limited to this.

[0133] Further, after the theoretical shipment data is predicted, the third reference index of the target supplier can be obtained in combination with the preset shipment data and the theoretical shipment data. The preset shipment data and the preset reference index can be pre-established as a linear relationship (such as but not limited to y=k*x+b, y corresponds to the reference index, and x corresponds to the shipment data), and the predicted theoretical shipment data can be substituted into the linear relationship to obtain the third reference index of the target supplier.

[0134] Further, the target reference index of the target supplier can be calculated by summing up the reference index of the target supplier and the third reference index of the target supplier, and the target supplier can be screened according to the target reference index of the target supplier.

[0135] Of course, the target reference index of the target supplier can also be obtained by combining product evaluation information of the target supplier, and the target reference index of the target supplier can be combined with historical shipment data, product information and product evaluation information of the target supplier to further improve the accuracy and reliability of the preliminary screening of the target supplier, which will not be described in detail here.

[0136] As another alternative of the embodiment of the present application, the screening of the target supplier based on the target reference index of the target supplier includes:

[0137] When the target reference index of the target supplier is lower than the preset index threshold, the name of the target supplier is removed in the supplier database;

[0138] When the target reference index of the target supplier is higher than the preset index threshold, the name of the target supplier is marked in the supplier database.

[0139] Specifically, when it is detected that the target reference index of the target supplier is lower than the preset index threshold, it can be indicated that the target supplier does not meet the preliminary requirements of the bidding, and thus the name of the target supplier can be removed in the supplier database; when it is detected that the target reference index of the target supplier is higher than the preset index threshold, it can be indicated that the target supplier meets the preliminary requirements of the bidding, and thus the name of the target supplier can be marked in the supplier database, and the target supplier after the marking can be screened subsequently.

[0140] Please refer to Figure 4 , Figure 4 A structure schematic diagram of an intelligent screening design device based on multi-dimensional collaboration provided by the embodiment of the present application is shown.

[0141] As shown in Figure 4 , the intelligent screening design device based on multi-dimensional collaboration can at least include a data query module 401, a first calculation module 402, a second calculation module 403 and a data analysis module 404, wherein:

[0142] The data query module 401 is used to determine the target supplier name in the supplier database and query at least two kinds of product information corresponding to the target supplier name; wherein the supplier database includes at least two supplier names, and the number of supplier names in the supplier database is updated according to a preset time interval;

[0143] The first calculation module 402 is configured to extract at least two target keywords from at least two product information, and calculate a weight index of each target keyword based on a preset keyword level and a total number of the at least two target keywords;

[0144] The second calculation module 403 is configured to input the at least two product information into a trained first convolutional neural network, and obtain a classification feature of each product information according to a prediction result of the first convolutional neural network; wherein the first convolutional neural network is trained by a plurality of sample product information with known classification features and the second convolutional neural network.

[0145] The data analysis module 404 is configured to calculate a reference index of a target supplier according to the weight index of each target keyword and the classification feature of each product information, and perform screening processing on the target supplier based on the reference index of the target supplier.

[0146] In some possible embodiments, the first calculation module comprises:

[0147] The first calculation unit is configured to obtain a proportion parameter of each target keyword according to a total number of the at least two target keywords and a number of each target keyword;

[0148] The second calculation unit is configured to determine a level index corresponding to each target keyword based on the preset keyword level, and calculate a weight index of each target keyword according to the proportion parameter of each target keyword and the level index corresponding to each target keyword.

[0149] In some possible embodiments, the second calculation module comprises:

[0150] The division unit is configured to divide a prediction value corresponding to each product information from the prediction result of the first convolutional neural network;

[0151] The first determination unit is configured to determine that the classification feature of the product information is a first category when the prediction value corresponding to the product information is in a preset first interval;

[0152] The second determination unit is configured to determine that the classification feature of the product information is a second category when the prediction value corresponding to the product information is in a preset second interval;

[0153] The third determination unit is configured to query associated information corresponding to the product information when the prediction value corresponding to the product information is in a preset third interval;

[0154] The processing unit is configured to perform superposition processing on the product information and the associated information corresponding to the product information, and perform normalization processing on the product information after the superposition processing;

[0155] a prediction unit, configured to input the product information after the normalization processing into the first convolutional neural network, and obtain a classification feature of the product information according to a prediction result of the first convolutional neural network.

[0156] In some possible embodiments, the data analysis module comprises:

[0157] a first summarizing unit, configured to perform summarizing processing on all product information with the classification feature being the first category, to obtain a first set of target keywords;

[0158] a third calculating unit, configured to calculate a first reference index of the target supplier according to the weight index of each target keyword and the first set of target keywords;

[0159] a second summarizing unit, configured to perform summarizing processing on all product information with the classification feature being the second category, to obtain a second set of target keywords;

[0160] a fourth calculating unit, configured to calculate a second reference index of the target supplier according to the weight index of each target keyword and the second set of target keywords;

[0161] a fifth calculating unit, configured to calculate a reference index of the target supplier based on a preset first parameter corresponding to the first category, the first reference index of the target supplier, a preset second parameter corresponding to the second category, and the second reference index of the target supplier.

[0162] In some possible embodiments, the data analysis module further comprises:

[0163] a obtaining unit, configured to, after the reference index of the target supplier is calculated, and before the target supplier is selected based on the reference index of the target supplier, obtain historical shipment data corresponding to at least two kinds of product information;

[0164] a sixth calculating unit, configured to predict theoretical shipment data of the target supplier according to the historical shipment data, and obtain a third reference index of the target supplier based on the theoretical shipment data of the target supplier and preset shipment data;

[0165] The target supplier is selected based on the reference index of the target supplier, and is specifically configured to:

[0166] a target reference index of the target supplier is calculated according to the reference index of the target supplier and the third reference index of the target supplier;

[0167] The target supplier is selected based on the target reference index of the target supplier.

[0168] In some possible embodiments, the target supplier is screened based on the target reference index of the target supplier, and the screening is specifically used for:

[0169] When the target reference index of the target supplier is lower than the preset index threshold, the name of the target supplier is eliminated in the supplier database;

[0170] When the target reference index of the target supplier is higher than the preset index threshold, the name of the target supplier is marked in the supplier database.

[0171] In some possible embodiments, the first convolutional neural network includes one hourglass structure, and the second convolutional neural network includes four hourglass structures; the loss function of the first convolutional neural network includes a loss parameter obtained after the second convolutional neural network is trained, and the second convolutional neural network is trained by sample product information of a plurality of known classification features.

[0172] Those skilled in the art can clearly understand that the technical solutions of the embodiments of the present application can be implemented by means of software and / or hardware. The "unit" and "module" in the specification refer to software and / or hardware that can independently complete or cooperate with other components to complete a specific function, and the hardware may, for example, be a field programmable gate array (FPGA), an integrated circuit (IC), and the like.

[0173] Please refer to Figure 5 , Figure 5 A structure schematic diagram of another intelligent screening design device based on multi-dimensional cooperation provided by an embodiment of the present application is shown.

[0174] As Figure 5 shown, the intelligent screening design device 500 based on multi-dimensional cooperation can include at least one processor 501, at least one network interface 504, a user interface 503, a memory 505, and at least one communication bus 502.

[0175] The communication bus 502 can be used to realize the connection and communication of the above-mentioned components.

[0176] The user interface 503 can include a key, and the optional user interface can further include a standard wired interface, a wireless interface.

[0177] The network interface 504 can include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, and the like.

[0178] The processor 501 can include one or more processing cores. The processor 501 connects various parts in the entire electronic device 500 by various interfaces and lines, executes various functions of the routing device 500 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 505, and calling data stored in the memory 505. Alternatively, the processor 501 can be implemented in at least one of the hardware forms of DSP, FPGA, PLA. The processor 501 can integrate one or a combination of CPU, GPU, and modem, etc. Among them, the CPU mainly processes operating systems, user interfaces, and application programs, etc.; the GPU is responsible for rendering and drawing the content required to be displayed on the display screen; and the modem is used to process wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 501, but can be realized by a separate chip.

[0179] The memory 505 can include RAM and can also include ROM. Alternatively, the memory 505 includes a non-transitory computer readable medium. The memory 505 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 505 can include a program storage area and a data storage area, wherein the program storage area can store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 505 can also be at least one storage device located away from the aforementioned processor 501. As shown, the memory 505 as a computer storage medium can include an operating system, a network communication module, a user interface module, and an intelligent filtering design application based on multi-dimensional collaboration. Figure 5 The processor 501 can be used to call the intelligent filtering design application based on multi-dimensional collaboration stored in the memory 505, and specifically perform the following operations:

[0180] Specifically, the processor 501 can be used to call the intelligent filtering design application based on multi-dimensional collaboration stored in the memory 505, and specifically perform the following operations:

[0181] Determine the target supplier name in the supplier database, and query at least two product information corresponding to the target supplier name; wherein the supplier database includes at least two supplier names, and the number of supplier names in the supplier database is updated at a preset time interval;

[0182] Extract at least two target keywords from the at least two product information, and calculate the weight index of each target keyword based on a preset keyword level and the total number of the at least two target keywords;

[0183] inputting at least two pieces of product information into the trained first convolutional neural network, and obtaining a classification feature of each piece of product information according to a prediction result of the first convolutional neural network; wherein the first convolutional neural network is trained by a plurality of sample product information with known classification features and the second convolutional neural network;

[0184] According to the weight index of each target keyword and the classification feature of each piece of product information, a reference index of the target supplier is calculated, and the target supplier is screened based on the reference index of the target supplier.

[0185] In some possible embodiments, the weight index of each target keyword is calculated based on a preset keyword level and a total number of the at least two target keywords, including:

[0186] According to the total number of the at least two target keywords and the number of each target keyword, a proportion parameter of each target keyword is obtained.

[0187] Based on the preset keyword level, a level index corresponding to each target keyword is determined, and the weight index of each target keyword is calculated according to the proportion parameter of each target keyword and the level index corresponding to each target keyword.

[0188] In some possible embodiments, the classification feature of each piece of product information is obtained according to the prediction result of the first convolutional neural network, including:

[0189] A prediction value corresponding to each piece of product information is divided from the prediction result of the first convolutional neural network.

[0190] When the prediction value corresponding to the product information is in a preset first interval, it is determined that the classification feature of the product information is of a first category.

[0191] When the prediction value corresponding to the product information is in a preset second interval, it is determined that the classification feature of the product information is of a second category.

[0192] When the prediction value corresponding to the product information is in a preset third interval, associated information corresponding to the product information is queried.

[0193] The product information and the associated information corresponding to the product information are subjected to superposition processing, and the product information subjected to the superposition processing is subjected to normalization processing.

[0194] The product information subjected to the normalization processing is re-input into the first convolutional neural network, and the classification feature of the product information is obtained according to the prediction result of the first convolutional neural network.

[0195] In some possible embodiments, the reference index of the target supplier is calculated according to the weight index of each target keyword and the classification features of each product information, and includes:

[0196] All product information with the classification feature of the first category is summarized to obtain a first set of target keywords;

[0197] The first reference index of the target supplier is calculated according to the weight index of each target keyword and the first set of target keywords;

[0198] All product information with the classification feature of the second category is summarized to obtain a second set of target keywords;

[0199] The second reference index of the target supplier is calculated according to the weight index of each target keyword and the second set of target keywords;

[0200] The reference index of the target supplier is calculated based on the preset first parameter corresponding to the first category, the first reference index of the target supplier, the preset second parameter corresponding to the second category, and the second reference index of the target supplier.

[0201] In some possible embodiments, after the reference index of the target supplier is calculated, the target supplier is further subjected to a screening process based on the reference index of the target supplier, which includes:

[0202] The historical shipment data corresponding to at least two product information is obtained;

[0203] The theoretical shipment data of the target supplier is predicted according to the historical shipment data, and the third reference index of the target supplier is obtained based on the theoretical shipment data of the target supplier and the preset shipment data;

[0204] The target supplier is subjected to a screening process based on the reference index of the target supplier, which includes:

[0205] The target reference index of the target supplier is calculated according to the reference index of the target supplier and the third reference index of the target supplier;

[0206] The target supplier is subjected to a screening process based on the target reference index of the target supplier.

[0207] In some possible embodiments, the target supplier is subjected to a screening process based on the target reference index of the target supplier, which includes:

[0208] When the target reference index of the target supplier is lower than a preset index threshold, the name of the target supplier is removed from the supplier database;

[0209] When the target reference index of the target supplier is higher than the preset index threshold, the name of the target supplier is marked in the supplier database.

[0210] In some possible embodiments, the first convolutional neural network comprises one hourglass structure, and the second convolutional neural network comprises four hourglass structures; the loss function of the first convolutional neural network comprises loss parameters obtained after the second convolutional neural network is trained, and the second convolutional neural network is trained by sample product information of a plurality of known classification features.

[0211] The application further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the steps of the above method. The computer readable storage medium can include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0212] It should be noted that, for the foregoing method embodiments, in order to simply describe, they are all described as a series of action combinations, but those skilled in the art should know that the application is not limited to the action sequence described, because according to the application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should know that the embodiments described in the specification all belong to preferred embodiments, and the actions and modules involved are not necessarily necessary for the application.

[0213] In the above embodiments, the description of each embodiment is focused on, and the part not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0214] In several embodiments provided by the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only schematic. The division of the units is only a logical function division. There can be another division manner for actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical or other forms.

[0215] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, i.e. may be located in one place, or may also be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0216] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0217] If the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable memory. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a memory, including a plurality of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the embodiments of the present application. The aforementioned memory includes: a U disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes.

[0218] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be instructed by a program to complete the relevant hardware, and the program can be stored in a computer readable memory, which can include a flash disk, a read-only memory (Read-Only Memory, ROM), a random access memory (Random Access Memory, RAM), a magnetic disk or an optical disk, etc.

[0219] The above are only exemplary embodiments of the present disclosure, and cannot limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure are still within the scope of the present disclosure. Those skilled in the art will easily think of embodiments of the present disclosure after considering the specification and practicing the disclosure herein. The present application is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional techniques in the art that are not described in the present disclosure. The specification and examples are only considered as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A multi-dimensional synergy based intelligent screening design method, characterized in that, The application relates to a method for screening a target supplier. The target supplier name is determined in a supplier database, and at least two kinds of product information corresponding to the target supplier name are queried out; the supplier database comprises at least two supplier names, and the number of supplier names in the supplier database is updated at a preset time interval; At least two target keywords are extracted from the at least two kinds of product information, and the weight index of each target keyword is calculated based on a preset keyword level and the total number of the at least two target keywords; The at least two kinds of product information are input into a trained first convolutional neural network, and the classification features of each kind of product information are obtained according to the prediction result of the first convolutional neural network; the first convolutional neural network is trained by a plurality of sample product information with known classification features and a second convolutional neural network; The reference index of the target supplier is calculated according to the weight index of each target keyword and the classification features of each kind of product information, and the target supplier is screened based on the reference index of the target supplier; The classification features of each kind of product information are obtained according to the prediction result of the first convolutional neural network, and the classification features of each kind of product information are obtained according to the prediction result of the first convolutional neural network; The prediction value corresponding to each kind of product information is divided from the prediction result of the first convolutional neural network; When the prediction value corresponding to the product information is in a preset first interval, the classification features of the product information are determined as a first category; When the prediction value corresponding to the product information is in a preset second interval, the classification features of the product information are determined as a second category; When the prediction value corresponding to the product information is in a preset third interval, the associated information corresponding to the product information is queried out; The product information and the associated information corresponding to the product information are superimposed, and the product information after superimposition is normalized; The product information after normalization is input into the first convolutional neural network again, and the classification features of the product information are obtained according to the prediction result of the first convolutional neural network; The reference index of the target supplier is calculated according to the weight index of each target keyword and the classification features of each kind of product information, and the target supplier is screened based on the reference index of the target supplier; All the product information with the classification features of the first category is summarized to obtain a first set of target keywords; The first reference index of the target supplier is calculated according to the weight index of each target keyword and the first set of target keywords; All the product information with the classification features of the second category is summarized to obtain a second set of target keywords; The second reference index of the target supplier is calculated according to the weight index of each target keyword and the second set of target keywords; The reference index of the target supplier is calculated based on a preset first parameter corresponding to the first category, the first reference index of the target supplier, a preset second parameter corresponding to the second category and the second reference index of the target supplier.

2. The method of claim 1, wherein, The weight index of each target keyword is calculated based on the total number of the at least two target keywords and the number of each target keyword. The proportion parameter of each target keyword is calculated based on the total number of the at least two target keywords and the number of each target keyword. The weight index of each target keyword is calculated based on the proportion parameter of each target keyword and the grade index corresponding to each target keyword.

3. The method of claim 1, wherein, After the reference index of the target supplier is calculated, before the target supplier is screened based on the reference index of the target supplier, the method further includes: Obtaining historical shipment data corresponding to the at least two product information; Predicting theoretical shipment data of the target supplier based on the historical shipment data, and obtaining a third reference index of the target supplier based on the theoretical shipment data of the target supplier and preset shipment data. The target reference index of the target supplier is calculated based on the reference index of the target supplier and the third reference index of the target supplier. The target supplier is screened based on the target reference index of the target supplier. When the target reference index of the target supplier is lower than a preset index threshold, the name of the target supplier is removed in the supplier database.

4. The method of claim 3, wherein, When the target reference index of the target supplier is higher than a preset index threshold, the name of the target supplier is marked in the supplier database. The first convolutional neural network includes a sandglass structure, and the second convolutional neural network includes four sandglass structures; the loss function of the first convolutional neural network includes a loss parameter obtained after the second convolutional neural network is trained, and the second convolutional neural network is trained by a plurality of sample product information of known classification features. The method includes:

5. The method of claim 1, wherein, A data query module is configured to determine a target supplier name in a supplier database and query at least two product information corresponding to the target supplier name; the supplier database includes at least two supplier names, and the number of supplier names in the supplier database is updated at a preset time interval; 6. A multi-dimensional collaboration based intelligent screening design apparatus, characterized by, A first calculation module is configured to extract at least two target keywords from the at least two product information, and calculate a weight index of each target keyword based on a preset keyword level and a total number of the at least two target keywords; ​ ​ The second computing module is configured to input the at least two product information into the trained first convolutional neural network, and obtain a classification feature of each product information according to a prediction result of the first convolutional neural network; wherein the first convolutional neural network is trained by a plurality of sample product information with known classification features and a second convolutional neural network; The data analysis module is configured to calculate a reference index of the target supplier according to the weight index of each target keyword and the classification feature of each product information, and perform screening processing on the target supplier based on the reference index of the target supplier; The classification feature of each product information is obtained according to the prediction result of the first convolutional neural network, including: Dividing a prediction value corresponding to each product information from the prediction result of the first convolutional neural network; When the prediction value corresponding to the product information is in a preset first interval, determining that the classification feature of the product information is a first category; When the prediction value corresponding to the product information is in a preset second interval, determining that the classification feature of the product information is a second category; When the prediction value corresponding to the product information is in a preset third interval, querying associated information corresponding to the product information; Performing superposition processing on the product information and the associated information corresponding to the product information, and performing normalization processing on the product information after the superposition processing; Re-inputting the product information after the normalization processing into the first convolutional neural network, and obtaining the classification feature of the product information according to the prediction result of the first convolutional neural network; The reference index of the target supplier is calculated according to the weight index of each target keyword and the classification feature of each product information, including: Performing aggregation processing on all the product information with the classification feature of the first category to obtain a first set of target keywords; According to the weight index of each target keyword and the first set of target keywords, calculating a first reference index of the target supplier; Performing aggregation processing on all the product information with the classification feature of the second category to obtain a second set of target keywords; According to the weight index of each target keyword and the second set of target keywords, calculating a second reference index of the target supplier; Based on a preset first parameter corresponding to the first category, the first reference index of the target supplier, a preset second parameter corresponding to the second category, and the second reference index of the target supplier, calculating the reference index of the target supplier.

7. A multi-dimensional collaboration based intelligent screening design apparatus, characterized by, including a processor and a memory; The processor is connected with the memory; The memory is configured to store executable program codes; The processor runs a program corresponding to the executable program codes by reading the executable program codes stored in the memory, so as to execute the steps of the method according to any one of claims 1-5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer readable storage medium stores instructions which, when executed on a computer or processor, cause the computer or processor to perform the steps of the method of any one of claims 1-5.

Images