Computer-implemented vendor recommendation method, computer device, and storage medium

Abstract

The application discloses a computer-implemented supplier recommendation method, a computer device and a storage medium. In a specific embodiment, the method comprises: training a PageRank model using a historical bidding data set, wherein the historical bidding data set comprises subcategory information as input of the model and supplier quotes corresponding to all bidding sessions under the subcategory, and winning probability of each supplier as output of the model; inputting the subcategory information into the trained PageRank model, and outputting recommended suppliers for inviting the recommended suppliers to participate in bidding before bidding.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of supplier recommendation. More particularly, it relates to a computer-implemented supplier recommendation method, a computer device and a storage medium. BACKGROUND

[0002] Auction refers to a transaction mode in which a buyer or a seller publishes an offer to the outside world through a transaction market spot auction transaction system, with information such as the brand, specifications, and main attributes of the goods, the delivery location, the delivery time, the quantity, the bottom price, and the like, and the qualified counterparties autonomously raise or reduce the price, and the transaction is completed with the highest bid or the lowest offer within the specified time according to the principle of "price priority", and an electronic purchase and sale contract is signed through the transaction market, and the physical delivery is made according to the contract.

[0003] For example, in a procurement scenario, the buyer needs to enter the supplier information (such as the name, bank account information, etc.) in the auction item, and the selection of the supplier directly leads to the difference in the final bid in this auction. Before the auction, the selection of which suppliers to invite to participate in the auction is set by the buyer manually, and this way cannot optimize the suppliers, resulting in a high price for the company to complete the procurement process. SUMMARY

[0004] An object of the present application is to provide a computer-implemented supplier recommendation method, a computer device and a storage medium.

[0005] To achieve the above object, the present application adopts the following technical solutions:

[0006] The present application provides a computer-implemented supplier recommendation method in the first aspect, comprising:

[0007] The historical auction data set is used to train the PageRank model, wherein the historical auction data set includes the subcategory information as the input of the model and the supplier bid corresponding to all bidding sessions under the subcategory, and the winning probability of each supplier as the output of the model;

[0008] The subcategory information is input into the trained PageRank model, and the recommended supplier is output for inviting the recommended supplier to participate in the bidding before the bidding.

[0009] Optionally, the historical auction data set is used to train the PageRank model, comprising:

[0010] The historical auction data set is processed, and the ranking of the corresponding supplier of each bidding under the subcategory is obtained according to the final supplier bid;

[0011] Converting the ranking of the corresponding supplier of each bidding into a mutual relative win-loss matrix, to obtain N mutual relative win-loss matrices, where N is the total number of bidding under the sub-category, and there are M suppliers participating in N bidding;

[0012] Matrix aggregation is performed on the N mutual relative win-loss matrices to obtain an M*M matrix;

[0013] Comparing all values containing the diagonal line of the M*M matrix, if the diagonal line containing the value is empty, a first hyperparameter value is filled in;

[0014] The diagonal line of the M*M matrix is filled with a second hyperparameter value;

[0015] Rescaling each column of the M*M matrix to make the sum of each column equal to 1 to obtain a state transition matrix.

[0016] Optionally, the first hyperparameter value is 0.1, and the second hyperparameter value is 1.

[0017] Optionally, it further comprises:

[0018] According to the state transition matrix, a network relationship graph is obtained, which is composed of nodes and edges between nodes; each node in the network relationship graph corresponds to a supplier;

[0019] According to the random walk algorithm, the winning probability of each supplier is obtained by iterative calculation.

[0020] Optionally, the random walk algorithm comprises:

[0021] Starting from at least one node in the M suppliers, with a probability of d, the next node pointed to by the current node is walked, and with a probability of 1-d, it jumps back to any node in the M supplier set; wherein 0

[0022] Optionally, the random walk algorithm further comprises:

[0023] If the current node is a terminal node in the network relationship graph, jump back to any node in the M supplier set with a probability of d.

[0024] Optionally, the random walk algorithm iteratively calculates the winning probability of each supplier, comprising:

[0025] In each iteration calculation process, for the first node in the network relationship graph, the winning probability between the first node and the M supplier set is calculated according to the following formula:

[0026]

[0027] Wherein, the first node is any one node of M supplier set; PPR(t) is the corresponding coefficient in the state transition matrix before this iteration calculation; PPR s (t) is the winning probability between node t and the M supplier set obtained by this iteration calculation; d is the walk probability; N is the number of nodes pointing to the node t; N out (i) represents the number of all outgoing edges of node i; N in (t) represents all nodes pointing to node t.

[0028] Optionally, the step of inputting the sub-category information into the trained PageRank model and outputting the recommended supplier includes:

[0029] The winning probability of the M suppliers is taken as the weight of the recommended supplier, and the corresponding supplier is generated by weighted and non-replacement sampling n times, wherein n is the optimal recommendation number of each category, which is automatically generated in the PageRank model training process.

[0030] The second aspect of the present application provides a computer device, comprising a processor and a memory, the memory has a computer program stored thereon, and the processor implements the method according to the first aspect of the present application when executing the program.

[0031] The third aspect of the present application provides a computer readable storage medium, which has a computer program stored thereon, and the program is executed by a processor to implement the method according to the first aspect of the present application.

[0032] The beneficial effects of the present application are as follows:

[0033] The technical solution of the present application trains the PageRank model by using the historical bidding data set, then inputs the sub-category information of the current bidding into the trained PageRank model, outputs the recommended supplier, and invites the recommended supplier to participate in the bidding, so that the auction party or the purchaser can complete the bidding at a price favorable to themselves. BRIEF DESCRIPTION OF DRAWINGS

[0034] The specific embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0035] Figure 1 A hardware architecture diagram for implementing the method of the present application according to an embodiment of the present application is shown.

[0036] Figure 2 A flowchart of a computer-implemented supplier recommendation method according to an embodiment of the present application is shown.

[0037] Figure 3An information input interface schematic diagram under the bidding subcategory of the embodiment of the present application is shown.

[0038] Figure 4 A display interface schematic diagram of the supplier recommendation method of the embodiment of the present application is shown.

[0039] Figure 5 A structural schematic diagram of a computer system implementing the method provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0040] In order to more clearly illustrate the present application, the present application is further described below in conjunction with preferred embodiments and the accompanying drawings. Similar components are denoted by the same reference numerals in the drawings. It should be understood by those skilled in the art that the specific descriptions below are illustrative rather than limiting, and should not limit the protection scope of the present application.

[0041] One embodiment of the present application is described by taking a procurement scenario as an example.

[0042] Before the procurement process starts, the purchaser needs to create a bidding matter, in which the information of the bidding subject (such as the name of the product to be procured, etc.), the supplier information (such as the name, bank account information, etc.) need to be entered. It is crucial to invite which suppliers to participate in this bidding in the bidding matter. For example, there are more than a thousand beer production supplier manufacturers. For this bidding information, which supplier manufacturers are invited to participate in the bidding so that the purchaser can complete the procurement process at a price favorable to itself and achieve the purpose of saving procurement costs is a particular concern of the purchaser.

[0043] Embodiment One

[0044] To this end, the embodiment of the present application proposes a computer-implemented supplier recommendation method. In one embodiment, the supplier recommendation method can be implemented on a hardware architecture such as Figure 1 .

[0045] The computer device 105 is provided with a supplier recommendation model. In the embodiment of the present application, the subcategory information is input into the trained PageRank model, and the PageRank model outputs n high-quality suppliers recommended from M suppliers, n < M. Further, the model provides the high-quality supplier information under the subcategory to the purchaser via the network 103, and the purchaser sets up the procurement management system installed in the computer device 107 according to the high-quality supplier information to invite these high-quality suppliers to participate in the bidding.

[0046] The network 103 is a medium for providing a communication link between the user terminal 101 and the computer devices 105 and 107. The network 103 can include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0047] The user terminal 101 can be various electronic devices with a display screen, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, etc.

[0048] The computer devices 105, 107 can be any device with a processor and memory, such as a server.

[0049] It should be noted that Figure 1 The number of user terminals, networks and computer devices in the above description is only illustrative. According to the needs of implementation, there can be any number of terminal devices, networks and servers.

[0050] The present application provides a computer-implemented supplier recommendation method. In one specific example, the supplier manufacturer can be a manufacturer of physical products, or a manufacturer of raw materials for producing products, such as a digital product manufacturer, a barley malt manufacturer for producing beer, etc. The present application is not limited thereto.

[0051] As Figure 2 shown, the method comprises the following steps:

[0052] S10, training a PageRank model using a historical bidding data set, wherein the historical bidding data set includes subcategory information as input of the model and supplier bids corresponding to all bidding sessions under the subcategory, and winning probability of each supplier as output of the model.

[0053] In order to use the PageRank model to output recommended high-quality suppliers, it needs to be trained first.

[0054] The PageRank model can be trained using any suitable data set. In the present embodiment, the PageRank model is trained using a historical bidding data set, which includes subcategory information as input of the model and supplier bids corresponding to all bidding sessions under the subcategory, and winning probability of each supplier as output of the model.

[0055] In one possible implementation, the training of the PageRank model using the historical bidding data set comprises:

[0056] Data processing is performed on the historical bidding data set to obtain the ranking of the corresponding supplier for each bidding session under the subcategory according to the final supplier bid;

[0057] Converting the ranking of the corresponding suppliers of each round of bidding into a mutual relative win-loss matrix, to obtain N mutual relative win-loss matrices, where N is the total number of rounds of bidding under the sub-category, and there are M suppliers participating in the N rounds of bidding;

[0058] Matrix aggregation is performed on the N mutual relative win-loss matrices to obtain a M*M matrix;

[0059] All values contained in the M*M matrix are compared along the diagonal line, and if the diagonal line of the contained values is empty, a first hyperparameter value is filled in;

[0060] The diagonal line of the M*M matrix is filled with a second hyperparameter value;

[0061] Each column of the M*M matrix is rescaled proportionally so that the sum of each column is 1, to obtain a state transition matrix.

[0062] In a specific example, for example, there are three rounds of bidding under this category information, and there are four suppliers participating in the three rounds of bidding.

[0063] For the first round of bidding, there are three suppliers A, B and C, and the mutual relative win-loss matrix converted from the final bids of the three suppliers is:

[0064] A B C A 1 1 B 1 C

[0065] For the second round of bidding, there are three suppliers B, C and D, and the mutual relative win-loss matrix converted from the final bids of the three suppliers is:

[0066] B C D B 1 1 C 1 D

[0067] For the third round of bidding, there are two suppliers C and D, and the mutual relative win-loss matrix converted from the final bids of the two suppliers is:

[0068] C D C D 1

[0069] Wherein, 1 in row A and column B indicates that A wins one round against B, and similarly, 1 in row A and column C indicates that A wins one round against C, 1 in row B and column C indicates that B wins one round against C, and so on.

[0070] Matrix aggregation is performed on each round of bidding under this category, and the mutual relative win-loss matrices of the above three rounds of bidding are matrix aggregated to obtain a 4*4 matrix:

[0071] A B C D A 1 1 B 2 1 C 1 D 1

[0072] All the values of the 4*4 matrix are compared along the diagonal line, and if the diagonal line of the values is empty, a first hyperparameter value is filled in. In a specific example, the first hyperparameter value is 0.1, indicating that the mutual victory and defeat is not absolute, and the weak also has a small probability of winning opportunity, and the following matrix is obtained:

[0073] A B C D A 1 1 B 0.1 2 1 C 0.1 0.1 1 D 0.1 1

[0074] The diagonal line of the 4*4 matrix is filled with a second hyperparameter value. In a specific example, the second hyperparameter value is 1, indicating that the strong has a high probability of winning opportunity, and the following matrix is obtained:

[0075] A B C D A 1 1 1 B 1 / 10 1 2 1 C 1 / 10 1 / 10 1 1 D 1 / 10 1 1

[0076] Each column of the 4*4 matrix is rescaled proportionally so that the sum of each column is 1, and a state transition matrix is obtained, which is:

[0077] A B C D A 10 / 12 10 / 22 1 / 5 B 1 / 12 10 / 22 2 / 5 1 / 3 C 1 / 12 1 / 22 1 / 5 1 / 3 D 1 / 22 1 / 5 1 / 3

[0078] The state transition matrix is a Markov matrix.

[0079] In a possible implementation, the method further includes:

[0080] According to the state transition matrix, a network relationship graph is obtained, the network relationship graph is composed of nodes and edges between nodes; each node in the network relationship graph corresponds to a supplier;

[0081] According to the random walk algorithm, the winning probability of each supplier is obtained by iterative calculation.

[0082] In a possible implementation, the random walk algorithm includes:

[0083] Starting from at least one node in the M suppliers, with a probability of d, walking to the next node pointed to by the current node, and with a probability of 1-d, jumping back to any node in the M supplier set; wherein 0

[0084] In a possible implementation, the random walk algorithm further includes:

[0085] If the current node is a terminal node in the network relationship graph, jump back to any node in the M supplier set with a probability of d.

[0086] In a possible implementation, the random walk algorithm iteratively calculates the winning probability of each supplier, including:

[0087] In each iteration of the calculation, for the first node in the network graph, the winning probability between the first node and the set of M suppliers is calculated using the following formula:

[0088]

[0089] Wherein, the first node is any node in the set of M suppliers; PPR(t) is the corresponding coefficient of node t in the state transition matrix before the current iteration calculation; PPR s (t) represents the winning probability of node t among the M supplier sets obtained in this iteration; d is the walk probability; N is the number of nodes pointing to node t; N out (i) represents the number of all outgoing edges from node i; N in (t) represents all nodes that point to node t.

[0090] In one possible implementation, inputting sub-category information into a trained PageRank model and outputting recommended suppliers includes:

[0091] The winning probabilities of the M suppliers are used as the weights of the recommended suppliers. The corresponding suppliers are generated by sampling n times with weights and without replacement, where n is the optimal number of recommendations for each category, which is automatically generated during the PageRank model training process.

[0092] In a specific example, the winning probabilities of the five suppliers numbered 01, 02, 03, 04, and 05 are 0.5, 0.2, 0.1, 0.1, and 0.1, respectively. The optimal number of recommendations for this bidding process using the PageRank model is 2. Therefore, two weighted samples without replacement are used to generate the corresponding two recommended suppliers. Specifically, the two weighted samples without replacement are used to generate the corresponding two recommended suppliers as follows:

[0093] The probability that suppliers numbered 01 and 02 are recommended is:

[0094]

[0095] The probability that suppliers numbered 01 and 03 are recommended is:

[0096]

[0097] The probability that suppliers numbered 01 and 04 are recommended is:

[0098]

[0099] The probability that the two suppliers numbered 01 and 05 are recommended is:

[0100]

[0101] The probability of the 2 suppliers numbered 02, 03 being recommended is:

[0102]

[0103] The probability of the 2 suppliers numbered 02, 04 being recommended is:

[0104]

[0105] The probability of the 2 suppliers numbered 02, 05 being recommended is:

[0106]

[0107] The probability of the 2 suppliers numbered 03, 04 being recommended is:

[0108]

[0109] The probability of the 2 suppliers numbered 03, 05 being recommended is:

[0110]

[0111] The probability of the 2 suppliers numbered 04, 05 being recommended is:

[0112]

[0113] In one specific example, the purchaser inputs the information under the current bidding subcategory into the visual interactive interface as shown in FIG. 4. Figure 3 The purchaser clicks the submit button, and the current bidding subcategory information as shown in FIG. 4 is sent to the PageRank model. Figure 3

[0114] As shown in FIG. 5, the number of recommended suppliers, the weight of the suppliers, the name and contact information of the recommended suppliers, and the like are presented on the interactive interface of the model, as follows: Figure 4

[0115] Recommended number: 5

[0116] Weight: 2.1, Email: None, Supplier name: A Company, Predefined: No, Feedback: Select;

[0117] Weight: 1.3, Email: None, Supplier name: B Company, Predefined: No, Feedback: Select;

[0118] Weight: 0.8, Email: aemail, Supplier name: C Company, Predefined: Yes, Feedback: Select; ​​

[0119] Weight: 0.8, Email: b email, Supplier Name: Company D, Predefined: Yes, Feedback: Select;

[0120] Weight: 0.8, Email: None, Supplier Name: Company E, Predefined: Yes, Feedback: Select.

[0121] On this interactive interface, purchasing staff can save supplier recommendations for this subcategory of bidding information by clicking "Submit Feedback," and then use these saved supplier recommendations to publish the bidding information when the purchasing party releases the bidding matter.

[0122] The technical solution described in this invention uses historical bidding datasets to train the PageRank model. Then, the sub-category information of the current bidding is input into the trained PageRank model, which outputs recommended suppliers. Bidders invite the recommended suppliers to participate in the bidding, which allows the auctioneer or purchaser to complete the bidding at a price that is favorable to themselves.

[0123] Example 2

[0124] like Figure 1 The computer devices 105 and 107 may include, for example: Figure 5 The architectures shown are suitable for implementing the vendor-recommended method for computer implementation provided in Embodiment 1 above. This architecture includes a central processing unit (CPU), which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) or a program loaded from storage into random access memory (RAM). The RAM also stores various programs and data required for the operation of the computer system. The CPU, ROM, and RAM are connected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0125] The following components are connected to the I / O interface: input sections including keyboards, mice, etc.; output sections including liquid crystal displays (LCDs) and speakers, etc.; storage sections including hard disks, etc.; and communication sections including network interface cards such as LAN cards and modems. The communication sections perform communication processing via networks such as the Internet. Drives are also connected to the I / O interface as needed. Removable media, such as disks, optical disks, magneto-optical disks, semiconductor memories, etc., are installed on the drive as needed so that computer programs read from them can be installed into the storage section as required.

[0126] In particular, according to the present embodiment, the processes described above in the flowcharts can be implemented as a computer software program. For example, the present embodiment includes a computer program product comprising a computer program tangibly embodied on a computer readable medium, the computer program containing program code for executing the methods illustrated in the flowcharts. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section, and / or installed from a removable medium.

[0127] The flowcharts and diagrams in the drawings illustrate the architecture, functionality, and operations of possible implementations of systems, methods and computer program products according to various embodiments. In this regard, each block in the flowcharts or diagrams can represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks can sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the diagram and / or flowchart illustrations, and combinations thereof, can be implemented by special purpose hardware-based systems that perform the specified functions or operations, or combinations of special purpose hardware and computer instructions.

[0128] Embodiment Three

[0129] As another aspect, the present embodiment also provides a non-transitory computer storage medium, which can be the non-transitory computer storage medium contained in the apparatus in the above embodiments, or can exist separately from the terminal and not be assembled into the terminal. The non-transitory computer storage medium stores one or more programs, when the one or more programs are executed by the apparatus as shown in the above embodiments, the computer-implemented vendor recommendation method provided by the above embodiment one is implemented. Figure 5

[0130] Those skilled in the art can understand that although the above embodiments are described in the procurement scenario, the teachings of the present application can obviously be used in other bidding scenarios, such as auctions, in which the final price is the highest price, and the procurement system can be replaced by a commercial auction system.

[0131] ​In the description of the present application, it needs to be explained that the terms "upper", "lower" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. Unless otherwise expressly specified and limited, the terms "mounting", "connecting", "connecting" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be the communication between two elements inside. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0132] It also needs to be explained that in the description of the present application, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or equipment. Without more limitation, the element defined by the sentence "including a…" does not exclude the presence of other identical elements in the process, method, article or equipment including the element.

[0133] Obviously, the above embodiments of the present application are only examples for clearly illustrating the present application, and are not a limitation on the embodiments of the present application. For those skilled in the art, on the basis of the above description, other different forms of changes or variations can also be made, and it is impossible to enumerate all the embodiments here. Any obvious changes or variations derived from the technical solutions of the present application are still within the protection scope of the present application.

Claims

1. A computer-implemented method of vendor recommendation, the method comprising: The method comprises: training a PageRank model using a historical bidding dataset, wherein the historical bidding dataset comprises subcategory information as input of the model, and supplier quotes corresponding to all bidding sessions under the subcategory, and winning probability of each supplier as output of the model; inputting the subcategory information into the trained PageRank model to output recommended suppliers for inviting to bid before bidding; the training of the PageRank model using the historical bidding dataset comprises: performing data processing on the historical bidding dataset to obtain the ranking of corresponding suppliers of each bidding session under the subcategory according to final supplier quotes; converting the ranking of corresponding suppliers of each bidding session into a mutual relative win-lose matrix to obtain N mutual relative win-lose matrices, wherein N is the total number of bidding sessions under the subcategory, and there are M suppliers participating in the N bidding sessions; performing matrix aggregation on the N mutual relative win-lose matrices to obtain an M*M matrix; comparing all values contained in the M*M matrix along the diagonal line, and if the diagonal line of the contained values is empty, a first hyperparameter value is filled in; filling the diagonal line of the M*M matrix with a second hyperparameter value; re-scaling each column of the M*M matrix proportionally to make the sum of each column equal to 1 to obtain a state transition matrix; the inputting of the subcategory information into the trained PageRank model to output recommended suppliers comprises: taking the winning probability of the M suppliers as the weight of the recommended suppliers, and generating corresponding suppliers through weighted and non-replacement sampling n times, wherein n is the optimal recommended number of each category and is automatically generated in the PageRank model training process; obtaining a network relationship graph according to the state transition matrix, wherein the network relationship graph is composed of nodes and edges between nodes; each node in the network relationship graph corresponds to a supplier; obtaining the winning probability of each supplier through iterative calculation according to the random walk algorithm.

2. The provider recommendation method of claim 1, wherein, The first hyperparameter value is 0.1, and the second hyperparameter value is 1.

3. The provider recommendation method of claim 2, wherein, The random walk algorithm comprises: starting from at least one node in the M suppliers, walking to the next node pointed to by the current node with a probability of d, and jumping back to any node in the M suppliers with a probability of 1-d; wherein 0 4. The provider recommendation method of claim 3, wherein, The random walk algorithm further comprises: if the current node is a terminal node in the network relationship graph, jumping back to any node in the M suppliers with a probability of d.

5. The provider recommendation method of claim 1, wherein, The random walk algorithm iteratively calculates to obtain the winning probability of each supplier, comprising: in each iteration calculation process, for the first node in the network relationship graph, the winning probability between the first node and the M suppliers is calculated according to the following formula: The first node is any one node of M suppliers; is a corresponding coefficient in the state transition matrix before the current iteration calculation of the node t; is the winning probability between the node t and the M suppliers obtained by the current iteration calculation; d is a wandering probability; K is the number of nodes pointing to the node t; represents the number of all outgoing edges of the node i; represents all nodes pointing to the node t. 6.A computer device, comprising a processor and a memory, wherein the memory has stored a computer program, and the computer device is characterized in that, The processor implements the method of any one of claims 1-5 when executing the program.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that The program is executed by the processor to implement the method of any one of claims 1-5.

Images