Supplier recommendation method, system and related equipment based on collaborative filtering algorithm

Through the RFMC model and clustering algorithm based on collaborative filtering algorithm, the problems of low supplier recommendation quality and data sparsity in the existing technology are solved, and high-precision supplier recommendations are realized for manufacturers, meeting the preferences of parts demand improves production efficiency and quality.

CN114840769BActive Publication Date: 2024-12-03GUANGDONG UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202210680497.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-06-16
Publication Date
2024-12-03
Estimated Expiration
2042-06-16

AI Technical Summary

Technical Problem

The existing supplier recommendation methods cannot effectively recommend the components based on the manufacturer's preferences, resulting in low recommendation quality and data sparsity problems.

Method used

Using a method based on collaborative filtering algorithm, the RFMC model is constructed, the RFMC value is calculated, the RFMC value is assigned, the scoring matrix is ​​filled, the clustering algorithm and Baseline algorithm are used to perform matrix filling, the product similarity is calculated, and the supplier recommendation list is generated, combining the RFMC model and Interest calculation supplier recommendations.

Benefits of technology

It improves the accuracy of supplier recommendations, solves the problem of data sparseness, meets the manufacturers' preference for parts, and improves purchase efficiency and production quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114840769B_ABST
    Figure CN114840769B_ABST
Patent Text Reader

Abstract

The present invention is applicable to the field of data analysis, and provides a supplier recommendation method, system and related equipment based on a collaborative filtering algorithm, the method comprising: S1, constructing an RFMC model between manufacturers and commodities, and calculating RFMC values; S2, weighted summing the RFMC values ​​to obtain RFMC scores; S3, constructing a scoring matrix using the RFMC scores; S4, dividing the scoring matrix using a clustering algorithm to obtain a scoring submatrix, and using a Baseline algorithm to calculate and fill the scoring matrix; S5, calculating the similarity between different commodities, and constructing a commodity similarity matrix; S6, calculating a plurality of similar commodities according to the commodity similarity matrix, grouping the commodities according to the categories of the suppliers, obtaining the interest of the manufacturer in the commodities, and generating a supplier recommendation list. The present invention realizes the supplier recommendation for the manufacturer by calculating the interest of the manufacturer in the commodities.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of data analysis, and in particular relates to a supplier recommendation method, system and related equipment based on a collaborative filtering algorithm. Background Art

[0002] Today, with the deep integration of industrialization and informatization, the total amount of data is growing exponentially. The manufacturing industry is facing a huge and complex amount of data. As a resource, how to mine, develop and utilize data has become a critical issue for enterprises. In previous studies, most of the research on the relationship between customers and suppliers has focused on customers, but there has been little attention and analysis on suppliers. The evaluation and selection of suppliers is a classic problem for optimizing resource allocation in the manufacturing process. With the development of the times, the manufacturing model has also undergone adaptive changes, and the relevant supply chain management theory has also been continuously innovated. Against this background, the connotation of this classic problem has also been updated accordingly.

[0003] Once the recommendation algorithm was proposed, it was accompanied by continuous evolution and optimization. In this process, various algorithms and technical forms emerged, such as clustering, association rules, collaborative recommendation, etc. Among them, collaborative recommendation technology is the most successful recommendation technology. The basic idea of ​​the collaborative algorithm is to find the rating data in the user cluster with similar ratings to the target user, and form a recommendation list for the target user based on the rating data. However, there is an important problem in the collaborative filtering algorithm, which is the mitigation of the sparsity of the rating matrix. The traditional way to alleviate the sparsity of the rating matrix is ​​mainly to use the mean filling method and the majority filling method. These methods all have the problem of single filling results and low recommendation quality. In addition, in terms of the evaluation of recommended objects, the RFM model has strong interpretability and operability, and has great advantages, but its application in different industries is different, and it needs to be adaptively improved.

[0004] For example, in the field of automobile manufacturing, the recommendation of parts suppliers is crucial for automobile manufacturers to prepare for production. A highly accurate supplier recommendation method can largely meet the manufacturer's parts preference needs, improve procurement efficiency, enhance production quality, reduce costs and inventory, and optimize the overall benefits of the production group. Therefore, how to better recommend suppliers has become an urgent problem to be solved in the industry. Summary of the invention

[0005] To solve the above problems, an embodiment of the present invention provides a supplier recommendation method, system and related equipment based on a collaborative filtering algorithm to solve the problem that existing recommendation methods cannot make recommendations based on manufacturers' preferences for parts.

[0006] In a first aspect, an embodiment of the present invention provides a supplier recommendation method based on a collaborative filtering algorithm, the method comprising the following steps:

[0007] S1. Construct an RFMC model between manufacturers and commodities, and calculate the RFMC value corresponding to the RFMC model. The manufacturers and commodities include multiple types, each of which corresponds to a supplier. The RFMC value includes four indicators: purchase time span R, purchase number F, total purchase amount M, and total purchase amount C.

[0008] S2, assigning weights to different RFMC values, and performing weighted summation on the RFMC values ​​according to the weights to obtain RFMC scores between the manufacturers and the products;

[0009] S3, using the RFMC score to fill and construct a score matrix between the manufacturer and the product;

[0010] S4. Use a clustering algorithm to divide the scoring matrix to obtain a scoring sub-matrix, and use a Baseline algorithm combined with the scoring sub-matrix to fill the scoring matrix to obtain a filled scoring matrix;

[0011] S5. Calculate the similarities between different commodities according to the filled rating matrix, and construct a commodity similarity matrix;

[0012] S6. Query the commodities purchased by the manufacturer according to the RFMC model, calculate multiple commodities with high similarity according to the commodity similarity matrix, group the commodities according to the categories of the suppliers, obtain the interest of the manufacturer in the commodities, and generate a supplier recommendation list according to the similarity and the interest.

[0013] Furthermore, in step S1, when calculating the RFMC value corresponding to the RFMC model, a three-point evaluation method is used that is simplified based on a five-point evaluation method.

[0014] Furthermore, in step S2, the priority diagram method is used to calculate the respective weights of different RFMC values.

[0015] Furthermore, in step S4, the clustering algorithm is a K-means clustering algorithm.

[0016] Furthermore, in step S6, in the supplier recommendation list, the sorting strategy of the suppliers is a Top-N strategy.

[0017] Furthermore, in step S6, the manufacturer is defined as f, the product is defined as c, and the interest level of the manufacturer f in the product c is defined as I fc , the interest level I fc Satisfies the following calculation formula:

[0018]

[0019] Among them, N(f) represents the set of commodities preferred by the manufacturer f, x c1c2 represents the similarity between the product c1 and the product c2, p fc2 represents the preference of the manufacturer f for the product c2, and X(c1,K) represents the set of the first K products that have the highest similarity with the product c1.

[0020] Furthermore, in step S6, the supplier is defined as s, and the manufacturer interest of the manufacturer f in the supplier s is Q fs , the manufacturer's interest Q fs Satisfies the following calculation formula:

[0021]

[0022] Wherein, T(s) represents the set of all the commodities provided by the supplier s.

[0023] In a second aspect, an embodiment of the present invention further provides a supplier recommendation system based on a collaborative filtering algorithm, comprising:

[0024] The RFMC calculation module is used to construct an RFMC model between manufacturers and commodities, and calculate the RFMC value corresponding to the RFMC model. The manufacturers and commodities include multiple types, each of which corresponds to a supplier. The RFMC value includes four indicators: purchase time span R, purchase number F, total purchase amount M, and total purchase amount C.

[0025] A weight calculation module, used for assigning weights to different RFMC values, and performing weighted summation on the RFMC values ​​according to the weights to obtain RFMC scores between the manufacturers and the products;

[0026] A rating matrix construction module, used to construct a rating matrix between the manufacturer and the product using the RFMC rating;

[0027] A matrix filling module, used to divide the scoring matrix using a clustering algorithm to obtain a scoring sub-matrix, and fill the scoring matrix using a Baseline algorithm combined with the scoring sub-matrix to obtain a filled scoring matrix;

[0028] A similarity matrix construction module, used to calculate the similarities between different commodities according to the filled rating matrix, and to construct a commodity similarity matrix;

[0029] The recommendation output module is used to query the commodities purchased by the manufacturer according to the RFMC model, calculate multiple commodities with high similarity according to the commodity similarity matrix, group the commodities according to the categories of the suppliers, and obtain the interest of the manufacturer in the commodities, and generate a supplier recommendation list according to the similarity and the interest.

[0030] In a third aspect, an embodiment of the present invention further provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps in the supplier recommendation method based on the collaborative filtering algorithm as described in any one of the above embodiments are implemented.

[0031] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the supplier recommendation method based on the collaborative filtering algorithm as described in any one of the above embodiments are implemented.

[0032] The beneficial effect achieved by the embodiment of the present invention is that by calculating the manufacturer's interest in the product and applying the conversion idea, the manufacturer's interest in the corresponding supplier is obtained, and the product supplier recommendation for the manufacturer is realized. At the same time, the improved RFMC model is used and clustered in combination with the clustering algorithm to obtain the scoring matrix, which can improve the accuracy of the recommendation and solve the data sparsity problem in the recommendation algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flowchart of the steps of a supplier recommendation method based on a collaborative filtering algorithm provided by an embodiment of the present invention;

[0034] Figure 2 is a schematic diagram of the structure of a supplier recommendation system 200 based on a collaborative filtering algorithm provided in an embodiment of the present invention;

[0035] Figure 3 It is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to clearly express the purpose, technical solutions and advantages of the present invention, the following is a more complete description in combination with the accompanying drawings, tables and embodiments. It should be noted that the following embodiments are only intended to illustrate the present invention, but not to limit the scope of the present invention.

[0037] Please refer to Figure 1 , Figure 1 : is a flowchart of a method for recommending suppliers based on a collaborative filtering algorithm provided by an embodiment of the present invention, wherein the method comprises the following steps:

[0038] S1. Construct an RFMC model between manufacturers and commodities, and calculate the RFMC value corresponding to the RFMC model. The manufacturers and commodities include multiple types, each of which corresponds to a supplier. The RFMC value includes four indicators: purchase time span R, purchase number F, total purchase amount M, and total purchase amount C.

[0039] First, the RFM model is introduced. The RFM model is used to measure the customer value and customer profitability in the business model. It includes three indicators: the most recent consumption (Recency), consumption frequency (Frequency), and consumption amount (Monetary). In the embodiment of the present invention, the automobile manufacturing industry is taken as an example for explanation. In the automobile manufacturing industry, the customer is the automobile manufacturer, and the manufacturer produces automobiles by purchasing automobile parts produced by suppliers. The embodiment of the present invention improves the original RFM model to obtain the RFMC model. The RFMC model has an additional indicator of the total purchase quantity (Consumption quantity) compared to the original RFM model. In the embodiment of the present invention, each indicator is defined as the time span R, the number of purchases F, the total purchase amount M, and the total purchase amount C. For example, please refer to Table 1. Table 1 is the calculation method of the RFMC value provided by the embodiment of the present invention.

[0040] Table 1 Calculation method of RFMC value

[0041]

[0042]

[0043] S2. Assign weights to the different RFMC values, and perform weighted summation on the RFMC values ​​according to the weights to obtain RFMC scores between the manufacturers and the products.

[0044] Exemplarily, after obtaining each value in the RFMC model according to the calculation method shown in Table 1, in order to achieve the purpose of evaluating the product and then evaluating the supplier, the embodiment of the present invention needs to perform scoring processing on the obtained RFMC value to obtain the manufacturer's RFMC score for the product. The scoring processing of the four RFMC values ​​mainly combines the manufacturer's historical purchase behavior data on the product, reasonably divides and assigns scores, and forms an evaluation standard.

[0045] In the embodiment of the present invention, the transaction behavior data of the relevant manufacturers for the goods have obvious repeated purchases. In order to facilitate the analysis operation, reduce the classification categories and make it easier to fill in the scoring matrix, a three-point evaluation method simplified on the basis of the five-point evaluation method is used to evaluate the scores of the four RFMC values. For example, the specific score evaluation is shown in Table 2.

[0046] Table 2 RFMC value evaluation

[0047]

[0048]

[0049] Furthermore, before performing the weighted summation of the RFMC values, the priority diagram method is used to calculate the respective weights of different RFMC values. The basic principle is to compare multiple features that need to determine the weights in pairs to determine the importance priority between them.

[0050] Exemplarily, the process of calculating the respective weights of different RFMC values ​​using the priority diagram method in the embodiment of the present invention is as follows:

[0051] S21. Establish a chessboard table for comparing the importance of RFMC values, wherein each RFMC value is listed in the first row and the first column of the table, and the remaining table spaces are filled with the importance between two adjacent values, as shown in Table 3.

[0052] Table 3 Chessboard table of importance between RFMC values

[0053] item R F M C R <![CDATA[s 11 ]]> <![CDATA[s 12 ]]> <![CDATA[s 13 ]]> <![CDATA[s 14 ]]> F <![CDATA[s 21 ]]> <![CDATA[s 22 ]]> <![CDATA[s 23 ]]> <![CDATA[s 24 ]]> M <![CDATA[s 31 ]]> <![CDATA[s 32 ]]> <![CDATA[s 33 ]]> <![CDATA[s 34 <!-- 5 -->]]> C <![CDATA[s 41 ]]> <![CDATA[s 42 ]]> <![CDATA[s 43 ]]> <![CDATA[s 44 ]]>

[0054] S22, compare different RFMC values ​​in pairs to determine their importance priorities. The priority is S ij Indicates that i represents the row value and j represents the column value. If the i-th feature is more important than the j-th feature, then S ij is equal to 1; if feature i is as important as feature j, then S ij is equal to 0.5; if feature i is less important than feature j, then S ij is equal to 0.

[0055] S23, summing up the importance values ​​of the RFMC values ​​in the chessboard table horizontally, and obtaining the sum of the importance scores of the four different RFMC values, and storing it on the far right of the chessboard table. The sum is expressed as S i As shown in Table 4.

[0056] Table 4 The sum of importance scores of RFMC values

[0057] item R F M C The sum of importance R <![CDATA[s 11 ]]> <![CDATA[s 12 ]]> <![CDATA[s 13 ]]> <![CDATA[s 14 ]]> <![CDATA[S1]]> F <![CDATA[s 21 ]]> <![CDATA[s 22 ]]> <![CDATA[s 23 ]]> <![CDATA[s 24 ]]> <![CDATA[S2]]> M <![CDATA[s 31 ]]> <![CDATA[s 32 ]]> <![CDATA[s 33 ]]> <![CDATA[s 34 ]]> <![CDATA[S3]]> C <![CDATA[s 41 ]]> <![CDATA[s 42 ]]> <![CDATA[s 43 ]]> <![CDATA[s 44 ]]> <![CDATA[S4]]>

[0058] S24. Calculate the weight of each indicator according to the obtained sum of the importance of the RFMC values. The weight ω of the i-th indicator satisfies:

[0059]

[0060] After determining the respective weights of the four RFMC values, the RFMC values ​​of each manufacturer for each product can be weighted summed to obtain the final RFMC score.

[0061] S3. Use the RFMC rating to populate and construct a rating matrix between the manufacturer and the product.

[0062] Specifically, in the embodiment of the present invention, the important medium for manufacturers to recommend product suppliers is the scoring matrix. The similarity between products that needs to be obtained later is also obtained through the scoring matrix. For example, the scoring matrix constructed by the RFMC scores obtained in the above steps in the embodiment of the present invention is shown in Table 5, where f i Indicates the manufacturer, c j Indicates the product, g ij represents the rating of manufacturer i on product j.

[0063] Table 5 Example of scoring matrix

[0064] <![CDATA[c1]]> <![CDATA[c2]]> <![CDATA[c3]]> … <![CDATA[c j ]]> <![CDATA[f1]]> <![CDATA[g 11 ]]> <![CDATA[g 13 ]]> … <![CDATA[f2]]> <![CDATA[g 22 ]]> … <![CDATA[g 2j ]]> <![CDATA[f3]]> <![CDATA[g 32 ]]> <![CDATA[g 33 ]]> … … … … … … … <![CDATA[f i ]]> <![CDATA[g i1 ]]> … <![CDATA[g ij ]]>

[0065] S4. Use a clustering algorithm to divide the scoring matrix to obtain a scoring sub-matrix, and use a Baseline algorithm in combination with the scoring sub-matrix to fill the scoring matrix to obtain a filled scoring matrix.

[0066] Specifically, the clustering algorithm used in the embodiment of the present invention is a K-means clustering algorithm. Based on the above rating matrix, when the K-means clustering algorithm is used, k manufacturer vectors and k product vectors are randomly selected as their respective initial cluster centers, and then the remaining manufacturer vectors (g i1 ,g i2 ,…,g ij ) and product vector (g 1j ,g 2j ,…,g ij) with their respective cluster centers, divide these vectors into categories to which the cluster centers with the closest cosine distances belong, calculate the average value of each category, and use the average value to form a new cluster center. Repeat the above steps until the generated cluster centers no longer change, and finally form k clusters, that is, k categories. Run the K-means clustering algorithm again from the beginning on the above data until the SSE values ​​of the k clusters formed at a certain time are all at the inflection point, that is, take the generated k clusters as the best categories.

[0067] For example, a vector 1(x1,x2,…,x n ) and vector 2(y1,y2,…,y n ) as an example, the cosine distance is expressed as:

[0068]

[0069] The clustering optimal category objective function uses the sum of squared errors within the cluster (Sum of the Squared Error, referred to as SSE):

[0070]

[0071] Among them, k represents the number of cluster centers, Z i represents the ith center, dist() represents the cosine distance, and x represents other vectors in the cluster. In the embodiment of the present invention, the rating matrix is ​​decomposed according to the k manufacturer categories and k product categories generated by the above process to form k manufacturer rating sub-matrices and k product rating sub-matrices, and then the generated categories are divided according to the attributes and preferences of the manufacturer itself and the attributes of the product itself.

[0072] For example, please refer to Table 6, Table 7, and Table 8, which are respectively examples of the rating matrix after clustering, an example of the manufacturer rating sub-matrix, and an example of the product rating sub-matrix obtained according to the above process in an embodiment of the present invention.

[0073] Table 6 Example of rating matrix after clustering

[0074]

[0075]

[0076] Table 7 Example of manufacturer scoring sub-matrix

[0077]

[0078] Table 8 Example of product rating submatrix

[0079]

[0080]

[0081] In the embodiment of the present invention, the filling of the rating matrix is ​​mainly based on the product rating sub-matrix and manufacturer rating sub-matrix obtained above. By applying the Baseline algorithm and setting a baseline, the manufacturer's rating and the product's rating deviation are combined to finally obtain a predicted rating.

[0082] Exemplarily, the process of calculating the prediction score using the Baseline algorithm is as follows:

[0083] An important idea of ​​the Baseline algorithm is based on the premise that manufacturers have their own preferences in rating. Some manufacturers generally rate their products higher than other manufacturers. On the contrary, some manufacturers are more demanding about their products, so their ratings will be lower. On this basis, the baseline is expressed as:

[0084]

[0085] in, represents the predicted score of manufacturer f for a product c in the baseline model established above; λ represents the mean score of all manufacturers for the product; b f represents the deviation of the manufacturer's average rating of purchased parts from the overall average. If the manufacturer is more demanding and gives a lower rating to the product, then b f Take a negative value, on the contrary, if the manufacturer's ratings for the products are all high, then b f Take positive value; b c It represents the deviation of the mean rating of the product given by the manufacturers who have purchased the product from the overall mean, that is, whether the product is popular.

[0086] S5. Calculate the similarities between different commodities according to the filled rating matrix, and construct a commodity similarity matrix.

[0087] Specifically, in the embodiment of the present invention, after the manufacturer's rating matrix for the product is filled, the product similarity matrix is ​​constructed by calculating the cosine similarity between two different product vectors. The cosine similarity is expressed as:

[0088]

[0089] In the above formula, c1 and c2 are the vector representations of product c1 and product c2 respectively. Both c1 and c2 are multidimensional vectors, and each dimension takes the corresponding manufacturer's rating value for the product.

[0090] After calculating the similarity between all commodities by the above formula, the commodities are filled into the first row and the first column of the similarity matrix, and the remaining intersecting spaces are filled with the similarity values ​​between each two commodities. Exemplarily, the commodity similarity matrix constructed based on the above data of the embodiment of the present invention is shown in Table 9.

[0091] Table 9 Product similarity matrix

[0092] <![CDATA[c1]]> <![CDATA[c2]]> <![CDATA[c3]]> … <![CDATA[c n <!-- 10 -->]]> <![CDATA[c1]]> 1 <![CDATA[x 12 ]]> <![CDATA[x 13 ]]> … <![CDATA[x 1n ]]> <![CDATA[c2]]> <![CDATA[x 21 ]]> 1 <![CDATA[x 23 ]]> … <![CDATA[x 2n ]]> <![CDATA[c3]]> <![CDATA[x 31 ]]> <![CDATA[x 32 ]]> 1 … <![CDATA[x 3n ]]> … … … … … … <![CDATA[c n ]]> <![CDATA[x n1 ]]> <![CDATA[x n2 ]]> <![CDATA[x n3 ]]> … 1

[0093] S6. Query the commodities purchased by the manufacturer according to the RFMC model, calculate multiple commodities with high similarity according to the commodity similarity matrix, group the commodities according to the categories of the suppliers, obtain the interest of the manufacturer in the commodities, and generate a supplier recommendation list according to the similarity and the interest.

[0094] Specifically, in the embodiment of the present invention, after the product similarity matrix is ​​constructed, in order to give the manufacturer an interesting supplier recommendation result, a Top-N recommendation strategy is adopted, where N represents the number of products in the recommendation list. In addition, the generation of recommendation results requires a standard for measuring the level of recommendation. In the embodiment of the present invention, interest is used as the measurement standard. More specifically, the manufacturer is defined as f, the product is defined as c, and the interest of the manufacturer f in the product c is I fc , the interest level I fc Satisfies the following calculation formula:

[0095]

[0096] Among them, N(f) represents the set of commodities preferred by the manufacturer f, x c1c2 represents the similarity between the product c1 and the product c2, p fc2 represents the preference of the manufacturer f for the product c2, and X(c1,K) represents the set of the first K products that have the highest similarity with the product c1.

[0097] In the embodiment of the present invention, if a manufacturer's interest value for a certain product is high, it means that the manufacturer is more interested in the product and is more likely to trade with it. Therefore, according to the principle of reverse order of interest, a manufacturer's interest list for parts is first generated.

[0098] Define that if producer f has purchased product c, it means that the product is preferred by producer f, P fc The value of is 1.

[0099] Since the ultimate recommendation target is the supplier of the goods, and the goods themselves are only used to express the interest preferences of the manufacturers, after obtaining the manufacturer's recommendation list for the goods, that is, after obtaining the manufacturer's interest in all the goods, these goods are grouped according to the supplier's attributes, that is, the manufacturer's interest in the supplier is the sum of the interest in the goods supplied by the supplier.

[0100] Define the supplier as s, and the manufacturer interest of the manufacturer f in the supplier s as Q fs , the manufacturer's interest Q fs Satisfies the following calculation formula:

[0101]

[0102] Wherein, T(s) represents the set of all the commodities provided by the supplier s.

[0103] Finally, the interest in suppliers derived from the manufacturer's interest in the product is sorted in descending order, and the top N items are taken to generate the final manufacturer's recommendation list for product suppliers.

[0104] The beneficial effect achieved by the embodiment of the present invention is that by calculating the manufacturer's interest in the product and applying the conversion idea, the manufacturer's interest in the corresponding supplier is obtained, and the product supplier recommendation for the manufacturer is realized. At the same time, the improved RFMC model is used and clustered in combination with the clustering algorithm to obtain the scoring matrix, which can improve the accuracy of the recommendation and solve the data sparsity problem in the recommendation algorithm.

[0105] The embodiment of the present invention further provides a supplier recommendation system based on a collaborative filtering algorithm, as shown in FIG. 2 . Figure 2 2 is a schematic diagram of the structure of a supplier recommendation system 200 based on a collaborative filtering algorithm provided by an embodiment of the present invention, including:

[0106] The RFMC calculation module 201 is used to construct an RFMC model between manufacturers and commodities, and calculate the RFMC value corresponding to the RFMC model. The manufacturers and commodities include multiple types, each of which corresponds to a supplier. The RFMC value includes four indicators: purchase time span R, purchase number F, total purchase amount M, and total purchase amount C.

[0107] A weight calculation module 202 is used to assign weights to different RFMC values, and perform weighted summation on the RFMC values ​​according to the weights to obtain RFMC scores between the manufacturers and the products;

[0108] A rating matrix construction module 203 is used to construct a rating matrix between the manufacturer and the product using the RFMC ratings;

[0109] A matrix filling module 204 is used to divide the scoring matrix using a clustering algorithm to obtain a scoring sub-matrix, and fill the scoring matrix using a Baseline algorithm in combination with the scoring sub-matrix to obtain a filled scoring matrix;

[0110] A similarity matrix construction module 205 is used to calculate the similarities between different commodities according to the filled rating matrix and construct a commodity similarity matrix;

[0111] The recommendation output module 206 is used to query the commodities purchased by the manufacturer according to the RFMC model, and calculate multiple commodities with high similarity according to the commodity similarity matrix, group the commodities according to the categories of the suppliers, and obtain the interest of the manufacturer in the commodities, and generate a supplier recommendation list according to the similarity and the interest.

[0112] The supplier recommendation system 200 based on collaborative filtering algorithm can implement the steps in the supplier recommendation method based on collaborative filtering algorithm in the above embodiment, and can achieve the same technical effect. Please refer to the description in the above embodiment, which will not be repeated here.

[0113] The embodiment of the present invention also provides a computer device, please refer to Figure 3 , Figure 3 300 is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. The computer device 300 includes: a memory 302, a processor 301, and a computer program stored in the memory 302 and executable on the processor 301.

[0114] The processor 301 calls the computer program stored in the memory 302 to execute the steps of the supplier recommendation method based on the collaborative filtering algorithm provided in the embodiment of the present invention. Figure 1 , specifically including:

[0115] S1. Construct an RFMC model between manufacturers and commodities, and calculate the RFMC value corresponding to the RFMC model. The manufacturers and commodities include multiple types, each of which corresponds to a supplier. The RFMC value includes four indicators: purchase time span R, purchase number F, total purchase amount M, and total purchase amount C.

[0116] Furthermore, in step S1, when calculating the RFMC value corresponding to the RFMC model, a three-point evaluation method is used that is simplified based on a five-point evaluation method.

[0117] S2. Assign weights to the different RFMC values, and perform weighted summation on the RFMC values ​​according to the weights to obtain RFMC scores between the manufacturers and the products.

[0118] Furthermore, in step S2, the priority diagram method is used to calculate the respective weights of different RFMC values.

[0119] S3. Use the RFMC rating to populate and construct a rating matrix between the manufacturer and the product.

[0120] S4. Use a clustering algorithm to divide the scoring matrix to obtain a scoring sub-matrix, and use a Baseline algorithm in combination with the scoring sub-matrix to fill the scoring matrix to obtain a filled scoring matrix.

[0121] Furthermore, in step S4, the clustering algorithm is a K-means clustering algorithm.

[0122] S5. Calculate the similarities between different commodities according to the filled rating matrix, and construct a commodity similarity matrix.

[0123] S6. Query the commodities purchased by the manufacturer according to the RFMC model, calculate multiple commodities with high similarity according to the commodity similarity matrix, group the commodities according to the categories of the suppliers, obtain the interest of the manufacturer in the commodities, and generate a supplier recommendation list according to the similarity and the interest.

[0124] Furthermore, in step S6, in the supplier recommendation list, the sorting strategy of the suppliers is a Top-N strategy.

[0125] Furthermore, in step S6, the manufacturer is defined as f, the product is defined as c, and the interest level of the manufacturer f in the product c is defined as I fc , the interest level I fc Satisfies the following calculation formula:

[0126]

[0127] Among them, N(f) represents the set of commodities preferred by the manufacturer f, x c1c2 represents the similarity between the product c1 and the product c2, p fc2 represents the preference of the manufacturer f for the product c2, and X(c1,K) represents the set of the first K products that have the highest similarity with the product c1.

[0128] Furthermore, in step S6, the supplier is defined as s, and the manufacturer interest of the manufacturer f in the supplier s is Q fs , the manufacturer's interest Q fs Satisfies the following calculation formula:

[0129]

[0130] Wherein, T(s) represents the set of all the commodities provided by the supplier s.

[0131] The computer device 300 provided in the embodiment of the present invention can implement the steps in the supplier recommendation method based on the collaborative filtering algorithm in the above embodiment, and can achieve the same technical effect. Please refer to the description in the above embodiment and will not be repeated here.

[0132] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes and steps in the supplier recommendation method based on the collaborative filtering algorithm provided in the embodiment of the present invention are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0133] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM).

[0134] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0135] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0136] The embodiments of the present invention are described above in conjunction with the accompanying drawings. What is disclosed is only the preferred embodiment of the present invention. However, the present invention is not limited to the above-mentioned specific implementation manner. The above-mentioned specific implementation manner is only illustrative rather than restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms and equivalent changes without departing from the scope of protection of the purpose of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. A supplier recommendation method based on collaborative filtering algorithm, characterized in that: The method comprises the following steps: S1. Construct an RFMC model between manufacturers and commodities, and calculate the RFMC value corresponding to the RFMC model. The manufacturers and commodities include multiple types, each of which corresponds to a supplier. The RFMC value includes four indicators: purchase time span R, purchase number F, total purchase amount M, and total purchase amount C. S2, assigning weights to different RFMC values, and performing weighted summation on the RFMC values ​​according to the weights to obtain RFMC scores between the manufacturers and the products; S3, using the RFMC score to fill and construct a score matrix between the manufacturer and the product; S4. Use a clustering algorithm to divide the scoring matrix to obtain a scoring sub-matrix, and use a Baseline algorithm combined with the scoring sub-matrix to fill the scoring matrix to obtain a filled scoring matrix; S5. Calculate the similarities between different commodities according to the filled rating matrix, and construct a commodity similarity matrix; S6. Query the commodities purchased by the manufacturer according to the RFMC model, calculate multiple commodities with high similarity according to the commodity similarity matrix, group the commodities according to the categories of the suppliers, obtain the interest of the manufacturer in the commodities, and generate a supplier recommendation list according to the similarity and the interest.

2. The supplier recommendation method based on collaborative filtering algorithm according to claim 1, characterized in that: In step S1, when calculating the RFMC value corresponding to the RFMC model, a three-point evaluation method is used that is simplified based on the five-point evaluation method.

3. The supplier recommendation method based on collaborative filtering algorithm according to claim 1, characterized in that: In step S2, the priority diagram method is used to calculate the respective weights of different RFMC values.

4. The supplier recommendation method based on collaborative filtering algorithm according to claim 1, characterized in that: In step S4, the clustering algorithm is a K-means clustering algorithm.

5. The supplier recommendation method based on collaborative filtering algorithm according to claim 1, characterized in that: In step S6, in the supplier recommendation list, the supplier sorting strategy is a Top-N strategy.

6. The supplier recommendation method based on collaborative filtering algorithm according to claim 1, characterized in that: In step S6, the manufacturer is defined as f, the product is defined as c, and the interest level of the manufacturer f in the product c is defined as I fc , the interest level I fc Satisfies the following calculation formula: Among them, N(f) represents the set of commodities preferred by the producer f, represents the similarity between the product c1 and the product c2, represents the preference of the manufacturer f for the product c2, and X(c1,K) represents the set of the first K products that have the highest similarity with the product c1.

7. The supplier recommendation method based on collaborative filtering algorithm according to claim 6, characterized in that: In step S6, the supplier is defined as s, and the manufacturer interest of the manufacturer f in the supplier s is Q fs , the manufacturer's interest Q fs Satisfies the following calculation formula: Wherein, T(s) represents the set of all the commodities provided by the supplier s.

8. A supplier recommendation system based on collaborative filtering algorithm, characterized in that: include: The RFMC calculation module is used to construct an RFMC model between manufacturers and commodities, and calculate the RFMC value corresponding to the RFMC model. The manufacturers and commodities include multiple types, each of which corresponds to a supplier. The RFMC value includes four indicators: purchase time span R, purchase number F, total purchase amount M, and total purchase amount C. A weight calculation module, used for assigning weights to different RFMC values, and performing weighted summation on the RFMC values ​​according to the weights to obtain RFMC scores between the manufacturers and the products; A rating matrix construction module, used to construct a rating matrix between the manufacturer and the product using the RFMC rating; A matrix filling module, used to divide the scoring matrix using a clustering algorithm to obtain a scoring sub-matrix, and fill the scoring matrix using a Baseline algorithm combined with the scoring sub-matrix to obtain a filled scoring matrix; A similarity matrix construction module, used to calculate the similarities between different commodities according to the filled rating matrix, and to construct a commodity similarity matrix; The recommendation output module is used to query the commodities purchased by the manufacturer according to the RFMC model, calculate multiple commodities with high similarity according to the commodity similarity matrix, group the commodities according to the categories of the suppliers, and obtain the interest of the manufacturer in the commodities, and generate a supplier recommendation list according to the similarity and the interest.

9. A computer device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the supplier recommendation method based on the collaborative filtering algorithm as claimed in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the supplier recommendation method based on a collaborative filtering algorithm as claimed in any one of claims 1 to 7 are implemented.

Citation Information

Patent Citations

  • Collaborative recommendation method and system based on agricultural product characteristic attribute comment tendency

    CN111523055A

  • Time-varying collaborative filtering recommendation method

    WO2016191959A1