Product intelligent recommendation method based on collaborative filtering algorithm

Abstract

The present invention relates to the technical field of product recommendation, and particularly relates to an intelligent product recommendation method based on a collaborative filtering algorithm. According to the changing trend of product sales volume of each product at different transaction times, the present invention obtains the popularity of each product; in combination with the transaction time characteristics of the corresponding historical purchased products, the present invention obtains the conformity of each user; for any user as the target user, according to the similarity between each emerging product and different historical purchased products, the present invention obtains the product similarity between the emerging product and the historical purchased products, and screens out emerging recommended products; according to the number of transactions and the distribution of transaction times of different historical purchased products, the present invention obtains the transaction performance of the user corresponding to each historical purchased product; furthermore, the present invention obtains the interest tendency of the target user corresponding to the historical purchased products; and obtains the recommendation index of the emerging recommended products corresponding to the target user. The present invention obtains the accurate recommendation index of each emerging product for users, and improves the accuracy of product recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of product recommendation, and in particular to a product intelligent recommendation method based on a collaborative filtering algorithm. Background Art

[0002] Product recommendation algorithm collaborative filtering, in order to make more accurate product recommendations to users, is usually based on the user's historical behavior data, and then recommends a collection of similar items. However, with the emergence of a large number of emerging products, there is a lack of sufficient historical data, making it difficult for collaborative filtering algorithms to effectively form product recommendations, which is the cold start problem of collaborative filtering algorithms.

[0003] In the prior art, the recommendation of emerging products is initialized by using a pre-trained model on a large data set, that is, a large amount of data is previewed to provide preliminary recommendations for the algorithm in the form of improving product ratings and exposure. However, the use of big data preview can only improve product exposure and consumes a lot of resources. It is impossible to analyze the correlation between emerging products and historical products, and the accuracy of product recommendations is poor. Summary of the invention

[0004] In order to solve the technical problem that the correlation between emerging products and historical products cannot be analyzed and the accuracy of product recommendations is poor, the purpose of the present invention is to provide a product intelligent recommendation method based on a collaborative filtering algorithm. The technical solution adopted is as follows:

[0005] The present invention proposes a product intelligent recommendation method based on a collaborative filtering algorithm, the method comprising:

[0006] Based on the product trading platform, obtain the product sales of emerging products and users' historically purchased products at each transaction time, as well as the number of transactions and transaction time of each historically purchased product by the user;

[0007] According to the sales trend of each product at different transaction times, the popularity of each product is obtained; according to the popularity of each user's different historical purchase products and the transaction time characteristics of the corresponding historical purchase products, the conformity of each user is obtained;

[0008] For any user as the target user, the product similarity between each emerging product and each corresponding historically purchased product is obtained based on the similarity between each emerging product and each corresponding historically purchased product, and the emerging recommended products are screened out;

[0009] Obtain the transaction performance of each historical purchased product corresponding to the user according to the transaction times and transaction time distributions of the user corresponding to different historical purchased products; obtain multiple highly similar product clusters according to the transaction performances of different user-purchased products corresponding to the target user; obtain the interest tendency of the target user corresponding to each historical purchased product according to the transaction feature distributions among different users corresponding to each historical purchased product within the highly similar product cluster.

[0010] Obtain the recommendation index of the emerging recommended product corresponding to the target user according to the popularity of the emerging recommended product, the herd mentality corresponding to the target user, the interest tendency, and the product similarity.

[0011] Furthermore, the method for obtaining the popularity includes:

[0012] For any product, select the transaction moment with the largest product sales value as the popularity moment.

[0013] Obtain the difference between the product sales of the corresponding product and the product sales of other products at each transaction moment within the neighborhood range of the popularity moment as the relative sales difference.

[0014] Obtain the cumulative sum of the relative sales differences of each product at all transaction moments within the neighborhood range of the popularity moment as the popularity of each product.

[0015] Furthermore, the method for obtaining the herd mentality includes:

[0016] For each historical purchased product of each user, obtain the time difference between the corresponding transaction moment and each transaction moment within the neighborhood range of the popularity moment, and select the one with the smallest time difference value as the popularity time difference.

[0017] Obtain the herd mentality of each user according to the popularity and popularity time difference of different historical purchased products of each user. The popularity is positively correlated with the herd mentality, and the popularity time difference is negatively correlated with the herd mentality.

[0018] Furthermore, the method for obtaining the product similarity includes:

[0019] Obtain the multi-dimensional information of the emerging product and the historical purchased products of the target user, and obtain the word vectors corresponding to each dimension information.

[0020] Obtain the cosine similarity of the word vectors of each same dimension information between the emerging product and each historical purchased product as the dimension information similarity; obtain the cumulative value of all dimension information similarities between the emerging product and each historical purchased product as the product similarity between the emerging product and each historical purchased product.

[0021] Furthermore, the method for obtaining the emerging recommended product includes:

[0022] For target users, accumulate the product similarities between emerging products and all historical purchased products, and perform normalization to obtain the overall product similarity between emerging products and all historical purchased products.

[0023] Obtain the mean value of the overall product similarities corresponding to all emerging products as the average overall product similarity.

[0024] If the overall product similarity corresponding to an emerging product is greater than the average overall product similarity, regard the corresponding emerging product as an emerging recommended product.

[0025] Furthermore, the method for obtaining the transaction performance includes:

[0026] Obtain the ratio of the number of transactions of a user corresponding to each historical purchased product to the sum of the number of transactions of all historical purchased products as the first performance coefficient; obtain the mean value of the differences in transaction times between the same historical purchased products and perform a negative correlation mapping as the second performance coefficient.

[0027] Calculate the product of the first performance coefficient and the second performance coefficient corresponding to each historical purchased product as the transaction performance of the user corresponding to each historical purchased product.

[0028] Furthermore, the method for obtaining the high-similar product clusters includes:

[0029] Perform ISODATA clustering on all historical purchased products according to the transaction performances of target users corresponding to different historical purchased products to obtain product clustering clusters.

[0030] Obtain the mean value of the transaction performances of historical purchased products within each product clustering cluster as the local transaction performance; obtain the mean value of the local transaction performances of all product clustering clusters as the overall transaction performance.

[0031] If the local transaction performance of a product clustering cluster is greater than the overall transaction performance, regard the corresponding product clustering cluster as the high-similar product cluster of the target user.

[0032] Furthermore, the method for obtaining the interest tendency includes:

[0033] Obtain the purchase tendency of the target user corresponding to each historical purchased product according to the transaction performances, the number of transactions, and the product sales volume characteristics corresponding to each historical purchased product among different users.

[0034] For the high-similar product clusters, obtain the differences between the transaction times of each historical purchased product corresponding to each user and the transaction times of each other historical purchased product, and perform a negative correlation mapping as the transaction activity; accumulate the transaction activities corresponding to all other purchased products as the transaction frequency.

[0035] Obtain the product of the first performance coefficient corresponding to each historical purchased product for each user and the transaction frequency as the transaction interest degree; according to the deviation of the transaction interest degree of each historical purchased product between the target user and each other user, and the purchase tendency of the target user, obtain the interest tendency of the target user corresponding to each historical purchased product. Both the deviation of the transaction interest degree and the purchase tendency are positively correlated with the interest tendency.

[0036] Further, the method for obtaining the purchase tendency includes:

[0037] For any historical purchased product, obtain the ratio of the number of transactions of the user corresponding to the historical purchased product to the sum of the product sales volumes of the historical purchased product at all transaction times as the user purchase ratio.

[0038] Obtain the product of the transaction performance of the user corresponding to the historical purchased product and the user purchase ratio as the purchase influence index of the user corresponding to the historical purchased product; obtain the average value of the purchase influence indexes of all other users corresponding to the historical purchased product as the average purchase influence index.

[0039] According to the deviation between the purchase influence index of the target user corresponding to the historical purchased product and the average purchase influence index, obtain the purchase tendency of the target user for the historical purchased product.

[0040] Further, the method for obtaining the recommendation index includes:

[0041] Select the one with the largest product similarity value between the emerging recommended product and each historical purchased product within the product cluster with high similarity to the target user, use the corresponding product similarity as the first recommendation coefficient, and use the interest tendency of the corresponding historical purchased product as the second recommendation coefficient.

[0042] According to the popularity of the emerging recommended product, the conformity of the target user, the first recommendation coefficient and the second recommendation coefficient, obtain the recommendation index of the emerging recommended product corresponding to the target user. The popularity, conformity, first recommendation coefficient and second recommendation coefficient are all positively correlated with the recommendation index.

[0043] The present invention has the following beneficial effects:

[0044] According to the changing trend of the sales volume of each product at different transaction times, the present invention obtains the popularity of each product. Through the changing trend of the sales volume, it can accurately capture the sales volume of different products in different time periods and evaluate the popularity of the products. According to the popularity of the products purchased by each user in different histories and the transaction time characteristics of the corresponding products purchased in history, the herd mentality of each user is obtained. For any user as the target user, according to the similarity between each emerging product and the corresponding different products purchased in history, the product similarity between each emerging product and each product purchased in history is obtained, and emerging recommended products are screened out, which can accurately match the emerging products with the products purchased by the user in history, so as to screen out emerging recommended products that meet the user's interests. According to the transaction times and transaction time distributions of the user corresponding to different products purchased in history, the transaction performance of the user corresponding to each product purchased in history is obtained, and the preference of the user's product purchase behavior is understood. According to the transaction performance of the target user corresponding to different products purchased by the user, multiple highly similar product clusters are obtained, and products with similar transaction performance can be classified into one category, which helps to discover the relevance between products. According to the transaction characteristic distributions between different users corresponding to each product purchased in history within the highly similar product cluster, the interest tendency of the target user corresponding to each product purchased in history is obtained, which helps to more deeply understand the needs and preferences of the user, so as to provide more personalized recommendations. The recommendation index of the emerging recommended products corresponding to the target user is obtained. The present invention obtains an accurate recommendation index of each emerging product for the user and improves the accuracy of product recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1 It is a flowchart of a product intelligent recommendation method based on a collaborative filtering algorithm provided by an embodiment of the present invention;

[0047] Figure 2 It is a flowchart of a method for obtaining popularity provided by an embodiment of the present invention;

[0048] Figure 3 It is a flowchart of a method for obtaining product similarity provided by an embodiment of the present invention;

[0049] Figure 4 It is a flowchart of a method for obtaining purchase tendency provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] To further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following, in combination with the accompanying drawings and preferred embodiments, details a product intelligent recommendation method based on collaborative filtering algorithm proposed according to the present invention, including its specific implementation manner, structure, features and effects, as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0052] The following specifically describes in conjunction with the accompanying drawings the specific solution of a product intelligent recommendation method provided by the present invention based on collaborative filtering algorithm.

[0053] Please refer to Figure 1 , which shows a flowchart of a product intelligent recommendation method based on collaborative filtering algorithm provided by an embodiment of the present invention. The specific method includes:

[0054] Step S1: Based on the product trading platform, obtain the product sales volume of emerging products and users' historical purchased products at each trading moment, as well as the transaction times and trading moments of users corresponding to each historical purchased product.

[0055] In the embodiment of the present invention, since emerging products have no historical information and often appear in the market for the first time, resulting in their inability to participate in recommendations, which is called the cold start of products. To improve users' awareness and purchase intention of emerging products, it is necessary to analyze the association between emerging products and users' historical purchased products, establish the social relationship between products, and conduct personalized product recommendations for users according to users' behavioral preferences. Therefore, it is necessary to analyze the attributes of products and users' behavioral habits. First, through the product trading platform, obtain the product sales volume of emerging products and users' historical purchased products at each trading moment, as well as the transaction times and trading moments of users corresponding to each historical purchased product.

[0056] It should be noted that in the embodiment of the present invention, the trading moment interval is one day, that is, the product sales volume of the product every day is obtained as the product sales volume at each trading moment. In other embodiments of the present invention, the size of the trading moment interval can be specifically set according to specific circumstances, and no limitation and elaboration will be made here.

[0057] It should be noted that in an embodiment of the present invention, the historical purchase situation of users on the product trading platform in the past year at the current moment is obtained for analysis, and the emerging product is a product that appears on the product trading platform within the first half month before the current moment.

[0058] It should be noted that in the embodiments of the present invention, for the convenience of subsequent data processing and to avoid differences in units and numerical magnitudes between data, the data is standardized to eliminate the influence of dimensions in data operations.

[0059] Step S2: According to the product sales volume change trend of each product at different transaction times, obtain the explosiveness of each product; according to the explosiveness of the products purchased by each user in different histories and the transaction time characteristics of the corresponding historically purchased products, obtain the conformity of each user.

[0060] Considering that some users have shopping herd behavior, even if the emerging product has no high similarity with the historically purchased products of the user, but because its product is a best-selling product in the short term and is purchased by most people, the corresponding emerging product should also be recommended. It is necessary to analyze the product sales volume change trend at different transaction times to reflect the market dynamics of the product at different times. The more rapidly the product sales volume increases in the short term, the more likely it is to be a best-selling product. According to the product sales volume change trend of each product at different transaction times, obtain the explosiveness of each product.

[0061] Preferably, in an embodiment of the present invention, for the method of obtaining explosiveness, please refer to Figure 2 , which shows a flowchart of a method for obtaining explosiveness, including:

[0062] Step S201: For any product, select the transaction time with the largest product sales volume value as the best-selling time.

[0063] Since best-selling products have the characteristics of rapidly becoming popular and a sharp increase in product sales volume in the short term, selecting the transaction time with the largest product sales volume value as the best-selling time represents the peak of product sales volume and is the moment when the product is most popular, which is a key reference point for evaluating the explosiveness of the product.

[0064] Step S202: Obtain the difference between the product sales volume of the corresponding product and the product sales volume of other products at each transaction time within the neighborhood range of the best-selling time as the relative sales volume difference.

[0065] Within the neighborhood range of the best-selling time, the product sales volumes of different products will vary, and the relative sales volume difference can more accurately reflect the relative performance of the best-selling product.

[0066] It should be noted that, in an embodiment of the present invention, the method for obtaining the neighborhood range includes: taking the moment of popularity as a reference, obtaining the difference in product sales volume between each transaction moment and the next transaction moment on both sides of the moment of popularity in sequence, and calculating the ratio of the difference result to the product sales volume at the next transaction moment as the sales volume change rate for each transaction moment; if there is a transaction moment on either side of the moment of popularity where the sales volume change rate is less than that of the previous adjacent transaction moment, the corresponding transaction moment is taken as the cut-off moment; the time range between the cut-off moments on both sides constitutes the neighborhood range of the moment of popularity.

[0067] Step S203: Obtain the sum of the relative sales volume differences of all transaction moments within the neighborhood range of each product at the moment of popularity as the popularity of each product.

[0068] In an embodiment of the present invention, the formula for popularity is expressed as:

[0069] ;

[0070] Wherein, represents the popularity of the th product; represents the product sales volume of the th product at the th transaction moment; represents the product sales volume of the th other product at the th transaction moment; represents the number of other products at the corresponding transaction moment; represents the number of transaction moments within the domain range of the moment of popularity.

[0071] In the formula for popularity, represents the product sales volume of the th product and the th other product at the th transaction moment, that is, the relative sales volume difference; represents calculating the sum of the relative sales volume differences of all moments within the neighborhood range of each product at the moment of popularity, that is, the popularity. The greater the relative sales volume difference, the greater the product sales volume of the th product at the th transaction moment corresponding to other products, and the more likely it is to be a popular product, and the higher the popularity.

[0072] By comprehensively analyzing the popularity of different historical purchased products of users and the transaction time characteristics of the corresponding historical purchased products, the herd mentality of users in purchasing products can be quantified. If a user frequently purchases products with a relatively high popularity, it is considered that the user has a strong herd mentality. The transaction time characteristics of historical purchased products help to understand consumers' preferences for different types of products at different time periods. Based on the popularity of different historical purchased products of each user and the corresponding transaction time characteristics of the historical purchased products, the herd mentality of each user can be obtained.

[0073] Preferably, in an embodiment of the present invention, the method for obtaining the herd mentality includes:

[0074] For each historical purchased product of each user, obtain the time difference between each transaction moment within the neighborhood range of the corresponding transaction moment and the popularity moment, and select the one with the smallest time difference value, and use the corresponding time difference as the popularity time difference;

[0075] Based on the popularity of different historical purchased products of each user and the popularity time difference, obtain the herd mentality of each user. The popularity is positively correlated with the herd mentality, and the popularity time difference is negatively correlated with the herd mentality.

[0076] In an embodiment of the present invention, the formula for the herd mentality is expressed as:

[0077] ;

[0078] Wherein, represents the herd mentality of the rd user; represents the popularity of the th historical purchased product; represents the transaction moment of the th historical purchased product; represents the transaction moment within the neighborhood range of the popularity moment of the th historical purchased product; represents the number of historical purchased products of the th user; represents taking the minimum value; represents the logistic function.

[0079] In the formula for the herd mentality, the greater the popularity of the th historical purchased product, the greater the probability that the user tends to buy popular products, and the greater the herd mentality; represents the time difference between the transaction moment of the th historical purchased product and each transaction moment within the neighborhood range of the popularity moment. Take the one with the smallest time difference value, that is, the popularity time difference. The greater the popularity time difference, the farther the transaction moment of the historical purchased product is from the neighborhood range of the corresponding popularity moment, and the smaller the herd mentality.

[0080] Step S3: For any user as the target user, according to the similarity between each emerging product and different corresponding historical purchased products, obtain the product similarity between each emerging product and each historical purchased product, and screen out emerging recommended products.

[0081] The interests and needs of each user are unique, and the performance characteristics of different products at different levels are different. By analyzing the similarity between emerging products and the user's historical purchased products, the internal connection of the products can be deeply explored, the relevance between emerging products and user needs can be captured more accurately, so as to provide recommendations that meet the user's preferences. For any user as the target user, according to the similarity between each emerging product and different corresponding historical purchased products, obtain the product similarity between each emerging product and each historical purchased product.

[0082] Preferably, in an embodiment of the present invention, for the method of obtaining the product similarity, please refer to Figure 3 , which shows a flowchart of a method for obtaining product similarity, including:

[0083] Step S301: Obtain multi-dimensional information of the emerging product and the historical purchased products of the target user, and obtain word vectors corresponding to each dimension information.

[0084] It should be noted that obtaining the multi-dimensional information of the product based on the product trading platform, including product uses, parameters, prices, brands, etc., which reflect the attributes of the product, helps to understand the product; for the convenience of data processing, first use the natural language NLP algorithm to numerically process the multi-dimensional information, use the jieba algorithm to remove meaningless function words and prepositions from the multi-dimensional information of emerging products and the user's historical purchased products, decompose them into word groups, and then use the BERT algorithm model of context embedding to generate word vectors of the information, which can be converted into a numerical form that the model can process, helping to capture the correlation relationship between products; the specific means are well-known technical means to those skilled in the art and will not be elaborated here.

[0085] Step S302: Obtain the cosine similarity of the word vectors of each same dimension information between the emerging product and each historical purchased product as the dimension information similarity; obtain the cumulative value of the similarity of all dimension information between the emerging product and each historical purchased product as the product similarity between the emerging product and each historical purchased product.

[0086] Word vectors not only contain numerical values with directions. Similar information often has similar directions, while word vectors with different directions represent different meanings. Cosine similarity can capture the directional similarity between word vectors and analyze the correlation relationship of dimensional information. The greater the cosine similarity of the word vectors of dimensional information, the greater the similarity between products.

[0087] In an embodiment of the present invention, for a target user, the formula for product similarity is expressed as:

[0088] ;

[0089] Wherein, represents the product similarity between the emerging product and the th historical purchased product; represents the word vector of the th dimensional information of the emerging product ; represents the word vector of the th dimensional information of the th historical purchased product; represents the number of dimensional information in the product; represents the number of historical purchased products of the user; represents the cosine similarity function.

[0090] In the formula for product similarity, represents calculating the cosine similarity of the word vectors of each same dimensional information between the emerging product and each historical purchased product, that is, the dimensional information similarity, and calculating the cumulative value of all dimensional information similarities, that is, the product similarity between the emerging product and each historical purchased product. The greater the similarity of the word vectors of all dimensional information, the greater the product similarity.

[0091] It should be noted that in other embodiments of the present invention, the Pearson correlation coefficient algorithm may also be used to obtain the similarity of the word vectors of dimensional information between products. The specific means are well-known technical means to those skilled in the art and will not be elaborated here.

[0092] The similarity between the emerging product and the historical purchased product can more accurately grasp the user's purchase preferences and needs, and can efficiently screen out emerging products similar to the user's historical purchase records, screen the emerging recommended products for each user, and improve the accuracy and efficiency of the recommendation.

[0093] Preferably, in an embodiment of the present invention, the method for obtaining emerging recommended products includes:

[0094] For the target user, accumulate the product similarities between the emerging products and all historical purchased products, and perform normalization, which is used as the overall product similarity between the emerging products and all historical purchased products; obtain the mean value of the overall product similarities corresponding to all emerging products as the average overall product similarity;

[0095] If the overall product similarity corresponding to the emerging product is greater than the average overall product similarity, the corresponding emerging product is used as an emerging recommended product.

[0096] Step S4: Obtain the transaction performance of the user corresponding to each historical purchased product according to the transaction times and transaction time distributions corresponding to different historical purchased products of the user; obtain multiple high-similar product clusters according to the transaction performances corresponding to different user-purchased products of the target user; obtain the interest tendency of the target user corresponding to each historical purchased product according to the transaction feature distributions between different users corresponding to each historical purchased product within the high-similar product cluster.

[0097] By analyzing the user's transaction times and transaction times, it helps to understand the changes in the user's purchase behavior in different time periods, and can reveal the user's purchase habits, preferences and activity levels, as well as whether the user shows a continuous purchase interest in certain products. Therefore, according to the transaction times and transaction time distributions corresponding to different historical purchased products of the user, obtain the transaction performance of the user corresponding to each historical purchased product.

[0098] Preferably, in an embodiment of the present invention, the method for obtaining the transaction performance includes:

[0099] Obtain the ratio of the transaction times of the user corresponding to each historical purchased product to the sum of the transaction times of all historical purchased products as the first performance coefficient; obtain the mean value of the differences in transaction times between the same historical purchased products and perform a negative correlation mapping as the second performance coefficient;

[0100] Calculate the product of the first performance coefficient and the second performance coefficient corresponding to each historical purchased product as the transaction performance of the user corresponding to each historical purchased product. In an embodiment of the present invention, the formula for the transaction performance is expressed as:

[0101] ;

[0102] Among them, represents the transaction performance of the th historical purchased product of the th user; represents the transaction times of the th historical purchased product of the th user; represents the sum of the transaction times of all historical purchased products of the th user; Represents the th user's mean difference in transaction times between the

[0103] In the formula for transaction performance, represents the ratio of the number of transactions of the th user's th historical purchased product to the sum of the number of transactions of all historical purchased products, that is, the first performance coefficient. The larger the ratio, the th user's th historical purchased product has a larger number of transactions. For the user, there are more historical purchased products for transactions, and the transaction performance is more obvious; represents the normalization of the mean difference in transaction times between the th user's th historical purchased product and the same historical purchased product, that is, the second performance coefficient. The larger the second performance coefficient, the greater the difference in transaction times between the same historical purchased products, and the smaller the purchase performance of the corresponding historical purchased products.

[0104] It should be noted that in an embodiment of the present invention, taking the reciprocal for negative correlation mapping constitutes a negative correlation relationship where the greater the mean difference in transaction times, the smaller the second performance coefficient, and the smaller the purchase performance. In other embodiments of the present invention, a function can also be used for negative correlation mapping. The specific means are well-known technical means to those skilled in the art and will not be elaborated here.

[0105] When a user purchases a product, they often exhibit certain behavior patterns or preferences. By analyzing the transaction performance of the target user for different users' purchased products, the behavior patterns of the user's product purchases can be identified. Classifying products with similar characteristics into one category helps to screen out products that meet the user's preferences for targeted analysis. Based on the transaction performance of the target user corresponding to different users' purchased products, multiple highly similar product clusters can be obtained.

[0106] Preferably, in an embodiment of the present invention, the method for obtaining highly similar product clusters includes:

[0107] Performing ISODATA clustering on all historical purchased products according to the transaction performance of the target user corresponding to different historical purchased products to obtain product clustering clusters;

[0108] Obtaining the mean value of the transaction performance of the historical purchased products within each product clustering cluster as the local transaction performance; obtaining the mean value of the local transaction performances of all product clustering clusters as the overall transaction performance;

[0109] If the local transaction performance of the product clustering cluster is greater than the overall transaction performance, the corresponding product clustering cluster is used as the highly similar product cluster of the target user.

[0110] It should be noted that ISODATA clustering is a well-known technical means in the art and will not be elaborated here.

[0111] The transaction characteristics of users can reflect users' purchase habits, preferences, and interests. In the highly similar product cluster, there is often a certain correlation between products. By analyzing the transaction characteristics of products among different users, the interest degree of the target user in different products can be inferred; according to the distribution of transaction characteristics among different users corresponding to each historical purchased product in the highly similar product cluster, the interest tendency of the target user corresponding to each historical purchased product is obtained.

[0112] Preferably, in an embodiment of the present invention, the method for obtaining the interest tendency includes:

[0113] According to the transaction performance, transaction times, and product sales volume characteristics of each historical purchased product corresponding to different users, the purchase tendency of the target user corresponding to each historical purchased product is obtained;

[0114] Preferably, in an embodiment of the present invention, for the method of obtaining the purchase tendency, please refer to Figure 4 , which shows a flowchart of a method for obtaining the purchase tendency, including:

[0115] Step S401: For any historical purchased product, obtain the ratio of the transaction times of the user corresponding to the historical purchased product to the sum of the product sales volume of the historical purchased product at all transaction times as the user purchase ratio.

[0116] By analyzing the ratio, the purchase preference of the user for the historical purchased product is reflected, and by comparing the transaction times of the user with the total sales volume of the product, the relative preference degree of the user for the product is quantified.

[0117] Step S402: Integrate the transaction performance of the user corresponding to the historical purchased product and the user purchase ratio to obtain the purchase influence index of the user corresponding to the historical purchased product; obtain the mean value of the purchase influence indexes of all other users corresponding to the historical purchased product as the average purchase influence index.

[0118] It should be noted that in some embodiments of the present invention, the transaction performance and the user purchase ratio can be integrated by addition or multiplication. The specific means are well-known technical means in the art and will not be elaborated here.

[0119] The transaction performance reflects the purchasing power and activity of users, and the user purchase proportion reflects the relative preference of users for the product. By combining the analysis, we can more comprehensively understand the purchase influence index of users, which is used to evaluate the purchase influence of users on specific products.

[0120] By quantifying the purchase influence indicators of all other users' corresponding historical purchased products through the method of taking the average, the average purchase influence index is obtained, which is used to compare the purchase influence levels of different user groups on historical purchased products.

[0121] Step S403: According to the deviation between the purchase influence index of the target user's corresponding historical purchased products and the average purchase influence index, obtain the purchase tendency of the target user for the historical purchased products.

[0122] It should be noted that in an embodiment of the present invention, the formula for the purchase tendency is expressed as:

[0123] ;

[0124] Wherein, represents the purchase tendency of the th historical purchased product of the target user; represents the transaction performance of the th historical purchased product of the target user; represents the number of transactions of the th historical purchased product of the target user; represents the sum of the product sales of the historical purchased product at all transaction times; represents the th number of other users corresponding to the historical purchased product; represents the number of transactions of the th historical purchased product of the th other user; represents the transaction performance of the th historical purchased product of the

[0125] In the formula for the purchase tendency, represents the ratio of the number of transactions of the th historical purchased product of the target user to the sum of the product sales of the historical purchased product at all transaction times, that is, the user purchase proportion. The larger the user purchase proportion, the more important the historical purchased product is to the user; represents the th The product of the transaction performance of a historical purchased product and the user purchase proportion, that is, the purchase influence index. The greater the purchase influence index, the greater the transaction performance, the greater the user purchase proportion, and the greater the purchase tendency of the target user for the corresponding historical purchased product; It represents calculating the mean of the purchase influence indices of all other users for their corresponding historical purchased products, that is, the average purchase influence index. It represents the target user The deviation between the purchase influence index of the

[0126] For highly similar product clusters, obtain the difference between the transaction moments of each historical purchased product corresponding to each user and those of each other historical purchased product, and perform a negative correlation mapping as the transaction activity; accumulate the transaction activities corresponding to all other purchased products as the transaction frequency;

[0127] Obtain the product of the first performance coefficient corresponding to each historical purchased product for each user and the transaction frequency as the transaction interest; based on the deviation of the transaction interest between the target user and each other user for each historical purchased product, and the purchase tendency of the target user, obtain the interest tendency of the target user for each historical purchased product. Both the deviation of the transaction interest and the purchase tendency are positively correlated with the interest tendency.

[0128] In an embodiment of the present invention, for any historical purchased product in a highly similar product cluster, the formula for the interest tendency is expressed as:

[0129] ;

[0130] Wherein, represents the interest tendency of the th historical purchased product of the target user; represents the number of other users corresponding to the th historical purchased product; represents the th purchase tendency of the th historical purchased product of the target user; represents the number of other historical purchased products within the highly similar product cluster corresponding to the th historical purchased product; Indicates the th historical purchased product corresponding to the th transaction moment of other users and the th transaction moment of other historical purchased products; Indicates the th transaction times of the historical purchased product of the target user; Indicates the sum of the transaction times of all historical purchased products of the target user ; Indicates the th transaction times of the th historical purchased product of other users; Indicates the sum of the transaction times of all historical purchased products of the th other user;

[0131] In the formula of interest tendency, indicates the ratio of the transaction times of the th historical purchased product of the target user to the sum of the transaction times of all historical purchased products, that is, the first performance coefficient of the target user; ; Indicates the negative correlation mapping of the difference between the transaction moment of the th historical purchased product corresponding to the target user and the transaction moment of the th other historical purchased product, that is, the transaction activity. The smaller the difference, the closer the transaction moments, and the greater the transaction activity; Indicates calculating the sum of the transaction activities corresponding to all other purchased products as the transaction frequency, Indicates calculating the product of the first performance coefficient and the transaction frequency as the transaction interest degree, that is, the greater the first performance coefficient, the greater the transaction activity, the corresponding greater the transaction frequency, and the more likely there is a greater demand situation, and the greater the corresponding transaction interest degree; Indicates calculating the deviation of the transaction interest degree of the target user and each other user for each historical purchased product. The greater the deviation, the greater the transaction interest degree of the historical purchased product corresponding to the target user relative to other users, the greater the interest tendency for the historical purchased product, the greater the purchase tendency, the greater the interest tendency, and the more likely it is a fast-moving consumer product.

[0132] Step S5: Obtain the recommendation index of the emerging recommended product corresponding to the target user according to the popularity of the emerging recommended product, the conformity corresponding to the target user, the interest tendency, and the product similarity.

[0133] Products with strong popularity can attract more user attention, improve the exposure rate and popularity of the products; the herd mentality of target users makes it easier for users to be influenced by popular products or popular recommendations, thus increasing the willingness to purchase; interest inclination helps to understand the degree of preference of users for specific types of products; product similarity reflects the similarity between emerging products and historical purchased products. The higher the similarity, the easier it is to be accepted and recognized by users. Therefore, according to the popularity of emerging recommended products, the herd mentality corresponding to target users, interest inclination, and product similarity, the recommendation index of emerging recommended products corresponding to target users is obtained.

[0134] Preferably, in an embodiment of the present invention, the method for obtaining the recommendation index includes:

[0135] Select the one with the largest product similarity value between each historical purchased product in the high-similar product cluster of the emerging recommended product and the target user, use the corresponding product similarity as the first recommendation coefficient, and use the interest inclination of the corresponding historical purchased product as the second recommendation coefficient;

[0136] According to the popularity of the emerging recommended product, the herd mentality corresponding to the target user, the first recommendation coefficient, and the second recommendation coefficient, obtain the recommendation index of the emerging recommended product corresponding to the target user. The popularity, herd mentality, first recommendation coefficient, and second recommendation coefficient are all positively correlated with the recommendation index.

[0137] Among them, the greater the popularity, the more the emerging recommended product is purchased by most users in the short term. The greater the herd mentality of users, the more frequently they purchase products with popularity. The greater the first recommendation coefficient and the second recommendation coefficient, it means that the similarity between the emerging recommended product and the historical purchased product is greater, and the historical purchased product meets the shopping preferences of users. Therefore, the emerging recommended product needs to be recommended more.

[0138] In an embodiment of the present invention, the formula of the recommendation index is expressed as:

[0139] ;

[0140] Among them, represents the recommendation index of the new recommended product corresponding to the target user ; represents the herd mentality of the target user ; represents the popularity of the new recommended product ; represents the new recommended product and the target user in the high-similar product cluster the product similarity between the represents the maximum value function; Indicates the selection of emerging recommended products with the target user among different historical purchased products within the highly similar product cluster The one with the largest product similarity value between them is the first recommendation coefficient; Indicates the interest tendency of the historical purchased product corresponding to the target user when the product similarity value is the largest, that is, the second recommendation coefficient; Indicates the logistic function. Indicates the logistic function.

[0141] In the formula of the recommendation index, when the emerging product is a blockbuster product, the stronger the herd mentality of the user, the higher the blockbuster nature of the new recommended product, then the larger it is, the greater the possibility of being recommended; the greater the product similarity, the greater the first recommendation coefficient, the higher the interest tendency of the historical purchased product, the greater the second recommendation coefficient, and the more likely they are products of the same category, and the greater the recommendation index of the emerging recommended product. The larger it is, the greater the possibility of being recommended; the greater the product similarity, the greater the first recommendation coefficient, the higher the interest tendency of the historical purchased product, the greater the second recommendation coefficient, and the more likely they are products of the same category, and the greater the recommendation index of the emerging recommended product.

[0142] Based on this, for each user as the target user, by obtaining the recommendation index of the emerging recommended product for each user, the overall product similarity between the emerging product and the historical purchased product can be weighted and adjusted, so that the emerging products with higher recommendation indices in the subsequent collaborative filtering algorithm occupy more important positions in the recommendation list, increasing the exposure of the emerging products and helping to accurately recommend products to users. The specific collaborative filtering algorithm is a well-known technical means to those skilled in the art and will not be elaborated here.

[0143] In summary, the present invention obtains the blockbuster nature of each product according to the product sales volume change trend of each product at different transaction times; combines the transaction time characteristics of the corresponding historical purchased products to obtain the herd mentality of each user; for any user as the target user, according to the similarity between each emerging product and different historical purchased products, obtains the product similarity between the emerging product and each historical purchased product, and screens out the emerging recommended products; obtains the transaction performance of the user corresponding to each historical purchased product according to the transaction times and transaction time distributions of different historical purchased products; and then obtains the interest tendency of the target user corresponding to the historical purchased products; obtains the recommendation index of the emerging recommended product corresponding to the target user. The present invention obtains the accurate recommendation index of each emerging product for the user, improving the accuracy of product recommendation.

[0144] It should be noted that: the above sequence of embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0145] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the differences between each embodiment and other embodiments are emphasized.

Claims

1. A product intelligent recommendation method based on collaborative filtering algorithm, characterized in that: The method comprises: Based on the product trading platform, obtain the product sales of emerging products and users' historically purchased products at each transaction time, as well as the number of transactions and transaction time of each historically purchased product by the user; According to the sales trend of each product at different transaction times, the popularity of each product is obtained; according to the popularity of each user's different historical purchase products and the transaction time characteristics of the corresponding historical purchase products, the conformity of each user is obtained; For any user as the target user, the product similarity between each emerging product and each corresponding historically purchased product is obtained based on the similarity between each emerging product and each corresponding historically purchased product, and the emerging recommended products are screened out; According to the transaction times and transaction time distribution of different historically purchased products by the user, the transaction performance of each historically purchased product by the user is obtained; according to the transaction performance of the target user corresponding to the products purchased by different users, multiple highly similar product clusters are obtained; according to the transaction feature distribution between different users for each historically purchased product in the highly similar product cluster, the interest tendency of the target user corresponding to each historically purchased product is obtained; According to the popularity of the emerging recommended products, the conformity, interest tendency and product similarity of the target users, the recommendation index of the emerging recommended products corresponding to the target users is obtained; The method for obtaining the recommendation index includes: Select the product with the largest similarity value between the emerging recommended product and each historically purchased product in the high similarity product cluster of the target user, and use the corresponding product similarity as the first recommendation coefficient, and use the interest tendency of the corresponding historically purchased product as the second recommendation coefficient; According to the popularity of the emerging recommended products, the corresponding conformity of the target users, the first recommendation coefficient and the second recommendation coefficient, the recommendation index of the emerging recommended products corresponding to the target users is obtained. The popularity, conformity, the first recommendation coefficient and the second recommendation coefficient are all positively correlated with the recommendation index.

2. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 1, characterized in that: The method for obtaining the explosive product includes: For any product, select the transaction moment with the largest sales value as the hot selling moment; Obtain the difference between the product sales of the corresponding product and the product sales of other products at each transaction moment in the neighborhood of the hot-selling moment as the relative sales difference; The cumulative sum of the relative sales volume differences of each product at all transaction moments within the neighborhood of the hot-selling moment is obtained as the hot-selling nature of each product.

3. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 2 is characterized in that: The method for obtaining the conformity includes: For each historically purchased product of each user, obtain the time difference between each transaction time in the neighborhood of the corresponding transaction time and the hot product time, select the one with the smallest time difference value, and use the corresponding time difference as the hot product time difference; According to the popularity and time difference of each user's historical product purchases, the conformity of each user is obtained. The popularity is positively correlated with the conformity, while the time difference of the popularity is negatively correlated with the conformity.

4. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 1, characterized in that: The method for obtaining the product similarity includes: Obtain multi-dimensional information about emerging products and historically purchased products of target users, and obtain word vectors corresponding to each dimensional information; The cosine similarity of the word vectors of each identical dimensional information between the emerging product and each historically purchased product is obtained as the dimensional information similarity; the cumulative value of the similarities of all dimensional information between the emerging product and each historically purchased product is obtained as the product similarity between the emerging product and each historically purchased product.

5. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 1 is characterized in that: The method for obtaining the emerging recommended products includes: For the target user, the product similarities between the emerging product and all the historically purchased products are accumulated and normalized to obtain the overall product similarity between the emerging product and all the historically purchased products; Obtain the average of the overall similarities of all emerging products to the product as the average overall product similarity; If the overall similarity of an emerging product to the aforementioned product is greater than the average overall similarity of the products, the corresponding emerging product will be regarded as an emerging recommended product.

6. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 1 is characterized in that: The method for obtaining the transaction performance includes: The ratio of the number of transactions of each historically purchased product to the sum of the number of transactions of all historically purchased products is obtained as the first performance coefficient; the mean difference of transaction times between the same historically purchased products is obtained and negatively correlated mapping is performed as the second performance coefficient; The product of the first performance coefficient and the second performance coefficient corresponding to each historically purchased product is calculated as the transaction performance of the user corresponding to each historically purchased product.

7. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 6 is characterized in that: The method for obtaining the highly similar product cluster includes: According to the transaction performance of target users corresponding to different historically purchased products, ISODATA clustering is performed on all historically purchased products to obtain product clustering clusters; Obtain the mean of the transaction performance of historically purchased products in each product cluster as the local transaction performance; obtain the mean of the local transaction performance of all product clusters as the overall transaction performance; If the local transaction performance of a product cluster is greater than the overall transaction performance, the corresponding product cluster will be used as a high-similar product cluster for the target user.

8. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 7 is characterized in that: The method for obtaining the interest tendency includes: According to the transaction performance, transaction number and product sales characteristics of each historically purchased product between different users, the purchase tendency of the target user for each historically purchased product is obtained; For highly similar product clusters, obtain the difference between the transaction time of each user for each historically purchased product and the transaction time of each other historically purchased product, and perform negative correlation mapping as the transaction activity; accumulate the transaction activities corresponding to all other purchased products as the transaction frequency; The product of the first performance coefficient and the transaction frequency of each user for each historically purchased product is obtained as the transaction interest. According to the deviation of the transaction interest between the target user and each other user for each historically purchased product, and the purchase tendency of the target user, the interest tendency of the target user for each historically purchased product is obtained, and both the deviation of the transaction interest and the purchase tendency are positively correlated with the interest tendency.

9. The product intelligent recommendation method based on collaborative filtering algorithm according to claim 8, characterized in that: The method for obtaining the purchase tendency includes: For any historically purchased product, obtain the ratio of the number of transactions of the user corresponding to the historically purchased product to the sum of the product sales of the historically purchased product at all transaction times as the user's purchase ratio; The product of the transaction performance of the user's corresponding historical purchase product and the user's purchase proportion is obtained as the purchase influence index of the user's corresponding historical purchase product; the average of the purchase influence indexes of all other users' corresponding historical purchase products is obtained as the average purchase influence index; According to the deviation between the purchase influence index of the target user's corresponding historical purchase products and the average purchase influence index, the purchase tendency of the target user for the historical purchase products is obtained.

Images