CN112418978B - Product recommendation method, device, equipment and medium - Google Patents
Product recommendation method, device, equipment and medium Download PDFInfo
- Publication number
- CN112418978B CN112418978B CN202011284211.4A CN202011284211A CN112418978B CN 112418978 B CN112418978 B CN 112418978B CN 202011284211 A CN202011284211 A CN 202011284211A CN 112418978 B CN112418978 B CN 112418978B
- Authority
- CN
- China
- Prior art keywords
- client
- target
- recommendation
- recommended
- product
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000001514 detection method Methods 0.000 claims abstract description 33
- 238000005065 mining Methods 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims description 39
- 238000004422 calculation algorithm Methods 0.000 claims description 34
- 230000002159 abnormal effect Effects 0.000 claims description 23
- 238000004590 computer program Methods 0.000 claims description 17
- 238000012217 deletion Methods 0.000 claims description 17
- 230000037430 deletion Effects 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 13
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000013138 pruning Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 9
- 238000004458 analytical method Methods 0.000 description 5
- 238000007418 data mining Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000010354 integration Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012552 review Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Game Theory and Decision Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to the field of big data technologies, and in particular, to a product recommendation method, device, equipment, and medium. The product recommending method comprises the steps of obtaining client characteristic data of clients to be recommended; invoking a pre-trained target detection model to perform target detection on the client characteristic data, and determining whether the client to be recommended is a target recommendation client or not; if the client to be recommended is the target recommended client, acquiring historical transaction data of the target recommended client in a preset time period; performing association rule mining on the historical transaction data to obtain a target recommendation rule; wherein the target recommendation rule corresponds to a target product set; and recommending the target product set to the target recommendation client. The product recommendation method can effectively ensure the effectiveness and accuracy of recommendation.
Description
Technical Field
The present invention relates to the field of big data technologies, and in particular, to a product recommendation method, device, equipment, and medium.
Background
With the development of the internet and informatization technologies, various business fields are continuously electronic, for example, the online insurance business fields can be applied, online claims are settled, online product recommendation and the like.
In internet technology, websites often need to recommend various product information to users, such as e-commerce websites recommending goods on web pages that may be of interest to users, etc. By the recommendation mode, the path of a user for searching for a required product is shortened, and user experience is improved.
When recommending products, the website mainly analyzes the preference of the user according to the historical transaction data of the user on certain products, such as the product purchase historical data of the user, so as to recommend the products to all the clients, and does not analyze whether the clients are potential clients, so that the success rate of the recommendation mode is not high, and the effectiveness of the recommendation cannot be guaranteed.
Disclosure of Invention
The embodiment of the invention provides a product recommendation method, device, equipment and medium, which are used for solving the problem that the effectiveness of recommendation cannot be ensured in the traditional recommendation mode.
A product recommendation method comprising:
acquiring client characteristic data of a client to be recommended;
invoking a pre-trained target detection model to perform target detection on the client characteristic data, and determining whether the client to be recommended is a target recommendation client or not;
if the client to be recommended is the target recommended client, acquiring historical transaction data of the target recommended client in a preset time period;
Performing association rule mining on the historical transaction data to obtain a target recommendation rule; wherein the target recommendation rule corresponds to a target product set;
and recommending the target product set to the target recommendation client.
A product recommendation device, comprising:
the client characteristic data acquisition module is used for acquiring client characteristic data of a client to be recommended;
the target detection module is used for calling a pre-trained target detection model to carry out target detection on the client characteristic data and determining whether the client to be recommended is a target recommendation client or not;
the first recommending module is used for acquiring historical transaction data of the target recommending client in a preset time period if the client to be recommended is the target recommending client;
the rule mining module is used for carrying out association rule mining on the historical transaction data to obtain a target recommendation rule; wherein the target recommendation rule corresponds to a target product set;
and the product recommending module is used for recommending the target product set to the target recommending client. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the product recommendation method described above when the computer program is executed.
A computer storage medium storing a computer program which, when executed by a processor, implements the steps of the product recommendation method described above.
According to the product recommending method, device, equipment and medium, the client characteristic data of the client to be recommended is obtained so as to call the pre-trained target detection model to carry out target detection on the client characteristic data, whether the client to be recommended is the target recommending client is determined, and before the product is recommended to the client to be recommended, the target recommending group required to be recommended is determined firstly, so that unnecessary recommendation is avoided, and the success rate of recommendation can be improved. If the client to be recommended is the target recommending client, historical transaction data of the target recommending client in a preset time period is obtained, association rule mining is conducted on the historical transaction data, target recommending rules are obtained rapidly, then a target product set corresponding to the target recommending rules is recommended to the target recommending client, and product recommending efficiency is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic view of an application environment of a product recommendation method according to an embodiment of the invention;
FIG. 2 is a flow chart of a product recommendation method according to an embodiment of the invention;
FIG. 3 is a flow chart of a product recommendation method according to an embodiment of the invention;
FIG. 4 is a flow chart of a product recommendation method according to an embodiment of the invention;
FIG. 5 is a flowchart showing step S402 in FIG. 4;
FIG. 6 is a flowchart showing step S501 in FIG. 5;
FIG. 7 is a flowchart showing step S502 in FIG. 5;
FIG. 8 is a flowchart showing step S204 in FIG. 2;
FIG. 9 is a flowchart showing step S204 in FIG. 2;
FIG. 10 is a schematic diagram of a product recommendation device according to an embodiment of the invention;
FIG. 11 is a schematic diagram of a computer device in accordance with an embodiment of the invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The product recommendation method is applicable in an application environment as in fig. 1, wherein a computer device communicates with a server via a network. The computer devices may be, but are not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices. The server may be implemented as a stand-alone server.
In one embodiment, as shown in fig. 2, a product recommendation method is provided, and the method is applied to the server in fig. 1, and the method includes the following steps:
s201: and acquiring client characteristic data of the clients to be recommended.
The method can be applied to a product recommendation system, unnecessary recommendation is avoided by determining target groups to be recommended first, and the success rate of subsequent recommendation can be improved. In addition, the recommendation accuracy is further ensured by pertinently adopting different recommendation algorithms to determine the target product set for the new customer and the old customer. It should be noted that the present application may be applied to product recommendation in different fields, and the following description will take vehicle insurance product recommendation as an example.
Wherein the customer characteristic data may be obtained from data recorded by the insurance platform or a third party data source. The customer characteristic data includes, but is not limited to, customer characteristics and vehicle characteristics. The customer characteristics include, but are not limited to, age, gender, occupation, annual income, family status, marital status, premium amount, historical claims rate, product review records, underwrited insurance products, location territories, driving age, and the like. The vehicle characteristics include, but are not limited to, selling price, model, brand model, age, condition of the vehicle, driving zone, etc. The clients to be recommended refer to all clients in the platform.
S202: and calling a pre-trained target detection model to perform target detection on the client characteristic data, and determining whether the client to be recommended is a target recommended client or not.
The target recommended clients are clients which finally need to be recommended. Specifically, before a product recommendation is made to a to-be-recommended client, it is necessary to further determine whether the to-be-recommended client is a potential client, that is, a target recommended client, in order to reduce unnecessary recommendation. Specifically, the client characteristic data is input into the target detection model as input data by calling the pre-trained target detection model to perform target detection, so that whether the client to be recommended is a target recommendation client is determined.
S203: and if the client to be recommended is the target recommended client, acquiring historical transaction data of the target recommended client in a preset time period.
Wherein the preset time period is a preset transaction time period for acquiring historical transaction data, for example, the past year. Specifically, if the client to be recommended is the target recommended client, the client is considered to purchase a certain product before, so that the effectiveness and accuracy of the preference analysis can be ensured when the subsequent analysis of the user product preference is performed based on the historical transaction data, in other words, the subsequent historical transaction data analysis has high availability, otherwise, for the client who does not purchase the product before, the availability and accuracy of the analysis result obtained by analyzing the user preference by adopting the historical transaction data of the client are lower, and the recommendation success rate is not high. In this embodiment, after determining that the to-be-recommended client is the target recommended client, data mining is required according to historical transaction data corresponding to the target recommended client, so as to determine a product to be recommended.
S204: performing association rule mining on historical transaction data to obtain target recommendation rules; wherein the target recommendation rule corresponds to a target product set.
Wherein the target product set includes at least one product with high user purchase availability. For example, the target recommendation is (I2- > I1, I3), where I1, I2, I3 are used to represent different products, and the product set corresponding to the target recommendation rule is (I1, I3). Association rule mining of historical transaction data in this embodiment may be implemented using algorithms including, but not limited to, apriori (Association rule mining ), which are not limited herein.
The implementation step of mining by using the Apriori algorithm is as follows, assuming that the first historical transaction data is (I1, I2, I3), the second historical transaction data is (I1, I2), the two historical transaction data are subjected to data mining, that is, a K value is initialized to 1, then a candidate K item set (candidate 1 item set) is obtained according to the two historical transaction data and is { I1}, { I2}, { I3}, the corresponding support degree of the candidate 1 item set (that is, the ratio of the number of occurrences of each 1 item set in the historical transaction data to the total data amount of the historical transaction data) is 2,1 item sets with the support degree smaller than the minimum support degree are removed, that is { I3}, a frequent 1 item set is obtained, that is { I1, I2}, then it is judged whether the number of item sets contained in the frequent K item set is larger than K (K is 1), if the number is larger than k=2, then the number of item sets is larger than k=2, and the algorithm is continuously carried out according to the number of item sets that the number of the frequent item sets is larger than K2, and the number of item sets is not contained in the frequent item sets is obtained, and the algorithm is continuously carried out. The natural connection theorem is that when the first K-1 items in two K item sets are identical, natural connection can be performed. The pruning algorithm is an algorithm for obtaining frequent sets by judging whether the support degree of each set in frequent K items is greater than a preset support degree. According to the above example, when the number of frequent 1 item sets is 1 and is not greater than K, the algorithm is terminated, the acquisition target frequent set is { I1, I2}, the corresponding subset thereof includes { I1}, { I2}, and the candidate rule may be determined by the subset, that is, (I1- > I2), (I2- > I1). And then calculating the confidence coefficient corresponding to each candidate rule through a confidence coefficient formula, and taking a subset with the confidence coefficient larger than a preset confidence coefficient threshold value as a target recommendation rule. Wherein, the confidence coefficient calculation formula is C (X- > Y) =σ (X u Y)/σ (X), wherein σ (X u Y) represents the number of times that X and Y occur in the historical transaction data at the same time, σ (X) represents the number of times that X occurs in the historical transaction data, in this example, C (I1- > I2) =2/2=1, C (I2- > I1) =2/2=1, assuming that the preset confidence coefficient threshold is 0.5, the target recommendation rules are (I1- > I2) and (I2- > I1), and the target product set is { I1, I2}.
S205: and recommending the target product set to a target recommending client.
Specifically, after the target product set corresponding to the target recommendation rule is determined, the target product set can be recommended to a target recommendation client, so that accurate product recommendation is realized.
In this embodiment, the client feature data of the client to be recommended is obtained so as to invoke the pre-trained target detection model to perform target detection on the client feature data, so as to determine whether the client to be recommended is a target recommendation client, so that the target recommendation group to be recommended is determined before the product is recommended to the client to be recommended, unnecessary recommendation is avoided, and the success rate of recommendation can be improved. If the client to be recommended is the target recommending client, historical transaction data of the target recommending client in a preset time period is obtained, association rule mining is conducted on the historical transaction data, target recommending rules are obtained rapidly, then a target product set corresponding to the target recommending rules is recommended to the target recommending client, and product recommending efficiency is improved.
In one embodiment, as shown in fig. 3, after step 202, the method further includes the following steps:
s301: acquiring a target recommendation client set; wherein the target recommendation client set comprises a plurality of target recommendation clients.
S302: and judging whether each target recommended client is a new client or not.
S303: if so, comparing the characteristic similarity between the client characteristic data of the new client and the client characteristic data of the non-new client in the target recommended client set, and determining similar clients of the new client.
S304: historical transaction data of similar clients is obtained, and a target product set is determined based on the historical data.
S305: and recommending the target product set to a new client.
Wherein the non-recommended clients set includes a plurality of non-target recommended clients. Specifically, since the new customer does not have historical transaction data, the target recommended customers in the target recommended customer set can be classified into new customers and non-new customers (i.e., old customers) according to whether the products have been purchased, and the old customers with the feature similarity greater than the similarity threshold can be used as similar customers of the new customers by comparing the feature similarity of the customer feature data of the new customers with the feature similarity of the old customers, and then the products purchased by the old customers or purchased more times can be recommended to the new customers as target product sets.
In this embodiment, the feature similarity between the new customer's customer feature data a and the old customer's customer feature data B may be calculated by using cosine similarity, or other ways of calculating feature similarity may be used to calculate the feature similarity between a and B, which is not limited herein.
In the embodiment, the target product set is determined and recommended by adopting different recommendation algorithms for the new customer and the old customer in a targeted manner, so that the recommendation accuracy is further ensured.
In one embodiment, as shown in fig. 4, before step S201, the method further includes the following steps:
s401: acquiring a customer portrait set; wherein the customer representation set includes training sample data for a plurality of platform customers.
The client image set is a feature matrix constructed for client feature data, for example, a feature factor is used as a column field, and each client to be recommended is used as a row field. Wherein the training sample data may be obtained from data recorded by the insurance platform or a third party data source. The training sample data includes, but is not limited to, customer characteristics, vehicle characteristics, and corresponding identification of whether or not it is a potential customer. The customer characteristics include, but are not limited to, age, gender, occupation, annual income, family status, marital status, premium amount, historical claims rate, product review records, underwrited insurance products, location territories, driving age, and the like. The vehicle characteristics include, but are not limited to, selling price, model, brand model, age, condition of the vehicle, driving zone, etc.
Specifically, if a customer purchases a certain vehicle insurance product several times in the past year, the customer is marked as a potential customer, such as 1; and customers that did not purchase or purchased only once in the past year are marked as non-potential customers, such as 0.
S402: and preprocessing the client image set to obtain a target image set.
The preprocessing of the feature matrix corresponding to the customer portrait set includes but is not limited to outlier processing, missing value processing, feature selection and the like.
Specifically, since there may be an abnormal value or a missing value in the customer image set, it is necessary to process the abnormal value and the missing value in the customer image set in order to ensure the accuracy of the subsequent group detection. In addition, because the characteristics in the customer portrait set are more, redundant characteristics or irrelevant characteristics possibly exist, the characteristics are selected to delete the characteristics, so that the number of the characteristics is reduced, the accuracy of a subsequent training model is improved, and the purpose of reducing the running time is achieved.
S403: and training the target image set by adopting an XGboost algorithm to obtain a target detection model.
Wherein XGboost is one of the integration algorithms, the main idea of which is to integrate a plurality of weak learners together to form one strong classifier. The algorithm uses CART regression trees as weak learners for integration. The expression obtained by integrating K CART regression trees by the XGboost algorithm is Wherein (1)>Representing the predicted value of the ith sample, and t represents the t-th base model, namely a CART regression tree; the corresponding objective function of the algorithm is: />Wherein Ω (f) t ) For controlling the complexity of the t-th class tree, Ω (·) represents a penalty term, ++>Representing the loss function, cross entropy may be used in this embodiment, where N represents the number of samples, i represents the sample identity, and K represents the number of base models. Specifically, residual error transfer training is continuously carried out through the first t base models through a greedy algorithm to obtain the (t+1) th base model until K base models are obtained, and the algorithm is terminated, wherein the algorithm is terminated, and the (t) th base model is selected according to the (t+1) th base model, wherein the (t+1) th base model is selected according to the greedy algorithm>And solving the function to obtain the t+1th base model.
In one embodiment, as shown in fig. 5, in step S402, a target image set is obtained by preprocessing a client image set, which specifically includes the following steps:
s501: and performing outlier processing on the client portrait set to obtain a first effective portrait set.
The outlier processing may determine the outlier in the client image set by means of statistical analysis or box graph analysis, and may then delete the outlier or mark as a missing item, so as to perform subsequent missing value processing, which is not limited herein.
Specifically, the abnormal data in the first effective image set can be deleted or filled through abnormal value processing, so that the influence of abnormal values on sample data is removed. The outlier is data in which an anomaly occurs in the customer view set, such as a negative number or an abnormal fluctuation point.
S502: and carrying out missing value processing on the first effective image set to obtain a second effective image set.
The missing value refers to the missing of information caused by the reasons that part of information cannot be acquired temporarily in the customer portrait set, so that part of attribute values are empty, part of information is lost due to some human factors, or some attributes of some objects are unavailable or some information is acquired too much, so that data are not acquired.
S503: and performing feature selection on the second effective image set to obtain a target image set.
In this embodiment, the feature selection includes, but is not limited to, a Filter method, a Wrapper method, or an Embedded method, which are not limited herein. Specifically, the target image set is obtained by carrying out feature selection on the second effective image set so as to delete redundant features or features with less useful information expression, thereby reducing the number of features, improving the accuracy of a subsequent training model and reducing the running time.
In one embodiment, as shown in fig. 6, in step S501, the outlier processing is performed on the client image set to obtain a first effective image set, which specifically includes the following steps:
s601: and detecting the abnormal value of the customer portrait set, and determining the abnormal value of the customer portrait set.
S602: and removing the abnormal value in the customer portrait set to obtain a first effective portrait set. Or,
s603: and identifying the abnormal value as a missing item to obtain a first effective image set.
In this embodiment, the abnormal values in the client image set may be determined by statistical analysis or box-type graph analysis, and then the abnormal values may be deleted or marked as missing items for subsequent missing value processing, which is not limited herein.
Further, the first effective image set corresponds to a plurality of feature factors; as shown in fig. 7, in step S502, the missing value processing is performed on the first effective image set to obtain a second effective image, which specifically includes the following steps:
s701: and detecting the missing value of each characteristic factor, and determining the missing item corresponding to each characteristic factor.
S702: and counting the missing items corresponding to each characteristic factor, and determining the missing rate corresponding to the characteristic factor.
S703: and if the deletion rate is larger than a preset deletion rate threshold value, deleting the characteristic factors.
S704: and if the deletion rate is not greater than the preset deletion rate threshold, carrying out deletion value filling processing on the deletion item corresponding to the characteristic factor to obtain a second effective image set.
In this embodiment, the first effective image set includes a plurality of feature factor columns, and if the missing rate (the number of missing values/the number of samples) of a certain feature factor column is greater than a preset missing rate threshold, the feature factor can be directly removed; if not, the mode, median, mean or default constant of the feature factor column may be automatically filled, and it is understood that other missing value filling strategies may be used to fill the missing values to obtain the second effective image set.
In one embodiment, as shown in fig. 8, in step S204, association rule mining is performed on historical transaction data to obtain a target recommendation rule, which specifically includes the following steps:
s801: initializing a K value, traversing history transaction data, and obtaining a plurality of candidate sets; wherein K is a positive integer.
S802: and determining the corresponding support degree according to the occurrence times of each candidate set in the historical transaction data.
S803: and selecting a candidate set with the support degree not smaller than the preset support degree as a frequent K item set.
The frequent K item set refers to a set of K item sets with a support degree greater than or equal to a minimum support degree (min_sup). Specifically, the initial value of the K value is 1, and at this time, the frequent K item set is a frequent 1 item set, and the frequent 1 item set includes a plurality of 1 item sets. The candidate set refers to 1 set of items generated based on historical transaction data in cycle 1 of the algorithm.
Such as the current historical transaction data shown in the table below,
specifically, by traversing the transaction data, the number of occurrences of each item is determined as shown in the following table,
it can be understood that by calculating the support degree corresponding to each single item set (1 item set), the single item set with the support degree smaller than the preset support degree threshold value is removed, and then the frequent 1 item set can be obtained.
Wherein, the supporting degree calculation formula is as follows: s (x) =σ (x)/M, where σ (x) represents the number of occurrences of each K-term set in the historical transaction data among the frequent K-term sets, and M represents the total number of transactions, which is 10 in the above example.
Specifically, for the above example, assuming that the preset support threshold is 0.2, a single item with a support less than the preset support threshold (i.e., I 6 ) And removing to obtain frequent 1 item set.
S804: and if the number of the item sets contained in the frequent K item sets is greater than K, adding 1 to the K value, and updating the K value.
S805: and removing the K-1 item set where each element in the frequent K-1 item set appears in the historical transaction data less than K-1 times, and updating the frequent K-1 item set.
And if the number of the item sets contained in the frequent K item sets is greater than K, adding 1 to the K value to update the K value. For example, for a frequent 2-item set, such as L 2 ＝{{I 1 I 2 }，{I 1 I 3 }，{I 2 I 3 }，{I 3 I 4 -2 sets of 2 items among the frequent 2 sets of 2 items, the updated K value being 3
Specifically, before calculating the frequent 3 item set (i.e., frequent K item set), the frequent 2 item set (i.e., frequent K-1 item set) needs to be filtered, i.e., the (K-1) 2 item set where the element in the frequent 2 item set (i.e., frequent K-1 item set) occurs less frequently than the element in the historical transaction data (i.e., K-1) times is removed, and the frequent 2 item set is updated.
For the example in step S703, if the number of occurrences of an element is less than K-1 (where the K value is updated to 2), the 2 item set in which the element in the frequent 2 item set is located is removed to update the frequent 2 item set, e.g., the frequent 2 item set is L 2 ＝{{I 1 I 2 }，{I 1 I 3 }，{I 2 I 3 }，{I 3 I 4 }}. Wherein element I 1 ,I 2 ,I 3 ,I 4 The number of occurrences in the historical transaction data is 5,7,6,3, so that the element occurrence is removed from the 2-term set where the element occurrence is less than K-1 (where the K value has been updated to 3), i.e., 2 times, and the element I4 occurrence is removed from the 2-term set { I) where I4 is located, since the element I4 occurrence is less than 3 3 ,I 4 And updates the frequent 2 item set to L 2 ＝{{I 1 ,I 2 },{I 1 ,I 3 },{I 2 ,I 3 }}。
It can be understood that, because the number of occurrences corresponding to an element is low, that is, not frequent, the corresponding super item set is still not frequent, so in this embodiment, before calculating the frequent K item set, the K item set where each element in the frequent K item set occurs in the historical transaction data less than K-1 times is used in advance, so that the infrequent item set can be directly removed, thereby effectively reducing a large number of candidate sets with low support and more items, and improving the algorithm performance.
S806: and processing the frequent K-1 item set according to the natural connection theorem and the pruning algorithm to obtain an updated frequent K item set, and repeatedly executing S804-S806 until the number of item sets contained in the updated frequent K item set is not more than K, so as to obtain a target frequent set.
Specifically, the frequent K-1 item set is processed according to the natural connection theorem and the pruning algorithm, and the updated frequent K item set is obtained. Wherein, the natural connection theorem is that when the first K-1 items in two K item sets are the same, natural connection can be performed, for example, there are two 2 item sets in frequent 2 item sets: { I1, I2} and { I1, I3}, which are naturally connected because the first 1 element is the same, i.e., 3 item sets { I1, I2, I3} are generated by connecting, and, for example, two 3 item sets { I1, I2} and { I3, I4}, which are not connectable because the first 1 element is different, are also described.
The pruning algorithm is an algorithm for obtaining frequent sets by judging whether the support degree of each set in frequent K items is greater than a preset support degree. Specifically, assuming that the preset support degree is 0.2, for example, the candidate set obtained according to the natural connection theorem is { I1, I2, I3}, the candidate set includes 1 set of 3 items, and the support degree of the set of 3 items is 2/10=0.2 (available according to the example table in step S703), the frequent 3 items updated at this time are { I1, I2, I3}, according to the pruning algorithm.
For example, if the frequent 1 item set includes 5 1 item sets, according to the natural connection theorem, the combination of the frequent 1 item sets may be connected to obtain 2 item candidate sets, { I1, I2}, { I1, I3}, { I1, I4}, { I1, I5}, { I2, I3}, { I2, I4}, { I2, I5}, { I3, I4}, { I4, I5}, then 2 item sets with the support degree smaller than the preset support degree are removed according to the pruning algorithm, the number of times each 2 item set appears in the historical transaction data is 4,2,1,1,6,2,0,3,0,0 according to the example table in the step S803, the support degree of each 2 item set is calculated according to the support degree calculation formula, and the 2 item sets with the support degree smaller than the preset support degree are deleted, so as to obtain the frequent 2 item set as L 2 ＝{{I 1 I 2 }，{I 1 I 3 }，{I 2 I 3 }，{I 3 I 4 }}。
Specifically, steps S804 to S806 are repeatedly executed until the number of item sets included in the frequent K item sets is not greater than K, to obtain a final frequent K item set, that is, a target frequent set.
S807: and determining a plurality of candidate rules according to the plurality of subsets corresponding to the target frequent set.
S808: and carrying out confidence calculation on each candidate rule, and taking a subset with the confidence degree larger than a preset confidence degree threshold value as a target recommendation rule.
Wherein, the confidence coefficient calculation formula is C (x- > y) =sigma (xU y)/sigma (x). For example, if the frequent set of targets is { I1, I2, I3}, then its corresponding subset includes { I1}, { I2}, { I3}, { I1, I2}, { I1, I3}, { I2, I3}, by which candidate rules can be determined, namely (I1- > I2, I3), (I2- > I1, I3), (I3- > I1, I2), (I1, I2- > I3), (I1, I3- > I2), (I2, I3- > I1).
And then, calculating the confidence coefficient corresponding to each candidate rule through a confidence coefficient formula, and taking a subset with the confidence coefficient larger than a preset confidence coefficient threshold value as a target recommendation rule.
In this embodiment, the improved association rule mining algorithm is adopted to perform association rule mining on historical transaction data, that is, before fusion of frequent K-1 term sets is performed, K-1 term sets where elements, in which the number of occurrences of each element in the frequent K-1 term sets is less than K-1, in the historical transaction data are located are removed first, so that infrequent term sets are removed, a large number of candidate sets with low support and more projects are effectively reduced, and algorithm performance is improved.
As another embodiment, as shown in fig. 9, in step S204, association rule mining is performed on historical transaction data to obtain a target recommendation rule, which specifically includes the following steps:
s901: initializing a K value and traversing the history transaction data to obtain a plurality of candidate sets.
Specifically, the step S901 is consistent with the step S801, and will not be described herein again to avoid repetition.
S902: and determining the corresponding support degree according to the occurrence times of each candidate set in the historical transaction data.
Specifically, step S902 is consistent with step S802, and is not repeated here.
S903: and selecting a candidate set with the support degree not smaller than the preset support degree as a frequent K item set.
Specifically, the step S903 is consistent with the step S803, and in order to avoid repetition, a detailed description is omitted here.
S904: and if the number of the K item sets in the frequent K item sets is greater than K, adding 1 to the K value, and updating the K value.
Specifically, the step S904 is consistent with the step S804, and is not repeated here.
S905: and removing the K-1 item set where each element in the frequent K-1 item set appears in the historical transaction data less than K-1 times, and updating the frequent K-1 item set.
Specifically, step S905 is consistent with step S805, and is not repeated here.
S906: and processing the frequent K-1 item set according to the natural connection theorem and the pruning algorithm to obtain an updated frequent K item set, and repeatedly executing the steps S904-S906 until the K value is not in the preset K value range to obtain a target frequent set.
It should be noted that, the termination conditions of the Apriori algorithm in this embodiment may include two types, one type is that the number of item sets included in frequent K item sets is not greater than K; the other is that the K value is not within the preset K value range. Wherein, the preset K value range can be set according to experience, such as 1-3.
S907: and determining a plurality of candidate rules according to the plurality of subsets corresponding to the target frequent set.
Specifically, step S907 is consistent with step S807, and is not repeated here.
S908: and carrying out confidence calculation on each candidate rule, and taking a subset with the confidence degree larger than a preset confidence degree threshold value as a target recommendation rule.
Specifically, step S908 is consistent with step S808, and is not repeated here.
It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present invention.
In an embodiment, a product recommendation device is provided, where the product recommendation device corresponds to the product recommendation method in the above embodiment one by one. As shown in fig. 10, the product recommendation apparatus includes a customer characteristic data acquisition module 10, a target detection module 20, a first recommendation module 30, a rule mining module 40, and a product recommendation module 50. The functional modules are described in detail as follows:
the client characteristic data acquisition module 10 is configured to acquire client characteristic data of a client to be recommended.
The target detection module 20 is configured to invoke a pre-trained target detection model to perform target detection on the client feature data, and determine whether the client to be recommended is a target recommendation client.
The first recommending module 30 is configured to obtain historical transaction data of the target recommending client in a preset time period if the client to be recommended is the target recommending client.
The rule mining module 40 is configured to perform association rule mining on the historical transaction data to obtain a target recommendation rule; wherein the target recommendation rule corresponds to a target product set.
The product recommendation module 50 is configured to recommend the target product set to a target recommendation client.
Specifically, the product recommendation device further comprises a target client set acquisition module, a client judgment module, a second recommendation module, a historical transaction data acquisition module and a product recommendation module.
The client set acquisition module is used for acquiring a target recommended client set; wherein the target recommendation client set comprises a plurality of target recommendation clients.
And the client judging module is used for judging whether each target recommended client is a new client or not.
And the second recommendation module is used for comparing the characteristic similarity between the client characteristic data of the new client and the client characteristic data of the non-new client in the target recommendation client set if the client characteristic data of the new client is positive, and determining similar clients of the new client.
The historical transaction data acquisition module is used for acquiring historical transaction data of similar clients and determining a target product set based on the historical data.
And the product recommending module is used for recommending the target product set to a new customer.
Specifically, the product recommendation device further comprises a customer portrait set acquisition module, a preprocessing module and a training module.
The client portrait set acquisition module is used for acquiring a client portrait set; wherein the customer representation set includes training sample data for a plurality of platform customers.
And the preprocessing module is used for preprocessing the client image set to obtain a target image set.
And the training module is used for training the target image set by adopting an XGboost algorithm to obtain a target detection model.
Specifically, the preprocessing module includes an outlier processing unit, a missing value processing unit, and a feature selection unit.
And the outlier processing unit is used for performing outlier processing on the client portrait set to obtain a first effective portrait set.
And the missing value processing unit is used for carrying out missing value processing on the first effective image set to obtain a second effective image set.
And the feature selection unit is used for performing feature selection on the second effective image set to obtain a target image set.
Specifically, the outlier processing unit includes an outlier detection subunit, a first outlier processing subunit, and a second outlier processing subunit.
And the abnormal value detection subunit is used for detecting the abnormal value of the customer portrait set and determining the abnormal value of the customer portrait set.
A first abnormal value processing subunit, configured to remove abnormal values in the client image set, to obtain a first effective image set; or,
and the second abnormal value processing subunit is used for identifying the abnormal value as a missing item to obtain a first effective image set.
Specifically, the first effective image set corresponds to a plurality of feature factors; the missing value processing unit comprises a missing value detection subunit, a missing rate statistics subunit, a first missing value processing subunit and a second missing value processing subunit.
And the missing value detection subunit is used for carrying out missing value detection on each characteristic factor and determining a missing item corresponding to each characteristic factor.
And the deletion rate statistics subunit is used for counting the deletion items corresponding to each characteristic factor and determining the deletion rate corresponding to the characteristic factor.
And the first missing value processing subunit is used for deleting the characteristic factors if the missing rate is greater than a preset missing rate threshold value.
And the second missing value processing subunit is used for carrying out missing value filling processing on the missing items corresponding to the characteristic factors if the missing rate is not greater than a preset missing rate threshold value to obtain a second effective image set.
Specifically, the rule mining module comprises an initialization unit, a support degree determining unit, a frequent K item set obtaining unit, a K value updating unit, a frequent set updating unit, a loop processing unit, a candidate rule determining unit and a target recommendation rule obtaining unit.
The initialization unit is used for initializing a K value and traversing history transaction data to obtain a plurality of candidate sets; wherein K is a positive integer.
And the support degree determining unit is used for determining the corresponding support degree according to the occurrence times of each candidate set in the historical transaction data.
The frequent K item set acquisition unit is used for selecting a candidate set with the support degree not smaller than the preset support degree as a frequent K item set.
And the K value updating unit is used for adding 1 to the K value and updating the K value if the number of the item sets contained in the frequent K item sets is greater than K.
The frequent set updating unit is used for removing the frequent K-1 item set, the K-1 item set where the element with the frequency smaller than K-1 times appears in the historical transaction data is located, and updating the frequent K-1 item set.
And the cyclic processing unit is used for processing the frequent K-1 item set according to the natural connection theorem and the pruning algorithm to obtain an updated frequent K item set, and repeatedly executing the step of adding 1 to the K value and updating the K value if the number of the item sets contained in the frequent K item set is larger than K until the number of the item sets contained in the updated frequent K item set is not larger than K, so as to obtain the target frequent set.
And the candidate rule determining unit is used for determining a plurality of candidate rules according to the plurality of subsets corresponding to the target frequent set.
And the target recommendation rule acquisition unit is used for carrying out confidence calculation on each candidate rule and taking a subset with the confidence degree larger than a preset confidence degree threshold value as a target recommendation rule.
For specific limitations of the product recommendation device, reference may be made to the above limitations of the product recommendation method, and no further description is given here. The respective modules in the above-described product recommendation apparatus may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 10. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a computer storage medium, an internal memory. The computer storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the computer storage media. The database of the computer device is used for storing data, such as historical data, generated or acquired during the execution of the product recommendation method. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a product recommendation method.
In one embodiment, a computer device is provided that includes a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the product recommendation method in the above embodiments when the computer program is executed, such as steps S201-S205 shown in fig. 2, or steps shown in fig. 3-8. Alternatively, the processor may implement the functions of each module/unit in this embodiment of the product recommendation device when executing the computer program, for example, the functions of each module/unit shown in fig. 9, which are not described herein again for avoiding repetition.
In an embodiment, a computer storage medium is provided, and a computer program is stored on the computer storage medium, where the computer program when executed by a processor implements the steps of the product recommendation method in the foregoing embodiment, for example, steps S201 to S205 shown in fig. 2, or steps shown in fig. 3 to 8, which are not repeated herein. Alternatively, the computer program when executed by the processor implements the functions of each module/unit in the above embodiment of the product recommendation device, for example, the functions of each module/unit shown in fig. 9, which are not repeated herein.
Those skilled in the art will appreciate that implementing all or part of the above-described methods in accordance with the embodiments may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.
Claims (9)
1. A method of product recommendation, comprising:
acquiring client characteristic data of a client to be recommended;
invoking a pre-trained target detection model to perform target detection on the client characteristic data, and determining whether the client to be recommended is a target recommendation client or not;
Acquiring a target recommendation client set; wherein the target recommendation client set comprises a plurality of target recommendation clients;
judging whether each target recommended client is a new client or not;
if yes, comparing the characteristic similarity between the client characteristic data of the new client and the client characteristic data of the non-new client in the target recommended client set, and determining similar clients of the new client;
acquiring historical transaction data of the similar clients, and determining a target product set based on the historical transaction data;
recommending the target product set to the new client;
if the client to be recommended is the target recommended client, acquiring historical transaction data of the target recommended client in a preset time period;
performing association rule mining on the historical transaction data to obtain a target recommendation rule; wherein the target recommendation rule corresponds to a target product set;
and recommending the target product set to the target recommendation client.
2. The product recommendation method as claimed in claim 1, wherein prior to said obtaining customer characteristic data of a customer to be recommended, said product recommendation method further comprises:
acquiring a customer portrait set; wherein the customer representation set comprises training sample data of a plurality of platform customers;
Preprocessing the client image set to obtain a target image set;
and training the target portrait set by adopting an XGboost algorithm to obtain the target detection model.
3. The product recommendation method of claim 2, wherein preprocessing the client image set to obtain a target image set comprises:
performing outlier processing on the client portrait set to obtain a first effective portrait set;
performing missing value processing on the first effective image set to obtain a second effective image set;
and performing feature selection on the second effective image set to obtain the target image set.
4. The product recommendation method of claim 3 wherein said performing outlier processing on said set of customer representations to obtain a first set of valid images comprises:
detecting the abnormal value of the customer portrait set, and determining the abnormal value of the customer portrait set;
removing abnormal values in the client portrait set to obtain the first effective portrait set; or,
and marking the abnormal value as a missing item to obtain the first effective image set.
5. The product recommendation method of claim 3 wherein said first active image set corresponds to a plurality of feature factors;
Performing missing value processing on the first effective image set to obtain a second effective image set, including:
detecting the missing value of each characteristic factor, and determining the missing item corresponding to each characteristic factor;
counting the deletion items corresponding to each characteristic factor, and determining the deletion rate corresponding to the characteristic factors;
if the deletion rate is larger than a preset deletion rate threshold, deleting the characteristic factors;
and if the deletion rate is not greater than a preset deletion rate threshold, carrying out deletion value filling processing on the deletion item corresponding to the characteristic factor to obtain the second effective image set.
6. The product recommendation method as claimed in claim 1, wherein said performing association rule mining on said historical transaction data to obtain target recommendation rules comprises:
initializing a K value, traversing the historical transaction data, and obtaining a plurality of candidate sets; wherein K is a positive integer;
determining a corresponding support degree according to the occurrence times of each candidate set in the historical transaction data;
selecting a candidate set with the support degree not smaller than a preset support degree as a frequent K item set;
if the number of the item sets contained in the frequent K item sets is greater than K, adding 1 to the K value, and updating the K value;
Removing a frequent K-1 item set, and updating the frequent K-1 item set, wherein the K-1 item set is where elements with the frequency smaller than K-1 times appear in the historical transaction data;
processing the frequent K-1 item set according to a natural connection theorem and a pruning algorithm to obtain an updated frequent K item set, and repeatedly executing the steps of adding 1 to the K value and updating the K value if the number of item sets contained in the frequent K item set is greater than K until the number of item sets contained in the updated frequent K item set is not greater than K to obtain a target frequent set;
determining a plurality of candidate rules according to a plurality of subsets corresponding to the target frequent set;
and carrying out confidence calculation on each candidate rule, and taking a subset with the confidence degree larger than a preset confidence degree threshold value as the target recommendation rule.
7. A product recommendation device, comprising:
the client characteristic data acquisition module is used for acquiring client characteristic data of a client to be recommended;
the target detection module is used for calling a pre-trained target detection model to carry out target detection on the client characteristic data and determining whether the client to be recommended is a target recommendation client or not;
The client set acquisition module is used for acquiring a target recommended client set; wherein the target recommendation client set comprises a plurality of target recommendation clients;
the client judging module is used for judging whether each target recommended client is a new client or not;
the second recommendation module is used for comparing the characteristic similarity of the client characteristic data of the new client and the client characteristic data of the non-new client in the target recommendation client set if yes, and determining similar clients of the new client;
the historical transaction data acquisition module is used for acquiring the historical transaction data of the similar clients and determining a target product set based on the historical transaction data;
the product recommending module is used for recommending the target product set to the new client;
the first recommending module is used for acquiring historical transaction data of the target recommending client in a preset time period if the client to be recommended is the target recommending client;
the rule mining module is used for carrying out association rule mining on the historical transaction data to obtain a target recommendation rule; wherein the target recommendation rule corresponds to a target product set;
the product recommending module is further used for recommending the target product set to the target recommending client.
8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the product recommendation method according to any one of claims 1 to 6 when the computer program is executed.
9. A computer storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the product recommendation method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011284211.4A CN112418978B (en) | 2020-11-17 | 2020-11-17 | Product recommendation method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011284211.4A CN112418978B (en) | 2020-11-17 | 2020-11-17 | Product recommendation method, device, equipment and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112418978A CN112418978A (en) | 2021-02-26 |
| CN112418978B true CN112418978B (en) | 2024-04-09 |
Family
ID=74831828
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011284211.4A Active CN112418978B (en) | 2020-11-17 | 2020-11-17 | Product recommendation method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112418978B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113157766A (en) * | 2021-03-12 | 2021-07-23 | Oppo广东移动通信有限公司 | Application analysis method and device, electronic equipment and computer-readable storage medium |
| CN113204714A (en) * | 2021-03-23 | 2021-08-03 | 北京中交兴路信息科技有限公司 | User portrait based task recommendation method and device, storage medium and terminal |
| CN113610580B (en) * | 2021-08-10 | 2023-09-19 | 平安科技（深圳）有限公司 | Product recommendation method and device, electronic equipment and readable storage medium |
| CN115392592B (en) * | 2022-10-25 | 2023-03-24 | 浪潮电子信息产业股份有限公司 | Storage product parameter configuration recommendation method, device, equipment and medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102750336A (en) * | 2012-06-04 | 2012-10-24 | 电子科技大学 | Resource individuation recommendation method based on user relevance |
| CN104572863A (en) * | 2014-12-19 | 2015-04-29 | 阳珍秀 | Product recommending method and system |
| CN107590688A (en) * | 2017-08-24 | 2018-01-16 | 平安科技（深圳）有限公司 | The recognition methods of target customer and terminal device |
-
2020
- 2020-11-17 CN CN202011284211.4A patent/CN112418978B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102750336A (en) * | 2012-06-04 | 2012-10-24 | 电子科技大学 | Resource individuation recommendation method based on user relevance |
| CN104572863A (en) * | 2014-12-19 | 2015-04-29 | 阳珍秀 | Product recommending method and system |
| CN107590688A (en) * | 2017-08-24 | 2018-01-16 | 平安科技（深圳）有限公司 | The recognition methods of target customer and terminal device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112418978A (en) | 2021-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112418978B (en) | Product recommendation method, device, equipment and medium | |
| CN110489520B (en) | Knowledge graph-based event processing method, device, equipment and storage medium | |
| US20190295114A1 (en) | Digital banking platform and architecture | |
| CN111339427B (en) | Book information recommendation method, device and system and storage medium | |
| CN109582876B (en) | Tourist industry user portrait construction method and device and computer equipment | |
| CN110598845A (en) | Data processing method, data processing device, computer equipment and storage medium | |
| WO2018014109A1 (en) | System and method for analyzing and searching for features associated with objects | |
| CN114372573B (en) | User portrait information recognition method and device, computer equipment and storage medium | |
| CN115098803B (en) | Book recommendation method and system based on mobile library | |
| CN112035611B (en) | Target user recommendation method, device, computer equipment and storage medium | |
| CN112784168B (en) | Information push model training method and device, information push method and device | |
| CN111582932A (en) | Inter-scene information pushing method and device, computer equipment and storage medium | |
| CN112288279A (en) | Business risk assessment method and device based on natural language processing and linear regression | |
| CN114529399A (en) | User data processing method, device, computer equipment and storage medium | |
| CN115222112A (en) | Behavior prediction method, behavior prediction model generation method and electronic equipment | |
| CN112948681A (en) | Time series data recommendation method fusing multi-dimensional features | |
| CN113094595A (en) | Object recognition method, device, computer system and readable storage medium | |
| CN115689648B (en) | User information processing method and system applied to directional delivery | |
| CN115408620A (en) | Method and device for determining data tag object | |
| US20230368265A1 (en) | Online inference and learning for nonsymmetric determinantal point processes | |
| CN118037458A (en) | Service request processing method, device, computer equipment and storage medium | |
| CN116308651A (en) | Financial product recommendation method and device based on time sequence and computer equipment | |
| CN107943781B (en) | Keyword recognition method and device | |
| CN115757971A (en) | Business object recommendation method and device, computer equipment and storage medium | |
| CN116861071A (en) | Information pushing method, information pushing device, computer equipment, storage medium and program product |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |