CN117132401A

CN117132401A - Product recommendation method, device, electronic equipment, medium and computer program product

Info

Publication number: CN117132401A
Application number: CN202310378779.XA
Authority: CN
Inventors: 吴琪; 陈思念; 涂晴宇; 李艳莉
Original assignee: Industrial and Commercial Bank of China Ltd ICBC
Current assignee: Industrial and Commercial Bank of China Ltd ICBC
Priority date: 2023-04-11
Filing date: 2023-04-11
Publication date: 2023-11-28

Abstract

The present disclosure provides a product recommendation method, apparatus, electronic device, medium, and program product. The method and the device can be used in the technical field of artificial intelligence. The method comprises the steps of obtaining product information corresponding to m products one by one and interaction information between a user and the products; forming a label set of a product corresponding to the product information according to each product information; calculating the influence degree value of each label in each label set on the product corresponding to the label set according to the label sets of m products; calculating the interest value of the user on each overlapping label in the g overlapping labels according to the interaction information of the user and the product, the label set of the product and the preset interest labels set by the user; calculating the predicted value of the user's interest in each of m products according to the influence value of each label in each label set on the product corresponding to the label set and the interest value of the user in each of g overlapped labels; and recommending products to the user according to the interestingness predicted value.

Description

Product recommendation method, device, electronic equipment, medium and computer program product

Technical Field

The present disclosure relates to the field of artificial intelligence technology, and more particularly, to a product recommendation method, apparatus, electronic device, medium, and computer program product.

Background

The banking industry financial products increase exponentially, and how to find an effective financial product recommendation method to meet the personalized needs of users becomes one of the research hotspots at the present stage. Some recommendation methods in the related art are collaborative filtering methods based on similarity between products, and recommend similar products to a user based on the purchase behavior of the user. Because banking industry has rich user information, other recommendation methods in related technologies divide research guest groups first, and then allocate designated product sets to corresponding guest group users according to business experience.

Disclosure of Invention

In view of this, the present disclosure provides a product recommendation method, apparatus, electronic device, computer-readable storage medium, and computer program product that are intelligent and have accurate and comprehensive recommendation results.

One aspect of the present disclosure provides a product recommendation method, comprising: obtaining product information corresponding to m products one by one and interaction information of users and the products, wherein the interaction information of the users on the products comprises purchase information of the users on the products and browsing information of the users on the products, and m is an integer greater than or equal to 1; forming a label set of a product corresponding to the product information according to each piece of product information, wherein the label set comprises at least one label; calculating the influence degree value of each label in each label set on the product corresponding to the label set according to the label sets of the m products; calculating the interest value of the user on each of g overlapped labels according to the interaction information of the user and the product, the label set of the product and the preset user set interest labels, wherein g overlapped labels exist in the set interest labels, the labels of the product purchased by the user and the labels of the product browsed by the user, and g is an integer greater than or equal to 1; calculating the predicted value of the user's interest degree for each product in the m products according to the influence degree value of each label in each label set on the product corresponding to the label set and the interest degree value of the user on each overlapping label in the g overlapping labels; and recommending products to the user according to the ranking of the interestingness predictive value.

According to the product recommendation method, a label set of a product corresponding to the product information is formed according to the obtained information of each product in m products; according to the label sets of m products, the influence degree value of each label in each label set on the product corresponding to the label set can be calculated; according to the interaction information of the user and the product, the label set of the product and the preset interest labels set by the user, the interest value of the user on each of the g overlapped labels can be calculated; according to the influence value of each label in each label set on the product corresponding to the label set and the interest value of the user on each overlapping label in g overlapping labels, the interest prediction value of the user on each product in m products can be calculated; so that products can be recommended to the user according to the ranking of the interestingness prediction values.

According to the interest degree predicted value disclosed by the invention, the influence degree of the label on the product is considered, and the interest degree of the user on the label is considered, so that the interest degree predicted value is more comprehensive and accurate, and the recommendation result is more approximate to the product of the user's true interest. The method can realize intelligent product recommendation, reduces the pressure of service operation, can help users to screen financial products with various and complex information, and is convenient for users to make decisions to purchase products.

In some embodiments, the product information includes a product instruction, and the forming a label set of a product corresponding to the product information according to each product information includes: performing word segmentation processing on the product specifications of each product to obtain word segmentation results; screening the word segmentation result to obtain a first sub-tag set; and taking the first sub-label set as a label set of the product.

In some embodiments, the filtering the word segmentation result to obtain a first sub-tag set includes: constructing a word co-occurrence map according to the word segmentation results, wherein the word segmentation results are used as nodes of the word co-occurrence map, and the co-occurrence relationship between the word segmentation results is used as an edge of the word co-occurrence map; iteratively spreading the weight of the node according to the word co-occurrence map until convergence; taking the weight of the node during convergence as an importance value of a word segmentation result corresponding to the node; and selecting word segmentation results to form the first sub-tag set according to the ranking of the importance values.

In some embodiments, the product information further includes product attribute and/or product management information, and the forming a tag set of a product corresponding to the product information according to each product information includes: taking the product attribute of each product as a second sub-label set of the product; and/or responding to the label of the business personnel to the product according to the product management information, and taking the label as a third sub-label set of the product; and taking the second sub-label set and/or the third sub-label set as the label set of the product.

In some embodiments, the calculating, according to the label sets of the m products, the influence degree value of each label in each label set on the product corresponding to the label set includes: operation S51, determining, as a first frequency, an occurrence frequency of a first target tag in each of the tag sets in the tag set; operation S52 of determining, as a second frequency, a frequency at which the first target tag appears in the tag set of the m products; operation S53, calculating, according to the first frequency and the second frequency, a value of influence of the first target tag on a product corresponding to the tag set; and an operation S54, wherein each label in the label set is sequentially used as the first target label, and operations S51 to S53 are performed to obtain an influence value of each label in the label set on a product corresponding to the label set.

In some embodiments, the calculating the interest value of the user for each of the g overlay tags according to the interaction information of the user with the product, the tag set of the product, and the preset interest tag set of the user includes: determining a first sub-interest degree of a user on each of g overlapped labels of the purchased product according to the purchase information of the user on the product and the label set of the product; determining a second sub-interest degree of the user on each of g overlapped labels of the browsed product according to the browsing information of the user on the product and the label set of the product; presetting a third sub-interest degree of a user on each of the g overlapped labels; and calculating the interest degree value of the user for each of the g coincident labels according to the first sub-interest degree, the second sub-interest degree and the third sub-interest degree of the user for each of the g coincident labels.

In some embodiments, the determining the first sub-interest degree of the user for each of the g overlay labels of the purchased product according to the purchase information of the user for the product and the label set of the product includes: operation S61, determining a first interestingness score of the user for the i-th purchased product; operation S62 of determining the number of times that the second target tag is included in the products purchased all times by the user; operation S63, calculating a first sub-interest degree of the user on the second target label according to the first interest degree score and the times of including the second target label in the products purchased by the user all times; and S64, taking each of the g overlapped labels of the product purchased by the user as a second target label in sequence, and executing operations S61-S63 to obtain the first sub-interest degree of the user on each of the g overlapped labels of the product purchased.

In some embodiments, the determining, according to the browsing information of the product by the user and the tag set of the product, the second sub-interest degree of each of the g overlay tags of the browsed product by the user includes: in operation S71, a second interestingness score of the user for the j-th browsed product is determined according to the total browsing times and each browsing time of the user for the j-th browsed product. Operation S72, determining the number of times that the second target tag is included in the product browsed by the user for all times; operation S73, calculating a second sub-interest degree of the user on the second target label according to the second interest degree score and the times of including the second target label in the products browsed by the user for all times; and S74, taking each of the g overlapped labels of the product browsed by the user as a second target label in sequence, and executing operations S71-S73 to obtain a second sub-interest degree of the user on each of the g overlapped labels of the product browsed.

In some embodiments, the calculating the predicted value of the user's interest level for each of the m products according to the influence level of each tag in each of the tag sets on the product corresponding to the tag set and the interest level of the user on each of the g overlapping tags includes: according to the interest degree value of the user on each of the g overlapped labels, k interesting labels which are interested by the user are determined according to the interest degree value sequence of the g overlapped labels, wherein k is an integer which is more than or equal to 1 and less than or equal to g; determining the same number of the labels in the k interest labels and the label set of each product; and calculating the predicted value of the user's interest degree for each product in the m products according to the number of the k interest tags, the number of the tags in the tag set of each product, the same number of the tags, the influence degree value of each tag in the tag set of each product on the product and the interest degree value of the user on each of the g overlapped tags.

Another aspect of the present disclosure provides a product recommendation device, comprising: the system comprises an acquisition module, a storage module and a display module, wherein the acquisition module is used for executing acquisition of product information corresponding to m products one by one and interaction information of users and the products, wherein the interaction information of the users on the products comprises purchase information of the users on the products and browsing information of the users on the products, and m is an integer greater than or equal to 1; a forming module, configured to perform forming a tag set of a product corresponding to each piece of product information according to each piece of product information, where the tag set includes at least one tag; the first calculation module is used for executing label sets of the m products, and calculating the influence degree value of each label in each label set on the product corresponding to the label set; the second calculation module is used for executing calculation of interest values of the user on each of g overlapped labels according to the interaction information of the user and the product, the label set of the product and preset user set interest labels, wherein g overlapped labels exist in the set interest labels, the labels of the product purchased by the user and the labels of the product browsed by the user, and g is an integer greater than or equal to 1; the third calculation module is used for executing calculation of an interest degree predicted value of a user for each product in the m products according to the influence degree value of each tag in each tag set for the product corresponding to the tag set and the interest degree value of the user for each of the g overlapped tags; and the recommending module is used for executing ranking according to the interestingness predicted value and recommending products to the user.

Another aspect of the present disclosure provides an electronic device comprising one or more processors and one or more memories, wherein the memories are configured to store executable instructions that, when executed by the processors, implement the method as described above.

Another aspect of the present disclosure provides a computer-readable storage medium storing computer-executable instructions that, when executed, are configured to implement a method as described above.

Another aspect of the present disclosure provides a computer program product comprising a computer program comprising computer executable instructions which, when executed, are for implementing a method as described above.

Drawings

The above and other objects, features and advantages of the present disclosure will become more apparent from the following description of embodiments thereof with reference to the accompanying drawings in which:

FIG. 1 schematically illustrates an exemplary system architecture to which methods, apparatuses may be applied according to embodiments of the present disclosure;

FIG. 2 schematically illustrates a flow chart of a product recommendation method according to an embodiment of the present disclosure;

FIG. 3 schematically illustrates a flowchart of forming a tag set of a product corresponding to each product information according to an embodiment of the present disclosure;

FIG. 4 schematically illustrates a flowchart of screening a segmentation result to obtain a first set of sub-tags, according to an embodiment of the present disclosure;

FIG. 5 schematically illustrates a flowchart of forming a tag set of a product corresponding to each product information according to an embodiment of the present disclosure;

FIG. 6 schematically illustrates a flowchart of forming a tag set of a product corresponding to each product information according to an embodiment of the present disclosure;

FIG. 7 schematically illustrates a flowchart of forming a tag set of a product corresponding to each product information according to an embodiment of the present disclosure;

FIG. 8 schematically illustrates a flowchart of calculating a value of influence of each tag in each tag set on a product corresponding to the tag set according to a tag set of m products according to an embodiment of the present disclosure;

FIG. 9 schematically illustrates a flowchart for calculating a user's interest level value for each of g overlay tags based on user interaction information with a product, a set of tags for the product, and a preset user's set of interest tags, according to an embodiment of the present disclosure;

FIG. 10 schematically illustrates a flowchart for determining a first sub-interest level of a user for each of g overlay labels of a purchased product based on user purchase information for the product and a set of labels of the product, according to an embodiment of the present disclosure;

FIG. 11 schematically illustrates a flowchart for determining a second sub-interest level of a user for each of g overlay labels of a browsed product based on user browsing information of the product and a label set of the product, according to an embodiment of the present disclosure;

FIG. 12 schematically illustrates a flowchart for calculating a user's interestingness prediction value for each of m products based on the interestingness value for each of the labels in each of the label sets for the products corresponding to the label set and the user's interestingness value for each of the g overlapping labels, according to an embodiment of the present disclosure;

FIG. 13 schematically illustrates a flow chart of a product recommendation method according to an embodiment of the present disclosure;

FIG. 14 schematically illustrates a block diagram of a product recommendation device, according to an embodiment of the present disclosure;

fig. 15 schematically illustrates a block diagram of an electronic device according to an embodiment of the disclosure.

Detailed Description

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is only exemplary and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. In addition, in the following description, descriptions of well-known structures and techniques are omitted so as not to unnecessarily obscure the concepts of the present disclosure.

In the technical scheme of the disclosure, the acquisition, storage, application and the like of the related personal information of the user all conform to the regulations of related laws and regulations, necessary security measures are taken, and the public order harmony is not violated. In the technical scheme of the disclosure, the processes of acquiring, collecting, storing, using, processing, transmitting, providing, disclosing, applying and the like of the data all conform to the regulations of related laws and regulations, necessary security measures are adopted, and the public order harmony is not violated.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and/or the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.

Where a formulation similar to at least one of "A, B or C, etc." is used, in general such a formulation should be interpreted in accordance with the ordinary understanding of one skilled in the art (e.g. "a system with at least one of A, B or C" would include but not be limited to systems with a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.). The terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more of the described features.

The banking industry financial products increase exponentially, and how to find an effective financial product recommendation method to meet the personalized needs of users becomes one of the research hotspots at the present stage. Some recommendation methods in the related art are collaborative filtering methods based on similarity between products, and recommend similar products to a user based on the purchase behavior of the user. Financial product recommendation based on collaborative filtering algorithm has the following problems: firstly, the problem of cold start exists, and when a new product and/or a new user have no behavior data, effective recommendation cannot be performed; secondly, the problem of sparse scoring data cannot be solved, the scoring matrix is sparse under the general condition, and the model effect is worse; thirdly, potential interests of each user are mined by establishing binary relation between the user and the product, system transition trend of the label-based product label cannot be adapted, and hidden purchasing rules of the product label are ignored.

Because banking industry has rich user information, other recommendation methods in related technologies divide research guest groups first, and then allocate designated product sets to corresponding guest group users according to business experience. The recommendation effect of the recommendation method is very dependent on service experience, and the service operation pressure is increased sharply along with the abundance of passenger groups and products. With the increase of the quantity of financial products, a single page cannot display all the products, and users face the complex financial products with large information quantity and easily sink into information overload, so that decision cannot be made, and the loss of fund purchase potential customers is caused.

Embodiments of the present disclosure provide a product recommendation method, apparatus, electronic device, computer-readable storage medium, and computer program product. The product recommendation method comprises the following steps: obtaining product information corresponding to m products one by one and interaction information of users and the products, wherein the interaction information of the users on the products comprises purchase information of the users on the products and browsing information of the users on the products, and m is an integer greater than or equal to 1; forming a label set of a product corresponding to the product information according to each piece of product information, wherein the label set comprises at least one label; calculating the influence degree value of each label in each label set on the product corresponding to the label set according to the label sets of m products; calculating the interest value of the user on each overlapping label in g overlapping labels according to the interaction information of the user and the product, the label set of the product and the preset interest label set of the user, wherein g overlapping labels exist in the three of the interest label set, the label of the product purchased by the user and the label of the product browsed by the user, and g is an integer greater than or equal to 1; calculating the predicted value of the user's interest in each of m products according to the influence value of each label in each label set on the product corresponding to the label set and the interest value of the user in each of g overlapped labels; and recommending products to the user according to the ranking of the interestingness predictive value.

It should be noted that the product recommendation method, apparatus, electronic device, computer readable storage medium and computer program product of the present disclosure may be used in the field of artificial intelligence technology, and may also be used in any field other than the field of artificial intelligence technology, such as the financial field, and the field of the present disclosure is not limited herein.

Fig. 1 schematically illustrates an exemplary system architecture 100 to which product recommendation methods, apparatuses, electronic devices, computer-readable storage media, and computer program products may be applied, according to embodiments of the present disclosure. It should be noted that fig. 1 is only an example of a system architecture to which embodiments of the present disclosure may be applied to assist those skilled in the art in understanding the technical content of the present disclosure, but does not mean that embodiments of the present disclosure may not be used in other devices, systems, environments, or scenarios.

As shown in fig. 1, a system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

The user may interact with the server 105 via the network 104 using the terminal devices 101, 102, 103 to receive or send messages or the like. Various communication client applications, such as shopping class applications, web browser applications, search class applications, instant messaging tools, mailbox clients, social platform software, etc. (by way of example only) may be installed on the terminal devices 101, 102, 103.

The terminal devices 101, 102, 103 may be a variety of electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablets, laptop and desktop computers, and the like.

The server 105 may be a server providing various services, such as a background management server (by way of example only) providing support for websites browsed by users using the terminal devices 101, 102, 103. The background management server may analyze and process the received data such as the user request, and feed back the processing result (e.g., the web page, information, or data obtained or generated according to the user request) to the terminal device.

It should be noted that, the product recommendation method provided by the embodiments of the present disclosure may be generally performed by the server 105. Accordingly, the product recommendation device provided by the embodiments of the present disclosure may be generally disposed in the server 105. The product recommendation method provided by the embodiments of the present disclosure may also be performed by a server or a server cluster that is different from the server 105 and is capable of communicating with the terminal devices 101, 102, 103 and/or the server 105. Accordingly, the product recommendation apparatus provided by the embodiments of the present disclosure may also be provided in a server or a server cluster that is different from the server 105 and is capable of communicating with the terminal devices 101, 102, 103 and/or the server 105.

It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

The product recommendation method of the embodiment of the present disclosure will be described in detail below by way of fig. 2 to 12 based on the scenario described in fig. 1.

Fig. 2 schematically illustrates a flow chart of a product recommendation method according to an embodiment of the present disclosure.

As shown in fig. 2, the product recommendation method of this embodiment includes operations S210 to S260.

In operation S210, product information corresponding to m products one-to-one and interaction information between the user and the products are obtained, wherein the interaction information between the user and the products includes purchase information of the user and browsing information of the user and the products, and m is an integer greater than or equal to 1.

In operation S220, a tag set of a product corresponding to each product information is formed according to the product information, wherein the tag set includes at least one tag.

As an embodiment, the product information may include a product specification, and as shown in fig. 3, operation S220 forms a tag set of a product corresponding to the product information according to each product information, including operations S221 to S223.

In operation S221, a word segmentation process is performed on the product specification of each product, and a word segmentation result is obtained. It will be appreciated that the product description may be unstructured information, specifically documents describing financial product descriptions, such as documents describing suitable purchasing groups, sales ratings, investment ranges, and the like. The product specification is composed of paragraphs, the paragraphs are composed of sentences, and the sentences are composed of words, so that the product specification of each product is subjected to word segmentation processing to obtain word segmentation results, for example, the product specification can be segmented according to the paragraphs, the paragraphs are segmented according to the whole sentences, the segmented whole sentences are cleaned and filtered, the filtered whole sentences are segmented according to the words, the words are cleaned and filtered, namely, after punctuation is removed from the words, words are stopped and words are normalized, only the terms with specified parts of speech are reserved.

In operation S222, the segmentation result is filtered to obtain a first sub-tag set.

In some examples, as shown in fig. 4, operation S222 filters the word segmentation result to obtain a first sub-tag set, including operations S2221-S2224.

In operation S2221, a word co-occurrence graph is constructed according to the word segmentation results, wherein the word segmentation results are used as nodes of the word co-occurrence graph, and the co-occurrence relationship between the word segmentation results is used as edges of the word co-occurrence graph.

In operation S2222, the weights of the propagating nodes are iterated according to the word co-occurrence graph until convergence.

In operation S2223, the weight of the node at the time of convergence is used as the importance value of the word segmentation result corresponding to the node.

In operation S2224, the word segmentation results are selected to form a first sub-tag set according to the ranking of the importance values. The screening of the segmentation results can be conveniently achieved through operations S2221-S2224, and the first sub-tag set is obtained.

In operation S223, the first sub-label set is taken as a label set of the product.

The formation of the tag set of the product corresponding to each product information according to the product information can be facilitated through operations S221 to S223.

In some examples, the product information may also include product attributes and/or product management information, e.g., product attributes may include the nature of the financial product, the risk level, the distribution scale, and so forth; for example, the product management information may include financial product manager, fund company, and winning situation, etc.

As shown in fig. 5 to 7, operation S222 forms a tag set of a product corresponding to each product information according to the product information, including operation S2225 and/or operation S2226, and operation S2227.

In operation S2225, the product attributes of each product are used as the second sub-label set for the product.

In operation S2226, in response to the business person labeling the product according to the product management information, the label is taken as the third sub-label set of the product. For example, the label may be a gold medal financial product manager, a financial product manager is excellent, a large funds company and/or a gold cow prize is earned, or the like.

In operation S2227, the second sub-label set and/or the third sub-label set are used as label sets for the product. Thus, in one example, a labelset of one product may include a first sub-labelset, a second sub-labelset, and a third sub-labelset; in another example, a set of labels for one product may include a first set of sub-labels and a third set of sub-labels; in yet another example, a set of labels for one product may include a first sub-set of labels and a second sub-set of labels.

The operation S2225 and the operation S2227 may facilitate the formation of a tag set of a product corresponding to each product information according to the product information; the operation S2226 and the operation S2227 can also facilitate the formation of a label set of a product corresponding to each product information according to the product information; the forming of the tag set of the product corresponding to each product information according to the product information can be similarly facilitated through operation S2225, operation S2226 and operation S2227. The first sub-label set, the second sub-label set and the third sub-label set are extracted from the product specification, the second sub-label set is composed of product attributes, and the third sub-label set is composed of labels, so that the types and the contents of the labels in the label set are rich, the interestingness predicted value is accurate, and the recommendation result is good.

In operation S230, an influence degree value of each label in each label set on a product corresponding to the label set is calculated according to the label sets of m products.

As one possible implementation manner, as shown in fig. 8, operation S230 calculates, according to a tag set of m products, an influence degree value of each tag in each tag set on a product corresponding to the tag set, including operations S51 to S54.

In operation S51, the occurrence frequency of the first target tag in each tag set in the tag set is determined as the first frequency. For a product, there is one and only one label in the set of labels, so the frequency of occurrence of the first target label a in the set of labels of product a, i.e. the first frequency TF (a, a), can be obtained by formula (1).

In operation S52, a frequency at which the first target tag appears in the tag set of m products is determined as the second frequency. For example, the frequency of occurrence of the first target tag a in the tag set of m products, that is, the second frequency IDF (a), can be obtained by the formula (2).

In operation S53, a degree of influence value of the first target tag on the product corresponding to the tag set is calculated according to the first frequency and the second frequency. For example, the influence value Q of the first target label a on the product a can be obtained by the formula (3).

In operation S54, each tag in the tag set is sequentially used as a first target tag, and operations S51 to S53 are performed to obtain a value of influence of each tag in the tag set on a product corresponding to the tag set. Thus, it is possible to easily implement the tag sets according to m products by operations S51 to S54, and calculate the influence degree value of each tag in each tag set on the product corresponding to the tag set.

In operation S240, according to the interaction information of the user and the product, the tag set of the product, and the preset interest tag set of the user, the interest value of the user for each of the g overlay tags is calculated, where g overlay tags exist among the interest tag set, the tag of the product purchased by the user, and the tag of the product browsed by the user, and g is an integer greater than or equal to 1. It will be appreciated that since the tag set of each product has been determined in operation S220, the tag sets of the purchased product and the browsed product can be determined, and thus, the overlapping portions of the tag of the purchased product, the tag of the browsed product, and the set interest tag, that is, g overlapping tags, can be determined.

In some examples, the user may belong to a specific guest group, the information of the guest group to which the user belongs may be obtained in the banking system, each guest group may have a preset interest tag, each user may belong to a plurality of different guest groups, and the interest tag of the guest group to which the user belongs is the interest tag of the user. Of course, the interest tag may be set in advance for each user individually.

As an implementation manner, as shown in fig. 9, operation S240 calculates an interest value of the user for each of the g overlay tags according to the interaction information of the user with the product, the tag set of the product, and the preset interest tags set by the user, including operations S241 to S244.

In operation S241, a first sub-interest level of the user in each of g overlay tags of the purchased product is determined according to the purchase information of the product by the user and the tag set of the product.

In some examples, as shown in fig. 10, operation S241 determines a first sub-interest level of the user for each of g overlay labels of the purchased product according to the purchase information of the product by the user and the label set of the product, including operations S61 to S64.

In operation S61, a first interestingness score of the user for the i-th purchased product is determined, i being an integer of 1 or more, where the first interestingness score may be noted as S1 (u, i), S ₁ (u，i)＝1。

In operation S62, the number of times the second target tag is included in the product purchased all times by the user is determined. Here, the number of times that the second target tag b is included in the products purchased all times by the user u may be represented by N (u, a).

In the operation S63 of the present invention,and calculating the first sub-interestingness of the user on the second target label according to the first interestingness score and the times of including the second target label in the products purchased by the user all times. For example, the first sub-interest degree P of the user u on the second target tag b can be obtained through the formula (4) ₁ 。

And S64, taking each of the g overlapped labels of the product purchased by the user as a second target label in sequence, and executing operations S61-S63 to obtain the first sub-interest degree of the user on each of the g overlapped labels of the product purchased.

The operations S61 to S64 may facilitate determining a first sub-interest level of the user for each of the g overlay tags of the purchased product according to the purchase information of the product by the user and the tag set of the product.

In operation S242, a second sub-interest level of the user in each of the g overlay labels of the browsed product is determined according to the browsing information of the user on the product and the label set of the product.

As one implementation manner, as shown in fig. 11, operation S242 determines a second sub-interest level of the user for each of g overlay labels of the browsed product according to the browsing information of the user for the product and the label set of the product, including operations S71 to S74.

In operation S71, a second interestingness score of the user for the product browsed for the j-th time is determined according to the total browsing times and each browsing time of the user for the product browsed for the j-th time, and j is an integer greater than or equal to 1. Here, the second interestingness score of user u for the jth browsed product may be used as S ₂ (u, j) represents.

In some examples, when user u browses a product only once, user u has a second interestingness score S for the product ₂ (u, j) =1; when user u browses a product z times, z is an integer greater than 1And each browsing time is more than or equal to 300 seconds, the second interestingness score S of the product is given by the user u ₂ (u, j) =1; when user u browses a product z times, z is an integer greater than 1, and at least one of the z times browses less than 300 seconds, a second interestingness score S of the product is given to user u ₂ (u, j) can be obtained by the formula (5).

Wherein Time (j) represents the Time of the jth browsing of the product by user u, and Time (h) represents the Time of the jth browsing of the product by user u.

In operation S72, the number of times the second target tag is included in the product that the user browses all times is determined. Here, the number of times the second target tag b is included in the product browsed by the user u all times may be represented by M (u, a).

In operation S73, a second sub-interest level of the user in the second target tag is calculated according to the second interest level score and the number of times the second target tag is included in the product browsed by the user for all times. For example, a second sub-interest degree P of the user u on the second target tag b can be obtained through formula (6) ₂ 。

In operation S74, each of the g overlay labels of the product browsed by the user is sequentially used as a second target label, and operations S71 to S73 are performed to obtain a second sub-interest level of the user in each of the g overlay labels of the browsed product.

Through operations S71 to S74, it is possible to facilitate determining the second sub-interest level of the user for each of the g overlay labels of the browsed product according to the browsing information of the user for the product and the label set of the product.

In operation S243, the user is preset to each of the g overlay labelsIs a third sub-interestingness of (2). Here, the third sub-interest level of the user u in the tag a among the g tags may be represented by S ₃ (u, a) represents, set S ₃ (u，a)＝1。

In operation S244, an interest value of the user for each of the g overlay labels is calculated according to the first sub-interest level, the second sub-interest level, and the third sub-interest level of the user for each of the g overlay labels. In some examples, the first sub-interestingness, the second sub-interestingness, and the third sub-interestingness may be given a weight coefficient, e.g., the interestingness value of user u for an overlapping tag a of the g overlapping tags is denoted by F (u, a), which may be obtained by equation (7).

F(u，a)＝αP ₁ +βP ₂ +S ₃ (u，a) (7)

Wherein α and β are weight coefficients. Thus, through operations S241 to S244, it is possible to facilitate calculation of the user 'S interest value for each of the g overlay tags according to the user' S interaction information with the product, the product 'S tag set, and the preset user' S set interest tag.

In operation S250, a predicted value of interest of the user for each of the m products is calculated according to the influence value of each of the tags in each of the tag sets on the products corresponding to the tag set and the interest value of the user for each of the g overlapping tags.

As a possible implementation manner, as shown in fig. 12, operation S250 calculates an interest degree prediction value of a user for each of m products according to an influence degree value of each tag in each tag set for a product corresponding to the tag set and an interest degree value of the user for each of g overlapped tags, including operations S251 to S253.

In operation S251, k interest tags of interest to the user are determined according to the interest level value of the user for each of the g overlay tags and the interest level value ranking of the g overlay tags, where k is an integer greater than or equal to 1 and less than or equal to g.

In operation S252, the same number of tags among the k interest tags and the tag set of each product is determined. Here, the same number of tags in the k interest tags as the tag set of each product may be represented by L.

In operation S253, according to the number of k interest tags, the number of tags in the tag set of each product, the same number of tags, the influence value of each tag in the tag set of each product on the product, and the interest value of the user on each of the g overlapping tags, the interest prediction value of the user on each of the m products is calculated. Here, the predicted value of interest level of the user u for the product a among m products may be expressed as cos (u, a), and coS (u, a) may be obtained by the formula (8).

Wherein Y represents the number of tags in the tag set of the product, a represents the xth tag in the tag set, Q represents the influence value of the tag a on the product A, and F (u, a) represents the interest value of the user u on the overlapped tag a in the g overlapped tags.

In operation S260, products are recommended to the user according to the ranking of the interestingness prediction values. For example, the first products with highest interest degree of each customer can be screened out to form a product set, a strategy is formed by combining marketing rules, and the customers are reached by means of mobile banking or customer manager recommendation and the like.

Compared with collaborative filtering recommendation algorithm recommendation in the prior art, the method disclosed by the invention has no cold start problem, and can effectively recommend when a new product and/or a new user has no behavior data because the method disclosed by the invention has small dependency on the behavior data of the new product and/or the new user; because the method disclosed by the invention has small degree of dependence on product grading, the problem of sparse grading matrix and poor model effect can not occur under the conditions of sparse grading data and large data scale; the methods of the present disclosure may adapt the system transition trend of label-based product labels.

A product recommendation method according to an embodiment of the present disclosure is described in detail below with reference to fig. 13. It is to be understood that the following description is exemplary only and is not intended to limit the disclosure in any way.

The product recommendation method comprises the steps of obtaining various products, users and transaction information through a service system or data lake data; generating a product label in an automatic extraction mode of service labeling and product specifications; and attributing the user to the corresponding guest group through the existing guest group information. After the importance of the tag and the interest degree of the user on the product are calculated, the mapping of the user product in each guest group is realized, and the interest degree of the user on the product is calculated; and combining Top N user product relations with highest product interest degree screening values of all groups of users to form a user recommended product result list, and reaching users through channels such as mobile banking, online banking and the like.

Fig. 13 is a flowchart of a product recommendation method, including the following.

Step S1: and acquiring information, namely acquiring all user attribution guest groups, fund products, purchasing behavior data and the like of the banking system.

1. User attribution guest group data: including military personnel groups, generation wages groups, public service personnel groups, etc., wherein one customer may belong to a different group.

2. Fund product data.

1) The shared structured information is stored in the rows, mainly the fund attribute information, such as distribution mode, fund type, fund risk level, etc.

2) The shared unstructured information is stored in the rows, and is mainly a fund product specification document.

3) The structured information maintained by the business personnel mainly comprises the fund information which is researched by the client manager and the management department, such as the condition of the fund manager and the condition of the fund company.

3. Transaction information: the user purchases funds including funds purchased in the last year, transaction amount, time since transaction, etc.

Step S2: the product tags generate a generalized description of the processing of the funding product resources, i.e., by the funding product data. The method can be roughly divided into service system extraction, service labeling and product instruction extraction according to different label generation modes; when the product specification is extracted by a new-fund online no-service system and service labeling label information is carried out, labels can be generated through extraction, and potential purchasing customers matched with the labels in the label library can solve the problem of cold starting of the product.

1) And (3) extracting a service system: the general information of various products, which can be directly obtained through the service system or the data lake structured data, is generally the attribute information of the foundation products, including the foundation types, risk grades, release scales and the like.

2) Service labeling: i.e., a letter department or a customer manager adds descriptive labels to the product based on his own experience, such as fund manager is excellent, large fund company, earned a gold cow prize, etc.

3) And (3) extracting a product specification: the information which is not stored in the tag in the business system is extracted through the product specification, and the information is generally suitable for purchasing groups, sales ratings, investment ranges and the like. Specific product specification extraction steps are shown in S201-S208.

Step S3: the importance calculation of the label on the product, namely the influence degree of different product labels on the product is different, and the influence of each product label on the product needs to be quantified. Step S2 obtains a label set for the product, and the influence degree of the label on the foundation product is calculated based on the label set referencing a classical TF-IDF algorithm. The main idea is that a label is high in frequency of occurrence and rarely occurs in other products, and the label is considered to have good distinguishing capability. The specific steps of label importance calculation are shown in S301-S304.

Step S4: the user belongs to the guest group. And defining the passenger group participating in recommendation according to the service definition, and acquiring the user under the corresponding passenger group.

Step S5: the interest degree calculation of the user on the tag, namely, for the guest group outlined in the step S4, the interest degree of each user group on the tag is different, and the interest degree of the user needs to be quantified. The user's interest in the tag is represented by business definition, historical purchase information, and browsing data. For each of the delineated guest groups, the interest level of the user in the label is calculated, and the specific steps are shown in S501-S504.

Step S6: and calculating the interest degree of the user product in each guest group, namely calculating the interest degree of the user on the product through cosine similarity by combining the importance of the label on the product.

cOs (u, A) is the predicted value of interest of user u in product A, L is the same number of tags in product A and the tags of interest of user u, M is the total number of tags of interest of user u, N is the total number of tags in product A, a is the ith tag in the tag set, F (u, a) _i Representing user u's interest level in the ith tag a in the tag set, F (a, A) _i A value indicating the weight of the i-th tag a in the product a.

Step S7: and (3) forming a user recommended product result, namely combining the user product interestingness in each customer group calculated in the step (S6), screening a topN product set with the highest interestingness for each customer, forming a strategy by combining marketing rules, and touching the customers in a mobile banking or customer manager mode.

Step S201: the method mainly comprises the steps of obtaining data sources, wherein the data sources mainly comprise product specifications and an existing label set and a financial word stock of the product stored by a business system.

Step S202: the whole sentence division, i.e., the product specification, is made up of paragraphs, which are made up of sentences, and therefore the product specification needs to be divided in whole sentences, i.e., t= [ S1, S2, ], sn ].

Step S203: word segmentation, cleaning and filtering, i.e. sentences consisting of words of smaller granularity, so that for each sentence S during extraction _i E T, namely after the punctuation is removed from the segmented words, the stop words and the vocabulary are normalized after the segmented words are needed to be segmented, only the vocabulary entry with the assigned part of speech is reserved, namely S _i ＝[t _i，1 ，t _i，2 ，...，t _i，m Wherein t is _i，j ∈S _j Is a candidate keyword after reservation.

Step S204: constructing a candidate keyword graph, namely constructing G= (V, E), wherein V is a node set, and is composed of candidate keywords generated in the step S203, then constructing edges between any two points by adopting a co-occurrence relation (co-current), wherein edges exist between two nodes only when corresponding vocabularies coexist in a window with the length of K, and K represents the window size, namely K words at most coexist.

Step S205: and calculating the importance of the entry based on the network connection characteristics, and iteratively propagating the weight of each node until convergence.

Wherein d is a damping coefficient, the value range is 0 to 1, and the value is 0.85. I.e. the error rate at any point is less than the given limit value of 0.0001, convergence is achieved.

Step S206: and screening candidate keywords, namely taking Top N values with the maximum node weight values as candidate keywords.

Step S207: and (3) the candidate keywords are rescreened, namely, label descriptions of the foundation products in the business system are removed, and focused vocabulary in the non-financial field in the cut words is removed through a financial word stock.

Step S208: and obtaining a final product specification extraction label set through the steps.

Step S301: and acquiring product label information. I.e. taking the product id as the dimension, the product tag dataset generated by step S2 is obtained.

Step S302: word frequency calculation TF, i.e. calculating the frequency of occurrence of a certain tag in a certain fund product, since there is one and only one tag per tag for that product, and therefore

Step S303: the reverse file frequency calculates IDF, namely the frequency of a specific label in the whole product label library, so as to avoid the influence of the calculation result on the later recommendation, and the method ensures that

Step S304: the TF-IDF is calculated, namely the influence degree of a certain label on a certain foundation product is calculated.

A high frequency tag within a particular product, and the tag has a low frequency throughout the tag set. A high weight TF-IDF can be generated.

Step S501: and calculating the interest degree of the user in the product according to the purchase information. For a user A, each fund transaction information can be obtained by storing a shared transaction log table in a row, and the ith purchase record of the user u is corresponding to the purchased productThe interestingness score may be expressed as: s is S ₁ (u，i)＝1。

Step S502: and according to the interest degree of the browsing calculation user on the product, for a user A, the browsing information of the client can be obtained through a mobile phone bank and an internet banking buried point log table which are stored and shared in a row. If the user browses a certain product, the user is informed of a certain interest in the product; the stay time of the user on the page can reflect the interest degree of the user on the product to a certain extent, and the longer the user browses, the higher the interest degree is indicated, and the shorter the stay time is, the lower the interest degree is indicated. The browsing time corresponding to the ith browsing record is expressed as:

in order to make personalized recommendations to the user, the product scores that the user browses need to be quantified. Specifically quantified as follows, S ₂ (u, m) is the interestingness score for the mth related product among the products browsed by user u:

1) The user browses only once, S ₂ (u，m)＝1

2) If the browsing time of the user is greater than or equal to 300 seconds, S ₂ (u，m)＝1

3) Other cases

Step S503: the service defines the relationship between the user and the tag. The business self-matches the guest group with the label according to experience, and then any user u in the guest group has interest degree F of the user to the label a matched with the business ₃ (u，a)＝1。

Step S504: and integrating and calculating the interest degree of the user and the tag. Integrating purchase information, browsing information, service definition information, the interest degree F (u, a) of the user u in the tag a can be expressed as:

n (u, a) represents the number of times tag a is involved in the historical purchase data of user u, S ₁ (u, i) represents the interestingness score of the ith order of user u for the purchased product.

M (u, a) represents the number of times tag a is involved in the product browsed by user u, S ₂ (u, i) represents the product interestingness score of the tag a related to the mth word in the product browsed by the user u.

F3 (u, a) represents the product interestingness score for user u for tag a in the business definition.

Alpha and beta respectively represent weights of scores obtained by order data and product browsing data. The specific values are determined by business specialists and data analysis.

Based on the product recommendation method, the disclosure further provides a product recommendation device. The product recommendation device 10 will be described in detail below with reference to fig. 14.

Fig. 14 schematically illustrates a block diagram of a product recommendation device 10 according to an embodiment of the present disclosure.

The product recommendation device 10 includes an acquisition module 1, a formation module 2, a first calculation module 3, a second calculation module 4, a third calculation module 5, and a recommendation module 6.

Acquisition module 1, acquisition module 1 is configured to perform operation S210: product information corresponding to m products one by one and interaction information of users and the products are obtained, wherein the interaction information of the users on the products comprises purchase information of the users on the products and browsing information of the users on the products, and m is an integer greater than or equal to 1.

Forming a module 2, the forming module 2 being configured to perform operation S220: and forming a label set of a product corresponding to the product information according to each product information, wherein the label set comprises at least one label.

The first calculating module 3, the first calculating module 3 is configured to perform operation S230: and calculating the influence degree value of each label in each label set on the product corresponding to the label set according to the label sets of m products.

The second calculation module 4, the second calculation module 4 is configured to perform operation S240: according to the interaction information of the user and the product, the label set of the product and the preset interest label set of the user, calculating the interest value of the user on each of g superposition labels, wherein g superposition labels exist in the three of the interest label set, the label of the product purchased by the user and the label of the product browsed by the user, and g is an integer greater than or equal to 1.

The third calculation module 5, the third calculation module 5 is configured to perform operation S250: and calculating the predicted value of the user's interest degree for each product in the m products according to the influence degree value of each label in each label set on the product corresponding to the label set and the interest degree value of the user for each overlapping label in the g overlapping labels.

The recommendation module 6, the recommendation module 6 is configured to perform operation S260: and recommending products to the user according to the ranking of the interestingness predictive value.

According to some embodiments of the present disclosure, the product information includes a product specification, and the forming module may include a word segmentation unit, a screening unit, and a first determination unit.

The word segmentation unit is used for carrying out word segmentation processing on the product specifications of each product to obtain word segmentation results.

And the screening unit is used for screening the word segmentation result to obtain a first sub-tag set.

And the first determining unit is used for taking the first sub-label set as the label set of the product.

According to some embodiments of the present disclosure, the screening unit may include a construction element, an iteration element, a first determination element, and a selection element.

The construction element is used for constructing a word co-occurrence map according to word segmentation results, wherein the word segmentation results are used as nodes of the word co-occurrence map, and the co-occurrence relationship between the word segmentation results is used as an edge of the word co-occurrence map.

And the iteration element is used for iterating the weights of the propagation nodes according to the word co-occurrence map until convergence.

The first determining element is used for taking the weight of the node in convergence as the importance value of the word segmentation result corresponding to the node.

And the selecting element is used for selecting word segmentation results to form a first sub-tag set according to the ranking of the importance values.

According to some embodiments of the present disclosure, the product information further includes product attribute and/or product management information, and the forming module may include a second determining unit and/or a third determining unit, and a fourth determining unit.

A second determining unit, configured to use the product attribute of each product as a second sub-label set of the product; and/or

The third determining unit is used for responding to the labeling of the business personnel to the product according to the product management information, and taking the labeling as a third sub-label set of the product; and

and the fourth determining unit is used for taking the second sub-label set and/or the third sub-label set as the label set of the product.

According to some embodiments of the present disclosure, the first computing module may include a fifth determining unit, a sixth determining unit, a first computing unit, and a repeating executing unit.

A fifth determining unit for determining, as a first frequency, a frequency of occurrence of a first target tag in each tag set in the tag set, in operation S51.

A sixth determining unit for determining, as a second frequency, a frequency at which the first target tag appears in the tag set of m products, in operation S52.

The first calculating unit is configured to calculate, according to the first frequency and the second frequency, a value of influence of the first target tag on a product corresponding to the tag set in operation S53.

And the repeated execution unit is used for sequentially taking each label in the label set as a first target label and executing operations S51-S53 to obtain the influence degree value of each label in each label set on the product corresponding to the label set in operation S54.

According to some embodiments of the present disclosure, the second calculation module may include a seventh determination unit, an eighth determination unit, a setting unit, and a second calculation unit.

And a seventh determining unit for determining a first sub-interest degree of the user for each of g overlapping labels of the purchased product according to the purchase information of the user for the product and the label set of the product.

And the eighth determining unit is used for determining the second sub-interest degree of the user on each of the g overlapped labels of the browsed product according to the browsing information of the user on the product and the label set of the product.

The setting unit is used for presetting a third sub-interest degree of a user for each of the g overlapped labels.

The second calculation unit is used for calculating the interest degree value of the user for each of the g coincident labels according to the first sub-interest degree, the second sub-interest degree and the third sub-interest degree of the user for each of the g coincident labels.

According to some embodiments of the present disclosure, the seventh determining unit may include a second determining element, a third determining element, a first calculating element, and a first repeating executing element.

And a second determining element for determining a first interestingness score of the user for the i-th purchased product, in operation S61.

And a third determining unit for determining the number of times the second target tag is included in the product purchased all times by the user, in operation S62.

The first calculating element is configured to calculate a first sub-interest level of the user in the second target tag according to the first interest level score and the number of times the second target tag is included in the products purchased all times by the user in operation S63.

And the first repeated execution element is used for sequentially taking each of the g overlapped labels of the product purchased by the user as a second target label, and executing operations S61-S63 to obtain the first sub-interest degree of the user on each of the g overlapped labels of the purchased product.

According to some embodiments of the present disclosure, the eighth determination unit may include a fourth determination element, a fifth determination element, a second calculation element, and a second repetition execution element.

And a fourth determining unit for determining a second interestingness score of the user for the j-th browsed product according to the total browsing times of the user for the j-th browsed product and each browsing time of the user.

And a fifth determining element for determining the number of times the second target tag is included in the product that the user browses all times in operation S72.

The second calculating element is configured to calculate a second sub-interest level of the user in the second target tag according to the second interest level score and the number of times the second target tag is included in the product browsed by the user for all times in operation S73.

And the second repeating executing element is used for executing operation S74, wherein each overlapping label in g overlapping labels of the product browsed by the user is sequentially used as a second target label, and operations S71-S73 are executed to obtain a second sub-interest degree of the user on each overlapping label in g overlapping labels of the product browsed by the user.

According to some embodiments of the present disclosure, the third calculation module may include a ninth determination unit, a tenth determination unit, and a third calculation unit.

And the ninth determining unit is used for determining k interesting labels interested by the user according to the interest degree value of the user on each overlapping label in the g overlapping labels and the interest degree value sequence of the g overlapping labels, wherein k is an integer which is more than or equal to 1 and less than or equal to g.

And a tenth determining unit for determining the same number of tags in the k interest tags as the tag set of each product.

The third calculation unit is used for calculating the predicted value of the user's interest degree for each product in m products according to the number of k interest labels, the number of labels in the label set of each product, the same number of labels, the influence degree value of each label in the label set of each product on the product and the interest degree value of the user on each overlapping label in the g overlapping labels.

Since the product recommendation device 10 is set based on the product recommendation method, the beneficial effects of the product recommendation device 10 are the same as those of the product recommendation method, and will not be described here again.

In addition, according to the embodiment of the present disclosure, any of the plurality of modules of the acquisition module 1, the formation module 2, the first calculation module 3, the second calculation module 4, the third calculation module 5, and the recommendation module 6 may be incorporated in one module to be implemented, or any of the plurality of modules may be split into a plurality of modules. Alternatively, at least some of the functionality of one or more of the modules may be combined with at least some of the functionality of other modules and implemented in one module.

According to embodiments of the present disclosure, at least one of the acquisition module 1, the formation module 2, the first calculation module 3, the second calculation module 4, the third calculation module 5, and the recommendation module 6 may be implemented at least in part as hardware circuitry, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or as hardware or firmware in any other reasonable way of integrating or packaging the circuitry, or as any one of or a suitable combination of any of the three.

Alternatively, at least one of the acquisition module 1, the formation module 2, the first calculation module 3, the second calculation module 4, the third calculation module 5 and the recommendation module 6 may be at least partially implemented as computer program modules which, when executed, may perform the respective functions.

Fig. 15 schematically illustrates a block diagram of an electronic device adapted to implement the above-described method according to an embodiment of the present disclosure.

As shown in fig. 15, an electronic device 900 according to an embodiment of the present disclosure includes a processor 901 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 902 or a program loaded from a storage portion 908 into a Random Access Memory (RAM) 903. The processor 901 may include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or an associated chipset and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), or the like. Processor 901 may also include on-board memory for caching purposes. Processor 901 may include a single processing unit or multiple processing units for performing the different actions of the method flows according to embodiments of the present disclosure.

In the RAM 903, various programs and data necessary for the operation of the electronic device 900 are stored. The processor 901, the ROM 902, and the RAM 903 are connected to each other by a bus 904. The processor 901 performs various operations of the method flow according to the embodiments of the present disclosure by executing programs in the ROM 902 and/or the RAM 903. Note that the program may be stored in one or more memories other than the ROM 902 and the RAM 903. The processor 901 may also perform various operations of the method flow according to embodiments of the present disclosure by executing programs stored in the one or more memories.

According to an embodiment of the disclosure, the electronic device 900 may also include an input/output (I/O) interface 905, the input/output (I/O) interface 905 also being connected to the bus 904. The electronic device 900 may also include one or more of the following components connected to the I/O interface 905: an input section 906 including a keyboard, a mouse, and the like; an output portion 907 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and a speaker; a storage portion 908 including a hard disk or the like; and a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to an input/output (I/O) interface 905 as needed. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is installed as needed on the drive 910 so that a computer program read out therefrom is installed into the storage section 908 as needed.

The present disclosure also provides a computer-readable storage medium that may be embodied in the apparatus/device/system described in the above embodiments; or may exist alone without being assembled into the apparatus/device/system. The computer-readable storage medium carries one or more programs which, when executed, implement methods in accordance with embodiments of the present disclosure.

According to embodiments of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to embodiments of the present disclosure, the computer-readable storage medium may include ROM 902 and/or RAM 903 and/or one or more memories other than ROM 902 and RAM 903 described above.

Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowcharts. The program code, when executed in a computer system, causes the computer system to perform the methods of embodiments of the present disclosure.

The above-described functions defined in the system/apparatus of the embodiments of the present disclosure are performed when the computer program is executed by the processor 901. The systems, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted, distributed, and downloaded and installed in the form of a signal on a network medium, via communication portion 909, and/or installed from removable medium 911. The computer program may include program code that may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 909 and/or installed from the removable medium 911. The above-described functions defined in the system of the embodiments of the present disclosure are performed when the computer program is executed by the processor 901. The systems, devices, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

According to embodiments of the present disclosure, program code for performing computer programs provided by embodiments of the present disclosure may be written in any combination of one or more programming languages, and in particular, such computer programs may be implemented in high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. Programming languages include, but are not limited to, such as Java, c++, python, "C" or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).

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

Those skilled in the art will appreciate that the features recited in the various embodiments of the disclosure and/or in the claims may be combined in various combinations and/or combinations, even if such combinations or combinations are not explicitly recited in the disclosure. In particular, the features recited in the various embodiments of the present disclosure and/or the claims may be variously combined and/or combined without departing from the spirit and teachings of the present disclosure. All such combinations and/or combinations fall within the scope of the present disclosure.

The embodiments of the present disclosure are described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the embodiments cannot be used advantageously in combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be made by those skilled in the art without departing from the scope of the disclosure, and such alternatives and modifications are intended to fall within the scope of the disclosure.

Claims

1. A method of product recommendation, comprising:

obtaining product information corresponding to m products one by one and interaction information of users and the products, wherein the interaction information of the users on the products comprises purchase information of the users on the products and browsing information of the users on the products, and m is an integer greater than or equal to 1;

Forming a label set of a product corresponding to the product information according to each piece of product information, wherein the label set comprises at least one label;

calculating the influence degree value of each label in each label set on the product corresponding to the label set according to the label sets of the m products;

calculating the interest value of the user on each of g overlapped labels according to the interaction information of the user and the product, the label set of the product and the preset user set interest labels, wherein g overlapped labels exist in the set interest labels, the labels of the product purchased by the user and the labels of the product browsed by the user, and g is an integer greater than or equal to 1;

calculating the predicted value of the user's interest degree for each product in the m products according to the influence degree value of each label in each label set on the product corresponding to the label set and the interest degree value of the user on each overlapping label in the g overlapping labels; and

and recommending products to the user according to the ranking of the interestingness predictive value.

2. The method according to claim 1, wherein the product information includes a product specification, and the forming a tag set of a product corresponding to the product information according to each product information includes:

Performing word segmentation processing on the product specifications of each product to obtain word segmentation results;

screening the word segmentation result to obtain a first sub-tag set; and

and taking the first sub-label set as a label set of the product.

3. The method of claim 2, wherein the screening the word segmentation result to obtain a first sub-tag set includes:

constructing a word co-occurrence map according to the word segmentation results, wherein the word segmentation results are used as nodes of the word co-occurrence map, and the co-occurrence relationship between the word segmentation results is used as an edge of the word co-occurrence map;

iteratively spreading the weight of the node according to the word co-occurrence map until convergence;

taking the weight of the node during convergence as an importance value of a word segmentation result corresponding to the node; and

and selecting word segmentation results according to the ranking of the importance values to form the first sub-tag set.

4. The method according to claim 2, wherein the product information further includes product attribute and/or product management information, and the forming a tag set of a product corresponding to the product information according to each product information includes:

Taking the product attribute of each product as a second sub-label set of the product; and/or

Responding to the label of the business personnel to the product according to the product management information, and taking the label as a third sub-label set of the product; and

and taking the second sub-label set and/or the third sub-label set as the label set of the product.

5. The method according to claim 1, wherein calculating, from the tag sets of the m products, a value of influence of each tag in each of the tag sets on a product corresponding to the tag set includes:

operation S51, determining, as a first frequency, an occurrence frequency of a first target tag in each of the tag sets in the tag set;

operation S52 of determining, as a second frequency, a frequency at which the first target tag appears in the tag set of the m products;

operation S53, calculating, according to the first frequency and the second frequency, a value of influence of the first target tag on a product corresponding to the tag set; and

and (S54) taking each label in the label set as the first target label in sequence, and executing the operations (S51-S53) to obtain the influence degree value of each label in the label set on the product corresponding to the label set.

6. The method according to claim 1, wherein calculating the interest value of the user for each of the g overlay tags according to the interaction information of the user with the product, the tag set of the product and the preset interest tag set of the user comprises:

determining a first sub-interest degree of a user on each of g overlapped labels of the purchased product according to the purchase information of the user on the product and the label set of the product;

determining a second sub-interest degree of the user on each of g overlapped labels of the browsed product according to the browsing information of the user on the product and the label set of the product;

presetting a third sub-interest degree of a user on each of the g overlapped labels; and

and calculating the interest degree value of the user for each of the g coincident labels according to the first sub-interest degree, the second sub-interest degree and the third sub-interest degree of the user for each of the g coincident labels.

7. The method of claim 6, wherein determining a first sub-interest level of the user for each of g overlay labels of the purchased product based on the user's purchase information for the product and the set of labels of the product comprises:

Operation S61, determining a first interestingness score of the user for the i-th purchased product;

operation S62 of determining the number of times that the second target tag is included in the products purchased all times by the user;

operation S63, calculating a first sub-interest degree of the user on the second target label according to the first interest degree score and the times of including the second target label in the products purchased by the user all times; and

8. The method of claim 6, wherein determining the second sub-interest level of the user for each of the g overlay labels of the browsed product based on the browsing information of the user for the product and the label set of the product comprises:

in operation S71, a second interestingness score of the user for the j-th browsed product is determined according to the total browsing times and each browsing time of the user for the j-th browsed product.

Operation S72, determining the number of times that the second target tag is included in the product browsed by the user for all times;

Operation S73, calculating a second sub-interest degree of the user on the second target label according to the second interest degree score and the times of including the second target label in the products browsed by the user for all times; and

and S74, taking each of the g overlapped labels of the product browsed by the user as a second target label in sequence, and executing operations S71-S73 to obtain a second sub-interest degree of the user on each of the g overlapped labels of the product browsed.

9. The method according to claim 1, wherein calculating the predicted value of interest of the user for each of the m products according to the value of influence of each of the tags in each of the tag sets on the products corresponding to the tag set and the value of interest of the user for each of the g tags, comprises:

according to the interest degree value of the user on each of the g overlapped labels, k interesting labels which are interested by the user are determined according to the interest degree value sequence of the g overlapped labels, wherein k is an integer which is more than or equal to 1 and less than or equal to g;

determining the same number of the labels in the k interest labels and the label set of each product; and

Calculating the predicted value of the user's interest degree for each product in the m products according to the number of the k interest tags, the number of the tags in the tag set of each product, the same number of the tags, the influence degree value of each tag in the tag set of each product on the product and the interest degree value of the user on each overlapping tag in the g overlapping tags.

10. A product recommendation device, comprising:

the system comprises an acquisition module, a storage module and a display module, wherein the acquisition module is used for executing acquisition of product information corresponding to m products one by one and interaction information of users and the products, wherein the interaction information of the users on the products comprises purchase information of the users on the products and browsing information of the users on the products, and m is an integer greater than or equal to 1;

a forming module, configured to perform forming a tag set of a product corresponding to each piece of product information according to each piece of product information, where the tag set includes at least one tag;

the first calculation module is used for executing label sets of the m products, and calculating the influence degree value of each label in each label set on the product corresponding to the label set;

The second calculation module is used for executing calculation of interest values of the user on each of g overlapped labels according to the interaction information of the user and the product, the label set of the product and preset user set interest labels, wherein g overlapped labels exist in the set interest labels, the labels of the product purchased by the user and the labels of the product browsed by the user, and g is an integer greater than or equal to 1;

the third calculation module is used for executing calculation of an interest degree predicted value of a user for each product in the m products according to the influence degree value of each tag in each tag set for the product corresponding to the tag set and the interest degree value of the user for each of the g overlapped tags; and

and the recommending module is used for executing ranking according to the interestingness predictive value and recommending products to the user.

11. An electronic device, comprising:

one or more processors;

one or more memories for storing executable instructions which, when executed by the processor, implement the method of any of claims 1-9.

12. A computer readable storage medium, characterized in that the storage medium has stored thereon executable instructions which, when executed by a processor, implement the method according to any of claims 1-9.

13. A computer program product comprising a computer program comprising one or more executable instructions which when executed by a processor implement the method according to any one of claims 1 to 9.