CN113761393A

CN113761393A - Commodity collaborative recommendation method and device, equipment, medium and product thereof

Info

Publication number: CN113761393A
Application number: CN202111131828.7A
Authority: CN
Inventors: 黄丕帅
Original assignee: Guangzhou Huaduo Network Technology Co Ltd
Current assignee: Guangzhou Huaduo Network Technology Co Ltd
Priority date: 2021-09-26
Filing date: 2021-09-26
Publication date: 2021-12-07

Abstract

The application discloses a commodity collaborative recommendation method and a device, equipment, a medium and a product thereof, wherein the method comprises the following steps: acquiring interactive behavior data of a plurality of behavior types corresponding to the commodity objects in the commodity database, wherein at least one behavior type has a business logic precedence relationship with other behavior types; counting the total amount of the users corresponding to each behavior type between every two commodity objects; calculating similarity data between every two commodity objects, wherein the similarity data is a sum of similarity determined by calculation according to the total amount of users sharing each behavior type of the corresponding two commodity objects; and responding to the commodity matching instruction, and inquiring and determining a plurality of corresponding similar commodity objects for the target commodity object specified by the instruction according to the similarity data. The matching degree of the similar commodity objects recommended to the user can be improved, the real requirements of the user can be matched better, and the similar commodity objects can obtain higher click rate.

Description

Commodity collaborative recommendation method and device, equipment, medium and product thereof

Technical Field

The present application relates to the field of e-commerce information technologies, and in particular, to a method and a device for collaborative recommendation of a commodity, a computer device, a computer-readable storage medium, and a computer program product.

Background

The e-commerce recommendation system industry is mainly divided into two stages of recall and sorting, and in the recall stage, a multi-recall scheme is often used as a main point. The collaborative filtering is used as the most classical recall scheme, and the user ID and the commodity ID are used for interactive behavior for recommendation, so that the personalized recall of the user can be well reflected.

However, in the conventional collaborative filtering, similar commodity recall is performed only by using an ID feature, usually commodity portrayal or user portrayal is performed based on an ID, and commodity matching is realized according to portrait tags, in this case, development and utilization of specific behavior information of a user are neglected, while in practice, different interactive behaviors of the user and the commodity express different intentions for an article, for example, one user clicks on a certain commodity, and a more suitable scenario is to recommend the similar commodity to the user, and if the user purchases the commodity, the related commodity of the commodity is more suitable to be recommended to the user. Accordingly, the applicant thinks that the interactive behaviors of the user such as clicking, purchase adding, order placing, payment and the like have different meanings, and can further perform data mining on the interactive behaviors and perform secondary utilization on the mined information.

In view of this, the technologies related to commodity selection in e-commerce platforms still have a space to be excavated, and the applicant has long concentrated on research and development in related fields, and has made corresponding researches on the technologies.

Disclosure of Invention

A primary object of the present application is to solve at least one of the above problems and provide a product collaborative recommendation method and a corresponding apparatus, computer device, computer readable storage medium, and computer program product.

In order to meet various purposes of the application, the following technical scheme is adopted in the application:

the commodity collaborative recommendation method adaptive to one of the purposes of the application comprises the following steps:

acquiring interactive behavior data of a plurality of behavior types corresponding to the commodity objects in the commodity database, wherein at least one behavior type has a business logic precedence relationship with other behavior types;

counting the total amount of the common users corresponding to each behavior type between every two commodity objects according to the number of the same users implementing the same behavior type to generate the same type of interactive behavior data;

calculating similarity data between every two commodity objects, wherein the similarity data is a sum of similarity determined by calculation according to the total amount of users sharing each behavior type of the corresponding two commodity objects;

and responding to the commodity matching instruction, and inquiring and determining a plurality of corresponding similar commodity objects for the target commodity object specified by the instruction according to the similarity data.

In a further embodiment, the step of obtaining interactive behavior data of a plurality of behavior types corresponding to the commodity object in the commodity database includes the following steps:

acquiring user historical behavior data corresponding to commodity objects in a commodity database;

and performing data cleaning on the historical behavior data of the user, determining interactive behavior data of a plurality of behavior types corresponding to each commodity object, wherein the interactive behavior data correspondingly comprises click behavior data acting on the commodity object and at least any one of the following items: the shopping behavior data of the commodity object added to the shopping cart, the ordering behavior data of the commodity object and the payment behavior data of the order of the commodity object.

In a further embodiment, the method for calculating the similarity data between two commodity objects comprises the following steps:

determining similarity between two commodity objects about the behavior type based on each behavior type;

applying a preset weight matching scheme, and carrying out weighted summation on the similarity of a plurality of common behavior types of every two commodity objects to obtain similarity data between every two commodity objects;

constructing a similarity matrix for storing the similarity data, wherein each element is used for storing the similarity data between the commodity object pointed by the row coordinate and the commodity object pointed by the column coordinate;

and normalizing the similarity data in the similarity matrix.

In a preferred embodiment, in the step of determining the similarity between each two commodity objects with respect to the behavior type based on each behavior type, the similarity calculation is performed by using any one of the following algorithms: cosine similarity calculation, Euclidean distance algorithm, Pearson correlation coefficient algorithm, Jacard algorithm, log-likelihood algorithm, and Manhattan distance algorithm.

In a preferred embodiment, in the step of applying the preset weight matching scheme, the weight matching scheme is configured to match weights from small to large in the following order: click behavior data, purchase adding behavior data, order placing behavior data and payment behavior data.

In a further embodiment, in response to a commodity matching instruction, querying and determining a plurality of corresponding similar commodity objects for a target commodity object specified by the instruction according to the similarity data, the method includes the following steps:

finding out similarity data mapped to all the commodity objects with the target commodity object;

performing reverse sorting according to the similarity data;

selecting a set number of similar commodity objects with the maximum similarity data from the sequencing results;

and pushing the similar commodity object to a user triggering the commodity matching instruction.

The utility model provides a commodity collaborative recommendation device that adapts one of this application's purpose, includes: the system comprises a data acquisition module, a total amount counting module, a similarity calculation module and a response recommendation module, wherein the data acquisition module is used for acquiring interactive behavior data of a plurality of behavior types corresponding to commodity objects in a commodity database, and at least one behavior type has a business logic precedence relationship with other behavior types; the total amount counting module is used for counting the total amount of the common users corresponding to each behavior type between every two commodity objects according to the same user number of the same behavior types and the same interactive behavior data generated by implementing the same behavior types; the similarity calculation module is used for calculating similarity data between every two commodity objects, and the similarity data is a sum of similarity determined by calculation according to the total amount of users sharing each behavior type of the corresponding two commodity objects; and the response recommending module is used for responding to the commodity matching instruction and inquiring and determining a plurality of corresponding similar commodity objects for the target commodity object specified by the instruction according to the similarity data.

In a further embodiment, the data acquisition module comprises: the historical data acquisition submodule is used for acquiring user historical behavior data corresponding to the commodity object in the commodity database; the data cleaning submodule is used for performing data cleaning on the historical behavior data of the user and determining interactive behavior data of a plurality of behavior types corresponding to each commodity object, and the interactive behavior data correspondingly comprises click behavior data acting on the commodity object and at least any one of the following items: the shopping behavior data of the commodity object added to the shopping cart, the ordering behavior data of the commodity object and the payment behavior data of the order of the commodity object.

In a further embodiment, the similarity calculation module comprises: the behavior calculation submodule is used for determining the similarity of the behavior types between every two commodity objects based on the behavior types; the weighting calculation sub-module is used for applying a preset weight matching scheme and carrying out weighted summation on the similarity of a plurality of common behavior types of every two commodity objects to obtain similarity data between every two commodity objects; the matrix construction submodule is used for constructing a similarity matrix used for storing the similarity data, wherein each element is used for storing the similarity data between the commodity object pointed by the row coordinate and the commodity object pointed by the column coordinate; and the matrix normalization submodule is used for normalizing the similarity data in the similarity matrix.

In a preferred embodiment, the behavior calculation sub-module is configured to perform the similarity calculation using any one of the following algorithms: cosine similarity calculation, Euclidean distance algorithm, Pearson correlation coefficient algorithm, Jacard algorithm, log-likelihood algorithm, and Manhattan distance algorithm.

In a preferred embodiment, in the weight calculation sub-module, the weight matching scheme is configured to match weights from small to large in the following order: click behavior data, purchase adding behavior data, order placing behavior data and payment behavior data.

In a further embodiment, the merchandise recommendation module comprises: the similarity query submodule is used for searching similarity data mapped to all the commodity objects with the target commodity object; the result sorting submodule is used for performing reverse sorting according to the similarity data; the similarity selection submodule is used for selecting a set number of similar commodity objects with the maximum similarity data from the sequencing result; and the similar recommending submodule is used for pushing the similar commodity object to the user triggering the commodity matching instruction.

The computer device comprises a central processing unit and a memory, wherein the central processing unit is used for calling and running a computer program stored in the memory to execute the steps of the commodity collaborative recommendation method.

A computer-readable storage medium storing a computer program implemented according to the product collaborative recommendation method in the form of computer-readable instructions, the computer program, when called by a computer, executing the steps included in the method.

A computer program product, provided to adapt to another object of the present application, comprises computer programs/instructions which, when executed by a processor, implement the steps of the method described in any of the embodiments of the present application.

Compared with the prior art, the application has the following advantages:

the method comprises the steps of mining data of user behavior data to the level of specific behavior types, counting the total amount of users between every two commodity objects according to different behavior types, wherein the total amount of users represents the total amount of users who are subjected to interactive behaviors of the same behavior type between the two commodity objects, further counting the similarity between every two commodity objects according to the total amount of users for each behavior type, finally summarizing the similarities of a plurality of behavior types between every two commodity objects to obtain the similarity data between every two commodity objects, macroscopically inheriting logic association information between each specific behavior type, reflecting deep association between every two commodity objects, matching similar commodity objects for one commodity object by using the similarity data on the basis, and further matching user demand logic, the method and the system enable the user to reach the commodity corresponding to the real demand more easily, and further improve the click rate and the transaction rate of the recommended similar commodity objects.

Drawings

The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

fig. 1 is a schematic flowchart of an exemplary embodiment of a collaborative commodity recommendation method according to the present application;

FIG. 2 is a flowchart illustrating a process of obtaining interactive behavior data according to an embodiment of the present application;

FIG. 3 is a flowchart illustrating a process of calculating similarity data according to an embodiment of the present application;

FIG. 4 is a flowchart illustrating a process of querying similar merchandise objects according to an embodiment of the present application;

FIG. 5 is a schematic block diagram of a collaborative commodity recommendation device according to the present application;

fig. 6 is a schematic structural diagram of a computer device used in the present application.

Detailed Description

Reference will now be made in detail to embodiments of the present application, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are exemplary only for the purpose of explaining the present application and are not to be construed as limiting the present application.

As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or wirelessly coupled. As used herein, the term "and/or" includes all or any element and all combinations of one or more of the associated listed items.

It will be understood by those within the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

As will be appreciated by those skilled in the art, "client," "terminal," and "terminal device" as used herein include both devices that are wireless signal receivers, which are devices having only wireless signal receivers without transmit capability, and devices that are receive and transmit hardware, which have receive and transmit hardware capable of two-way communication over a two-way communication link. Such a device may include: cellular or other communication devices such as personal computers, tablets, etc. having single or multi-line displays or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service), which may combine voice, data processing, facsimile and/or data communication capabilities; a PDA (Personal Digital Assistant), which may include a radio frequency receiver, a pager, internet/intranet access, a web browser, a notepad, a calendar and/or a GPS (Global Positioning System) receiver; a conventional laptop and/or palmtop computer or other device having and/or including a radio frequency receiver. As used herein, a "client," "terminal device" can be portable, transportable, installed in a vehicle (aeronautical, maritime, and/or land-based), or situated and/or configured to operate locally and/or in a distributed fashion at any other location(s) on earth and/or in space. The "client", "terminal Device" used herein may also be a communication terminal, a web terminal, a music/video playing terminal, such as a PDA, an MID (Mobile Internet Device) and/or a Mobile phone with music/video playing function, and may also be a smart tv, a set-top box, and the like.

The hardware referred to by the names "server", "client", "service node", etc. is essentially an electronic device with the performance of a personal computer, and is a hardware device having necessary components disclosed by the von neumann principle such as a central processing unit (including an arithmetic unit and a controller), a memory, an input device, an output device, etc., a computer program is stored in the memory, and the central processing unit calls a program stored in an external memory into the internal memory to run, executes instructions in the program, and interacts with the input and output devices, thereby completing a specific function.

It should be noted that the concept of "server" as referred to in this application can be extended to the case of a server cluster. According to the network deployment principle understood by those skilled in the art, the servers should be logically divided, and in physical space, the servers may be independent from each other but can be called through an interface, or may be integrated into one physical computer or a set of computer clusters. Those skilled in the art will appreciate this variation and should not be so limited as to restrict the implementation of the network deployment of the present application.

One or more technical features of the present application, unless expressly specified otherwise, may be deployed to a server for implementation by a client remotely invoking an online service interface provided by a capture server for access, or may be deployed directly and run on the client for access.

Unless specified in clear text, the neural network model referred to or possibly referred to in the application can be deployed in a remote server and used for remote call at a client, and can also be deployed in a client with qualified equipment capability for direct call.

Various data referred to in the present application may be stored in a server remotely or in a local terminal device unless specified in the clear text, as long as the data is suitable for being called by the technical solution of the present application.

The person skilled in the art will know this: although the various methods of the present application are described based on the same concept so as to be common to each other, they may be independently performed unless otherwise specified. In the same way, for each embodiment disclosed in the present application, it is proposed based on the same inventive concept, and therefore, concepts of the same expression and concepts of which expressions are different but are appropriately changed only for convenience should be equally understood.

The embodiments to be disclosed herein can be flexibly constructed by cross-linking related technical features of the embodiments unless the mutual exclusion relationship between the related technical features is stated in the clear text, as long as the combination does not depart from the inventive spirit of the present application and can meet the needs of the prior art or solve the deficiencies of the prior art. Those skilled in the art will appreciate variations therefrom.

The commodity collaborative recommendation method can be programmed into a computer program product, is deployed in a client or a server to run, and is generally deployed in the server to implement, for example, in an e-commerce platform application scenario of the present application, so that the method can be executed by accessing an open interface after the computer program product runs and performing human-computer interaction with a process of the computer program product through a graphical user interface.

Referring to fig. 1, in an exemplary embodiment of a collaborative commodity recommendation method according to the present application, the collaborative commodity recommendation method includes the following steps:

step S1100, acquiring interactive behavior data of a plurality of behavior types corresponding to the commodity object in the commodity database, wherein at least one behavior type has a business logic precedence relationship with other behavior types:

taking the cross-border e-commerce platform based on the independent station as an example, each merchant instance has a commodity database for storing commodity information of the commodity object which is published online by the merchant instance, wherein the commodity information includes but is not limited to different types of information such as title texts, classification labels, attribute data, detailed texts, pictures, videos and the like corresponding to the commodity object. When a user accesses a commodity object in the independent station, the operation of a preset embedded point code in a corresponding page or an application program of terminal equipment where the user is located is triggered to submit a user behavior message to the independent station, and the behavior type implemented by the user and the incidence relation information of the data object acted by the user are submitted to a server of the independent station, so that the server can obtain corresponding interactive behavior data. For example, when a user clicks a commodity object to enter a detail page of the commodity object, the point-embedding code can acquire a corresponding click behavior and an acted commodity object, organize the commodity object into association relation information, package the association relation information into interaction behavior data, and provide the interaction behavior data for the server to analyze and acquire the interaction behavior data. Similarly, when the user further clicks the purchase control on the detail page to place an order, the corresponding embedded point code can also identify the order placing behavior of the user and the commodity object to be placed, and the corresponding embedded point code is constructed into corresponding interaction behavior data to be submitted to the server. Other types of behavior, such as a purchase-adding behavior in which the user adds the purchase amount of the commodity object, a payment behavior in which the user executes an order for a certain commodity object, and the like, may also be automatically generated in the background.

In another embodiment, the server may also push back the behavior type triggered by the user according to the specific page when the user clicks the link to pull the corresponding page, thereby automatically determining the behavior type corresponding to the user and the corresponding data object thereof, and generating interactive behavior data corresponding to the incidence relation information representing the behavior type and the data object acted by the behavior type. For example, when a user clicks a commodity object, the user pulls a detail page of the commodity object, and the server can confirm the click behavior of the user acting on the commodity object according to the detail page, so as to obtain corresponding interaction behavior data. Similarly, when the user further clicks the purchase control of the commodity object, the background generates corresponding interactive behavior data acting on the ordering behavior of the commodity object. Other types of behavior, such as a purchase-adding behavior in which the user adds the purchase amount of the commodity object, a payment behavior in which the user executes an order for a certain commodity object, and the like, may also be automatically generated in the background.

The types of actions involved in the user's implementation of the interaction action on the commodity object include, but are not limited to, the following: click behavior type, order placing behavior type, purchase adding behavior type and payment behavior type. The click behavior type mainly refers to a behavior type that a user clicks a commodity object to enter a detail page of the commodity object so as to place an order, the order placing behavior type refers to a behavior type that the user triggers an order placing logic of the commodity object, the purchase adding behavior type refers to a behavior type that the user adds the commodity object to a shopping cart, and the payment behavior type refers to a payment behavior type that the user successfully executes payment corresponding to an order created by the commodity object.

It is easy to see that the user behavior events corresponding to each behavior type form a complete business chain related to the purchasing behavior of the commodity object, for example, when a user wants to purchase a commodity, the user firstly clicks the corresponding commodity object on the page to enter the detailed page of the commodity object, so as to trigger the generation of interactive behavior data of the clicking behavior type, namely clicking behavior data; then adding the data into a shopping cart, thereby triggering and generating interactive behavior data of the shopping behavior type, namely the shopping behavior data; then, orders can be directly placed in the shopping cart to create orders corresponding to the commodity objects, and interactive behavior data corresponding to the order placing behavior types, namely order placing behavior data, is triggered and generated; and finally, the user executes payment operation on the corresponding bill, and correspondingly triggers and generates interactive behavior data of the payment behavior type, namely the payment behavior data.

For the business chain, according to the common business logic of the e-commerce platform, the user is usually allowed to stride over the link of adding to the shopping cart and directly execute the ordering action on the detail page; the user may not perform the ordering action and the payment action for a long time although the commodity object is added to the shopping cart. Or even if ordering is performed, payment may not be performed for a long period of time. That is, although the present application considers various behavior types involved in the business logic corresponding to the whole user purchasing process, that is, the interactive behavior data corresponding to all the behavior types involved in the whole chain of the business process purchased by the user is preferably obtained for each commodity object, this application does not constitute a limitation to the implementable embodiment of the present application, and the present application allows the implementation by using the interactive behavior data corresponding to some behavior types in the whole business chain, for example, one of the click behavior type and other behavior types may be used. In summary, the present application is intended to collect, for each commodity object, interaction behavior data corresponding to at least two behavior types implemented by a user for the commodity object as basic data required for data mining. Since the click behavior type corresponds to the entry of the commodity object, that is, the commodity object can be clicked on the first interface to enter the detail page thereof so as to implement the subsequent ordering process, in the exemplary embodiment, the acquired interaction behavior data at least includes the click behavior type and any one of the following behavior types: the purchase adding behavior type, the order placing behavior type and the payment behavior type. However, some alternative embodiments are possible, such as only including both the ordering action type and the payment action type. Or only three types of click behavior type, order placing behavior type and payment behavior type can be included. In addition, other behavior types can be expanded according to the requirement of the data mining depth, for example, sharing behavior types of the user sharing the commodity object to other users, and corresponding interactive behavior data can be obtained to be used for data mining of the application.

Therefore, it can be generally understood that the interaction behavior data adopted in the present application relates to behavior types having precedence relationships in business logic, including directly successive relationships that occur immediately, and also including indirectly successive relationships that do not occur immediately. Therefore, the method and the device can utilize the precedence relationship to dig out the logic information and the user intention decision information corresponding to the user behaviors so as to determine the similarity matrix according to the precedence relationship in the following and serve for commodity recommendation.

It should be noted that although the application emphasizes that there is a sequential relationship in business logic between multiple behavior types, in a preferred case, it may preferentially adopt the interactive behavior information corresponding to all users of the specified behavior types within the complete business logic implemented for the commodity object, but it should not be understood that the reverse narrowing of the interactive behavior information is that a single user must implement all the behavior types in business logic for a specific commodity object, because the object targeted by the application is mainly the commodity object rather than the user itself during data mining, and therefore, as long as the interactive behavior data corresponding to the commodity object includes such a sequential relationship in behavior types, the interactive behavior data macroscopically represents the associated information of the commodity object in the purchasing process on a big data level, and can be mined and utilized by the application.

The interactive behavior data adopted by the server are accumulated and stored in the background for a long time, and the interactive behavior data are called out so as to start data mining by using a statistical means. Generally, a historical time period is given when data acquisition is carried out, so that an interactive behavior data set belonging to the range of the historical time period is screened out for each commodity object, and the historical time period is generally a time period for carrying out backtracking of the current application on the same day. For example, if the historical time period is set to three months, it means that only three months are screened from the date of carrying out the present application and back to obtain the interactive behavior data generated during the three months. Of course, this historical period is flexibly set by those skilled in the art, depending on the need for information aging. The historical time period may be one month, half month, one week, etc., and may be set as desired.

Step S1200, according to the same user quantity of the same type of interactive behavior data generated by implementing the same behavior type, counting the total quantity of the common users corresponding to each behavior type between every two commodity objects:

and after the interactive behavior data set of the historical time period is obtained, starting statistical work, and counting the total amount of users between every two commodity objects corresponding to each behavior type aiming at every two commodity objects.

The interactive behavior data associated with each commodity object is derived from a plurality of users, and is irrelevant to the behavior types implemented by a single user in the stage of counting the total number of the users, so that whether the user completely implements all the behavior types of the service chain formed by the whole service logic is generally not required to be considered unless the requirements of individual embodiments are changed.

Specifically, when two commodity objects are respectively subjected to the interactive behaviors of the same behavior type by the same user, namely interactive behavior data corresponding to the same interactive behaviors exist between the two commodity objects, 1 is accumulated for the total number of the common users of the same behavior type, and finally, the total number of the common users corresponding to all the behavior types is obtained through statistics, wherein the total number of the common users corresponding to each behavior type represents how many users with the same identity perform the same corresponding interactive behaviors by each of the two commodity objects in a statistical sense, for example, the first commodity object of the two commodity objects is subjected to the same corresponding interactive behaviors by N_f,iEach user implements X₁A secondary click interactive behavior, wherein f represents a specific interactive behavior, namely a click interactive behavior, and i represents a first commodity object; a second commodity object, which is represented by N_f,jEach user implements X₂A secondary click interaction behavior, wherein j represents a second merchandise object. In this case, the intersection N of the two user groups is obtained_f,i∩N_f，jTo obtain the total amount N of the same user_fThe total amount of users between the two commodity objects corresponding to the click behavior event is obtained. In which, X is not considered in the statistical process₁And X₂These two data, i.e. the multi-click actions performed by the same user for the same commodity object, are essentially treated as equivalent to one-click actions. The same holds true for the statistics of the total number of common users corresponding to other behavior types. Broad, commercial productsAll commodity objects in the database can obtain the total amount of users corresponding to each behavior type between every two commodity objects.

Therefore, the total amount of the common users establishes the associated information of the two commodity objects about the same interactive behaviors of the same user group, and the interactive behaviors implemented by the users aiming at the same commodity object may cover a plurality of behavior types, so that the associated information of the interactive behaviors across the behavior types in the service logic chain is hidden between the total amount of the common users of different commodity object assemblies, and therefore, the efficient representation of the associated information on the service logic hidden by the interactive behavior data generated by the users is realized.

The total amount of users N corresponding to each behavior type in two commodity objects_fIs shown as having N_fThe two commodity objects are subjected to the same behavior type of interactive behaviors by the common users, so that the total amount of the common users between different behavior types may show differences. For example, the total number of users corresponding to the click behavior type is C_nThe total amount of the common users corresponding to the purchase behavior types is A_nThe total amount of the common users corresponding to the order placing behavior type is O_nThe total amount of the users corresponding to the payment behavior type is P_nCombining practical experience with simple logical reasoning, in general, C_n>A_n>O_n>P_n. Of course, it is not absolute, if the user is not used to add the merchandise object to the shopping cart, but directly places an order for purchase, then A_nNot necessarily greater than O_nAnd P_n. Therefore, the total amount of all the users sharing the same commodity object pair really carries the logic relation information between different interactive behaviors of the same user on the E-commerce platform, and the association relation between various behavior types is macroscopically embodied.

For convenience of calculation, the total amount of the common users can be constructed into a matrix, so that subsequent matrix operation is facilitated, and the total amount of the common users of each type between every two commodity objects is directly called. The matrix can be a two-dimensional matrix or a three-dimensional matrix, taking the two-dimensional matrix as an example, a matrix can be established for each behavior type, the row coordinates and the column coordinates of the matrix refer to each commodity object in the commodity database in sequence, so that a square matrix is formed, each element is used for storing the total amount of users shared between two commodity objects pointed by the row coordinates and the column coordinates of the element respectively, and the total amount of users shared by the corresponding two commodity objects about the behavior type can be obtained by referring to the element of the two-dimensional matrix.

Step S1300, calculating similarity data between every two commodity objects, wherein the similarity data is a sum value of similarity determined by calculating according to the total amount of the common users of each behavior type of the corresponding two commodity objects:

the data mining of the associated information between every two commodity objects is realized in the process, the corresponding associated information is hidden in the total amount of users of every two commodity objects related to each behavior type, on the basis, the total amount of the users can be referred to calculate the similarity between every two commodity objects, the similarity data is obtained, so that the association between every two commodity objects on the business logic of purchasing the commodity objects by the users is further reflected through the similarity data, and other similar commodity objects are conveniently determined for a certain target commodity object by referring to the similarity data.

The way to calculate the similarity between the data can be implemented by a person skilled in the art by using various well-known algorithms, including but not limited to any of the following: cosine similarity calculation, Euclidean distance algorithm, Pearson correlation coefficient algorithm, Jacard algorithm, log-likelihood algorithm, and Manhattan distance algorithm. Certainly, because a plurality of vector matrixes corresponding to the behavior types exist, in the process of calculating by using the similarity algorithm, the common total amount of users between every two commodity objects of each behavior type is referred to calculate the similarity of the behavior types between the two commodity objects, then the similarities are summed to obtain a sum value, and the sum value is finally determined similarity data between the two commodity objects.

In order to refer to the similarity data, a two-dimensional matrix can be constructed in the same way, the row coordinates and the column coordinates of the matrix refer to each commodity object in the commodity database in sequence, so that a square matrix is formed, each element is used for storing the similarity data between two commodity objects respectively pointed by the row coordinates and the column coordinates of the element, and the similarity data of the corresponding two commodity objects can be obtained by referring to the elements of the two-dimensional matrix. It can be understood that, each row vector stores similarity data between one commodity object and all commodity objects in the commodity database, and when a similar commodity object of a certain commodity object is to be determined, similarity matching is performed from the row vector.

Step S1400, responding to the commodity matching instruction, and inquiring and determining a plurality of corresponding similar commodity objects for the target commodity object specified by the instruction according to the similarity data:

when the background server of the e-commerce platform needs to generate a list of similar commodity objects for each target commodity object according to the similarity data, a commodity matching instruction can be issued to realize the purpose. Or, the consumer user of the merchant instance may issue the commodity matching instruction to drive the server to determine similar commodity objects corresponding to the target commodity object that the user is accessing, and push the similar commodity objects to the user.

When the commodity matching instruction is issued, the background server responds to the commodity matching instruction, a row vector (or a column vector) corresponding to a target commodity object specified by the instruction is called in the two-dimensional matrix storing the similarity data, then a commodity object pointed by the column coordinate of each element is correspondingly obtained, the commodity objects are ranked according to the similarity data of each element and then preferentially selected as similar commodity objects, namely, the matching process of the similar commodity objects of the target commodity object specified by the instruction is finished, and finally the similar commodity objects can be pushed to related users.

It can be known from the disclosure of the exemplary embodiment that the present application excavates the data of user behavior data deep into the level of a specific behavior type, counts the total amount of users between two commodity objects according to different behavior types, the total amount of users representing the total amount of users having the same behavior type of interaction between the two commodity objects, further counts the similarity between the two commodity objects according to the total amount of users, and finally summarizes the similarities of a plurality of behavior types between the two commodity objects to obtain the similarity data between the two commodity objects, the similarity data macroscopically inherits the logic association information between each specific behavior type to embody the deep association between the two commodity objects, on the basis, the similarity data is used as a commodity object to match with a similar commodity object, the method and the system match with the user demand logic more, so that the user can directly reach the commodity corresponding to the real demand more easily, and the click rate and the transaction rate of the recommended similar commodity object can be further improved.

Referring to fig. 2, in a further embodiment, the step S1100 of obtaining the interactive behavior data of the plurality of behavior types corresponding to the commodity object in the commodity database includes the following steps:

step S1110, obtaining user historical behavior data corresponding to the commodity object in the commodity database:

as mentioned above, when the user is in the e-commerce platform, the corresponding interactive behavior data is triggered to be generated, and after the background server obtains the interactive behavior data, the interactive behavior data is stored in the corresponding database, so that the historical behavior data of the user is formed.

In order to serve the technical solution of the present application, the historical behavior data of the user needs to be called from the data, and the data belonging to the historical time period is usually filtered according to a given historical time period, for example, the historical behavior data of the historical time period corresponding to march, january, half month and week. The person skilled in the art is flexible in this respect.

Step S1120, performing data cleaning on the historical behavior data of the user, and determining interactive behavior data of a plurality of behavior types corresponding to each commodity object, where the interactive behavior data correspondingly includes click behavior data acting on the commodity object and at least any one of the following: the purchase behavior data of the commodity object added to the shopping cart, the order placing behavior data of the commodity object, and the payment behavior data of the order of the commodity object are as follows:

when the historical user behavior data exists, the historical user behavior data is stored after interactive behavior data are correspondingly generated for user behavior messages triggered by user behaviors each time, and the historical user behavior data are relatively discrete and diversified, so that data cleaning needs to be carried out on the historical user behavior data.

One of the means is to remove the related behavior data irrelevant to the data required by the application, and only keep the interactive behavior data relevant to the behavior type to be referred to by the application, wherein the interactive behavior data generally comprises user characteristic information, commodity objects, behavior types and mapping relation information among time, so that the technical means corresponding to each step of the application is conveniently applied.

Another means may be to cluster the interactive behavior data corresponding to the same user implementing the same behavior type for the same commodity object. As mentioned above, the present application is mainly for counting the total amount of users sharing the same behavior type between two commodity objects, and does not pay attention to how many times a single user applies the same behavior type of interaction on the same commodity object, so that for subsequent statistics, clustering can be performed in advance to obtain more purified data, thereby improving the operation efficiency in the subsequent statistics stage.

Other conventional data cleaning means can be flexibly applied by the technical personnel in the field as long as the realization of the technical scheme of the application is not influenced.

According to the embodiment, the data are further cleaned, so that the subsequent data mining efficiency is promoted, and the related associated information is obtained more efficiently.

Referring to fig. 3, in a further embodiment, the similarity data is preferably calculated by a cosine similarity algorithm, so that the step S1300 of calculating the similarity data between two commodity objects includes the following steps:

step S1310, determining similarity between two commodity objects with respect to each behavior type based on each behavior type:

as described above, the similarity between each two commodities is calculated for each behavior type. In this embodiment, when calculating the similarity formula, a cosine similarity calculation method is adopted, and the calculation process may be logically embodied as applying the following formula:

wherein f represents a behavior type, and the value ranges respectively correspond to: click behavior type, purchase adding behavior type, order placing behavior type and payment behavior type. Omega_fRepresenting the weight of the type of behavior. i and j represent commodity objects i and j, respectively.

According to the formula, the total amount of users N of two commodity objects under each behavior type condition_f,iAnd N_f，jAre intersected, therefore, N_f，i∩N_f，jI.e. the total amount of users shared between the two merchandise objects. Since the total amount of the common users corresponding to each behavior type between two commodity objects has been calculated in a previous step, for example, the aforementioned matrix for storing the total amount of the common users, the total amount of the common users can be directly obtained from the previously calculated data.

It can be seen from the formula that for two commodity objects, the similarity needs to be calculated corresponding to each behavior type.

Step S1320, applying a preset weight matching scheme, carrying out weighted summation on the similarity of a plurality of common behavior types of every two commodity objects, and obtaining similarity data between every two commodity objects:

considering that there is a difference in information value between each specific behavior type, for example, the payment behavior type has a higher value for representing that the user has made a trade with a commodity object than the click behavior type, according to the above formula, in the process of calculating the similarity data, the similarity matching weight ω for each behavior type is determined according to the similarity matching weight ω_fWeighting is carried out to correspond toAs final similarity data sim between two commodity objects_i,j。

The method comprises the following steps of distributing weights among different behavior types to form a weight matching scheme, wherein the weight matching scheme can be preset and called when similarity data need to be calculated. The weight matching scheme can also be automatically adjusted according to recall data or ranking list information referring to other commodity objects, and therefore, the method can be flexibly implemented by a person skilled in the art. In consideration of the relationship between information values between the respective behavior types, the present embodiment recommends that the weight matching scheme be configured to match weights from small to large in the following order: click behavior data, purchase adding behavior data, order placing behavior data and payment behavior data. It can be understood that the weight matching scheme reflects the information value corresponding to each behavior type, so that the similarity data can better reflect the association degree of possible associated purchase by the user between two commodity objects.

Step S1330, constructing a similarity matrix for storing the similarity data, wherein each element is used for storing the similarity data between the commodity object pointed by the row coordinate and the commodity object pointed by the column coordinate:

as previously mentioned, to facilitate reference to the similarity data, a two-dimensional matrix [ i: [ j: sim ] is constructed_i,j,k:sim_i,k]…]The row coordinates and the column coordinates of the matrix refer to each commodity object in the commodity database in sequence, so that a square matrix is formed, each element is used for storing similarity data between two commodity objects respectively pointed by the row coordinates and the column coordinates of the matrix, and the similarity data of the two corresponding commodity objects can be obtained by referring to the elements of the two-dimensional matrix. It can be understood that, each row vector stores similarity data between one commodity object and all commodity objects in the commodity database, and when a similar commodity object of a certain commodity object is to be determined, similarity matching is performed from the row vector.

Step S1340, normalizing the similarity data in the similarity matrix:

when the similarity data are calculated, the similarity data of each element are not generally calculated based on the same statistical scale, and therefore the relative relationship between the similarity data and each other is not unified, so that the two-dimensional matrix storing the similarity data can be normalized by applying the following formula:

wherein, sim_i,jFor similarity data of commodity object i and commodity object j, n indicates different elements, Σ sim, varying along the column coordinates_i,nRepresenting the addition of similarity data of each element in the row vector of commodity object i, scale _ sim_i,jNamely the similarity data after the commodity object i and the commodity object j are normalized. So far, in the whole two-dimensional matrix, the similarity data of each row vector is unified to the same statistical scale and can be directly referred.

Referring to fig. 4, in a further embodiment, the step S1400, in response to a commodity matching instruction, determining a plurality of corresponding similar commodity objects for the target commodity object specified by the instruction according to the similarity data by querying, includes the following steps:

step S1410, finding out similarity data mapped to all the commodity objects with the target commodity object:

when a commodity object similar to a target commodity object needs to be determined, all the commodity objects having similarity data with the target commodity object are determined, and are specifically included in the two-dimensional matrix called in the foregoing embodiments, and the row vector corresponding to the target commodity object is extracted. As previously mentioned, the row vector stores similarity data mapping the target merchandise object to all merchandise objects in the merchandise database.

Step S1420, performing reverse sorting according to the similarity data:

in order to extract part of similar commodity objects, all similar commodity objects mapped by the row vector may be sorted inversely according to the similarity data, so that the similar commodity objects are arranged from large to small according to the similarity between the similar commodity objects and the target commodity object.

Step S1430, selecting a set number of similar commodity objects with the largest similarity data from the sorting results:

the number of similar commodity objects screened out is generally predefined to avoid producing too many similar commodity objects, so the Top _ K algorithm can be adopted to preset the value of K, and the corresponding K similar commodity objects ranked in the front are selected from the row vector to form a similar commodity object list.

Step S1440, pushing the similar commodity object to a user who triggers the commodity matching instruction:

when the commodity matching instruction is triggered by a certain consuming user, for example, the user enters a page of 'guessing you like', and then the server parses the page into similar commodity objects acquired according to a certain target commodity object which is just visited by the user to trigger the commodity matching instruction, in this case, after the server acquires the similar commodity objects of the target commodity object, a similar commodity object list formed by the similar commodity objects can be pushed to the user to serve as a response to the commodity matching instruction.

In this embodiment, by providing a business logic for searching similar goods for a user, the similarity data calculated in the present application is more practical, and similar goods objects serving for the user are matched, which is not difficult to understand, because in the present application, relevant information about a purchase business process corresponding to each goods object has been mined in advance, after the user receives the similar goods objects, the user can hit the similar goods objects with a higher probability, and even trigger further purchasing behavior.

Referring to fig. 5, a collaborative commodity recommendation apparatus adapted to one of the purposes of the present application is a functional implementation of the collaborative commodity recommendation method of the present application, and the apparatus includes: the system comprises a data acquisition module 1100, a total amount statistics module 1200, a similarity calculation module 1300 and a response recommendation module 1400, wherein the data acquisition module 1100 is used for acquiring interactive behavior data of a plurality of behavior types corresponding to a commodity object in a commodity database, wherein at least one behavior type has a business logic precedence relationship with other behavior types; the total amount counting module 1200 is configured to count a total amount of users corresponding to each behavior type between two commodity objects according to the number of the same users implementing the same behavior type and generating the same type of interactive behavior data; the similarity calculation module 1300 is configured to calculate similarity data between each two commodity objects, where the similarity data is a sum of similarities calculated and determined according to total users of each behavior type of the corresponding two commodity objects; the response recommending module 1400 is configured to respond to the commodity matching instruction, and query and determine a plurality of corresponding similar commodity objects for the target commodity object specified by the instruction according to the similarity data.

In a further embodiment, the data acquisition module 1100 includes: the historical data acquisition submodule is used for acquiring user historical behavior data corresponding to the commodity object in the commodity database; the data cleaning submodule is used for performing data cleaning on the historical behavior data of the user and determining interactive behavior data of a plurality of behavior types corresponding to each commodity object, and the interactive behavior data correspondingly comprises click behavior data acting on the commodity object and at least any one of the following items: the shopping behavior data of the commodity object added to the shopping cart, the ordering behavior data of the commodity object and the payment behavior data of the order of the commodity object.

In a further embodiment, the similarity calculation module 1300 includes: the behavior calculation submodule is used for determining the similarity of the behavior types between every two commodity objects based on the behavior types; the weighting calculation sub-module is used for applying a preset weight matching scheme and carrying out weighted summation on the similarity of a plurality of common behavior types of every two commodity objects to obtain similarity data between every two commodity objects; the matrix construction submodule is used for constructing a similarity matrix used for storing the similarity data, wherein each element is used for storing the similarity data between the commodity object pointed by the row coordinate and the commodity object pointed by the column coordinate; and the matrix normalization submodule is used for normalizing the similarity data in the similarity matrix.

In order to solve the technical problem, an embodiment of the present application further provides a computer device. As shown in fig. 6, the internal structure of the computer device is schematically illustrated. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected by a system bus. The computer-readable storage medium of the computer device stores an operating system, a database and computer-readable instructions, the database can store control information sequences, and the computer-readable instructions, when executed by the processor, can enable the processor to implement a collaborative commodity recommendation method. The processor of the computer device is used for providing calculation and control capability and supporting the operation of the whole computer device. The memory of the computer device may store computer readable instructions, and when the computer readable instructions are executed by the processor, the processor may execute the product collaborative recommendation method of the present application. The network interface of the computer device is used for connecting and communicating with the terminal. Those skilled in the art will appreciate that the architecture shown in fig. 6 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In this embodiment, the processor is configured to execute specific functions of each module and its sub-module in fig. 5, and the memory stores program codes and various data required for executing the modules or the sub-modules. The network interface is used for data transmission to and from a user terminal or a server. The memory in this embodiment stores program codes and data necessary for executing all modules/sub-modules in the product collaborative recommendation apparatus of the present application, and the server can call the program codes and data of the server to execute the functions of all sub-modules.

The present application further provides a storage medium storing computer-readable instructions, which when executed by one or more processors, cause the one or more processors to perform the steps of the product collaborative recommendation method according to any of the embodiments of the present application.

The present application also provides a computer program product comprising computer programs/instructions which, when executed by one or more processors, implement the steps of the method as described in any of the embodiments of the present application.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments of the present application can be implemented by a computer program, which can be stored in a computer-readable storage medium, and when the computer program is executed, the processes of the embodiments of the methods can be included. The storage medium may be a computer-readable storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM).

In summary, according to the method and the device, deep data mining is performed on the user behavior data by means of data mining, so that the behavior type information in the user behavior data is utilized, the matching degree of the similar commodity objects recommended to the user can be improved, the real requirements of the user can be matched, and the similar commodity objects can obtain higher click rate.

Those of skill in the art will appreciate that the various operations, methods, steps in the processes, acts, or solutions discussed in this application can be interchanged, modified, combined, or eliminated. Further, other steps, measures, or schemes in various operations, methods, or flows that have been discussed in this application can be alternated, altered, rearranged, broken down, combined, or deleted. Further, steps, measures, schemes in the prior art having various operations, methods, procedures disclosed in the present application may also be alternated, modified, rearranged, decomposed, combined, or deleted.

The foregoing is only a partial embodiment of the present application, and it should be noted that, for those skilled in the art, several modifications and decorations can be made without departing from the principle of the present application, and these modifications and decorations should also be regarded as the protection scope of the present application.

Claims

1. A collaborative commodity recommendation method is characterized by comprising the following steps:

2. The collaborative commodity recommendation method according to claim 1, wherein the step of obtaining interactive behavior data of a plurality of behavior types corresponding to the commodity object in the commodity database comprises the steps of:

3. The collaborative commodity recommendation method according to claim 2, wherein calculating similarity data between two commodity objects comprises the steps of:

and normalizing the similarity data in the similarity matrix.

4. The collaborative commodity recommendation method according to claim 3, wherein in the step of determining the similarity between two commodity objects with respect to each behavior type based on the behavior type, the similarity calculation is performed by using any one of the following algorithms: cosine similarity calculation, Euclidean distance algorithm, Pearson correlation coefficient algorithm, Jacard algorithm, log-likelihood algorithm, and Manhattan distance algorithm.

5. The collaborative commodity recommendation method according to claim 3, wherein in the step of applying a preset weight matching scheme, the weight matching scheme is configured to match weights from small to large in the following order: click behavior data, purchase adding behavior data, order placing behavior data and payment behavior data.

6. The collaborative commodity recommendation method according to any one of claims 1 to 5, wherein in response to a commodity matching instruction, a plurality of corresponding similar commodity objects are queried and determined for a target commodity object specified by the instruction according to the similarity data, comprising the steps of:

performing reverse sorting according to the similarity data;

7. A collaborative commodity recommendation apparatus, comprising:

the data acquisition module is used for acquiring interactive behavior data of a plurality of behavior types corresponding to the commodity objects in the commodity database, wherein at least one behavior type has a business logic precedence relationship with other behavior types;

the total amount counting module is used for counting the total amount of the common users corresponding to each behavior type between every two commodity objects according to the same user number of the same behavior types and the same interactive behavior data generated by implementing the same behavior types;

the similarity calculation module is used for calculating similarity data between every two commodity objects, and the similarity data is a sum of similarity determined by calculation according to the total amount of users sharing each behavior type of the corresponding two commodity objects;

and the response recommending module is used for responding to the commodity matching instruction and inquiring and determining a plurality of corresponding similar commodity objects for the target commodity object specified by the instruction according to the similarity data.

8. A computer device comprising a central processor and a memory, characterized in that the central processor is adapted to invoke execution of a computer program stored in the memory to perform the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that it stores, in the form of computer-readable instructions, a computer program implemented according to the method of any one of claims 1 to 7, which, when invoked by a computer, performs the steps comprised by the corresponding method.

10. A computer program product comprising computer program/instructions, characterized in that the computer program/instructions, when executed by a processor, implement the steps of the method as claimed in any one of claims 1 to 7.