TW201804400A - Data object pushing method, device and system - Google Patents

Abstract

A data object pushing method, device and system, the method comprising: receiving a data object access request associated with a second user and sent by a client corresponding to a first user, wherein the second user has multiple data objects associated therewith; determining one or more data objects with respect to the access request according to multiple data object sets determined by the multiple data objects, and determining a target data object set associated with the first user from the multiple data object sets; and selecting to send to the client at least one data object from the target data object set and the one or more data objects determined with respect to the access request. The method, device and system can be combined with the actual intention of a user and used to push the most relevant data object to the user.

Description

Method, device and system for pushing data objects

The present application relates to the technical field of data processing, and in particular, to a method for pushing data objects, a device for pushing data objects, and a system for pushing data objects.

Content push is a way to promote specific advertising content, news content, notification content, audio and video multimedia resource content to specific people.

In the prior art, the following methods are usually used to push content: The technology of using user tags to target the crowd is a common technology in the field of advertising and other content delivery. This technology is based on the user ’s historical behavior or the user ’s own attributes, etc. The data labels the user.

The tag technology based on the user's historical behavior is mainly aimed at the user's browsing, searching, and clicking on various web pages, and clicks or conversions on advertisements or content, to map the user to a certain label. For example, for an e-commerce website, according to the user's search, clicks on products, collections, adding shopping carts, transactions, and other historical behaviors, correspondingly generate shop veterans, new customers, and stores / add purchases in n days. Users who have seen items in the store, etc.

User attribute tags are generally attributes of the user, such as the user's gender, age, region, and occupation. It has nothing to do with the specific store.

Although the above-mentioned tags such as demographics and user behavior can meet the needs of segmentation and distribution of the crowd in some scenarios, they cannot accurately predict the actual intention of the user, and are tags that are not directly related to the conversion goal. At the same time, because this technology is a computing technology based on historical data, in the case where future users cannot be predicted, this user content matching strategy based on user tags is also relatively blind.

Therefore, a technical problem that needs to be urgently solved by those skilled in the art is to propose a data object push mechanism to push the most relevant data object to the user in combination with the actual intention of the user.

The technical problem to be solved in the embodiments of the present application is to provide a method for pushing data objects, combining the user's actual intention to push the most relevant data objects to the user.

Correspondingly, the embodiment of the present application further provides a device for pushing data objects and a system for pushing data objects to ensure the implementation and application of the above method.

In order to solve the above problems, an embodiment of the present application discloses a system for pushing data objects. The system includes: one or more processors; a memory; and one or more modules, the one or more modules. Stored in the record The memory is configured to be executed by the one or more processors, and the one or more modules have the following functions: receiving data objects associated with the second user sent by the client corresponding to the first user An access request, wherein the second user has an associated plurality of data objects, and a plurality of data object sets determined based on the plurality of data objects; determining one or more data objects for the access request, and From the plurality of data object sets, determining a target data object set related to the first user; from the target data object set and the one or more data objects determined for the access request , Selecting to send at least one data object to the client.

An embodiment of the present application further discloses a method for pushing a data object. The method includes: receiving a data object access request associated with a second user sent by a client corresponding to a first user, wherein the second user Having a plurality of associated data objects, and a plurality of data object sets determined according to the plurality of data objects; determining one or more data objects for the access request, and, from the plurality of data object sets, Determining a target data object set related to the first user; and selecting from the target data object set and the one or more data objects determined for the access request to send at least one data to the client object.

Preferably, the step of determining a target data object set related to the first user from the plurality of data object sets includes: determining the first user and the plurality of data object sets respectively. The first N data object sets with the largest actual preference value are selected as the target data object set related to the first user, where N is a positive integer and N is less than the number of data object sets.

Preferably, the step of separately determining actual preference values of the first user and the plurality of data object sets includes: obtaining predicted preference values of the plurality of data object sets separately; and determining the first The correlation between the user and the data object set; for each data object set, the correlation is used to correct the predicted preference value to obtain the actual preference of the first user and the data object set value.

Preferably, the step of separately obtaining the predicted preference values of the plurality of data object sets includes: separately obtaining a set feature vector of the data object set; obtaining a feature value of a user characteristic of a user of the entire network, and The feature value of the user feature determines a corresponding user feature vector; the set feature vector and the user feature vector are respectively input into a preset prediction model to obtain a prediction preference value of the data object set.

Preferably, the prediction model is established in the following manner: Acquire all the visitor feature vectors of all the visitors who access the associated page of the second user; obtain the attribute feature vectors of the data objects associated with the second user; obtain all the associated pages of the second user The visit behavior data of the visitors on the associated page is used as sample information; and a prediction model is generated based on the sample information, the attribute feature vector, and the visitor feature vector.

Preferably, the step of separately obtaining a set feature vector of the data object set includes: obtaining a feature value and a feature vector of attribute characteristics of each data object in the data object set; The feature values and feature vectors of the attribute features of the data objects are summarized to obtain a set feature vector of the data object set.

Preferably, the data object includes header information, and the step of separately determining the correlation between the first user and the data object set includes: separately calculating the data object set according to a preset word vector model. Collection of word vectors; obtaining a specified number of data objects recently viewed by the first user, and obtaining the intentional word vector of the first user based on the specified number of data objects; calculating the intentional word vector and each Similarity of set word vectors Degree as the correlation between the first user and the corresponding set of data objects.

Preferably, the step of selecting to send at least one data object to the client from the target data object set and the one or more data objects determined for the access request includes: from the target At least one data object is selected from the data object set and the one or more data objects determined for the access request; a target page is generated according to the selected data object; and the target page is returned to the client.

Preferably, the first user is a buyer user, the second user is a seller user, and the data object associated with the second user is a product sold by the seller user, and the data object A collection is a combination of related products in the same store.

An embodiment of the present application also discloses a device for pushing data objects. The device includes: an access request receiving module for receiving a data object access request associated with a second user sent by a client corresponding to a first user, The second user has a plurality of data objects associated with it, and a plurality of data object sets determined according to the plurality of data objects; a data object determination module is configured to determine one or more of the access requests; Data objects; a target data object set determining module, for determining a target data object set related to the first user from the plurality of data object sets; A data object sending module is configured to select to send at least one data object to the client from the target data object set and the one or more data objects determined for the access request.

Preferably, the target data object set determination module includes: an actual preference determination sub-module for determining actual preference values of the first user and the plurality of data object sets, respectively; a target data object set selection A sub-module for selecting the first N data object sets with the largest actual preference value as the target data object set related to the first user, where N is a positive integer and N is less than the number of data object sets .

Preferably, the actual preference determining sub-module includes: a prediction preference obtaining unit for obtaining the predicted preference values of the plurality of data object sets respectively; and a correlation calculation unit for separately determining the first user A correlation with the data object set; a correction unit, for each data object set, using the correlation to correct the predicted preference value to obtain the first user and the data object set Actual preference value.

Preferably, the prediction preference acquisition unit includes: a collection feature vector acquisition subunit, which is used to obtain the collection feature vector of the data object set; a user feature vector acquisition subunit, which is used to obtain the use of users across the network. The feature value of the user feature, and the corresponding user feature vector is determined according to the feature value of the user feature; A prediction preference calculation subunit is configured to respectively input the set feature vector and the user feature vector into a preset prediction model to obtain a prediction preference value of the data object set.

Preferably, the prediction model is established in the following manner: acquiring all visitor feature vectors of all visitors who visit the associated page of the second user in the entire network; obtaining attribute feature vectors of data objects associated with the second user Obtaining all visitor behavior data of the second user's associated page in the associated page as sample information; modeling based on the sample information, the attribute feature vector, and the visitor feature vector To generate a prediction model.

Preferably, the set feature vector acquisition subunit is further configured to: obtain feature values and feature vectors of the attribute features of each data object in the data object set; and attribute characteristics of all data objects in the data object set The eigenvalues and eigenvectors are summarized to obtain a set eigenvector of the data object set.

Preferably, the data object includes title information, and the relevance calculation unit includes: a set word vector calculation sub-unit for calculating a set word vector of the data object set respectively according to a preset word vector model; an intent Word vector calculation subunit, configured to obtain a specified number of data objects recently viewed by the first user, and based on the specified number of data objects To obtain the intent word vector of the first user; a similarity calculation subunit for calculating the similarity of the intent word vector and each set of word vectors as the first user and the corresponding data object Relevance of the collection.

Preferably, the data object sending module includes a data object selection sub-module for selecting at least one from the target data object set and the one or more data objects determined for the access request. A data object; a target page generation submodule for generating a target page according to the selected data object; a target page return submodule for returning the target page to the client.

Preferably, the first user is a buyer user, the second user is a seller user, and the data object associated with the second user is a product sold by the seller user, and the data object A collection is a combination of related products in the same store.

Compared with the prior art, the embodiment of the present application includes the following advantages: In the embodiment of the present application, the second user can preset one or more data object sets, and when receiving the When a data object access request is associated with a second user, one or more data objects are determined for the request, and a target data object set related to the first user is determined from the plurality of data object sets, and Among the set of data objects and one or more data objects determined for the request, at least one data object is selected to be sent to the client, so that the first use The data objects obtained by the client corresponding to the user are more in line with the preferences of the first user, enabling accurate data object push.

10‧‧‧ target page

20‧‧‧ Data Object

30‧‧‧Data Object Collection

201‧‧‧Access request receiving module

202‧‧‧Data Object Identification Module

203‧‧‧Target data object collection determination module

204‧‧‧Data Object Sending Module

300‧‧‧Server

322‧‧‧Central Processing Unit

326‧‧‧ Power

330‧‧‧Storage Media

332‧‧‧Memory

341‧‧‧operating system

342‧‧‧Apps

344‧‧‧ Information

350‧‧‧Wired or wireless network interface

356‧‧‧Keyboard

358‧‧‧I / O interface

FIG. 1 is a flowchart of steps in an embodiment of a method for pushing a data object in the present application; FIG. 1a is a schematic diagram of a target page in an embodiment of a method for pushing a data object in the present application; FIG. 2 is a kind of information in the application FIG. 3 is a schematic structural diagram of a server according to an embodiment of the present application.

In order to make the above-mentioned objects, features, and advantages of this application more comprehensible, the following further describes this application in detail with reference to the accompanying drawings and specific embodiments.

Referring to FIG. 1, there is shown a flowchart of steps in an embodiment of a method for pushing a data object in the present application, which may specifically include the following steps: Step 101: Receive a second user association sent by a client corresponding to a first user. A data object access request; wherein the second user has a plurality of data objects associated with it, and a plurality of data object sets determined according to the plurality of data objects.

Specifically, the users in the embodiment of the present application may include at least a first user and a second user, where the second user is a data object Provider, the first user is the pushed party.

The second user may have multiple data objects associated with it, and the data objects may be displayed on a display page of the designated website and / or on a page associated with the second user in the designated website.

In practice, the display page may also display related information of the page associated with the second user, for example, page identification information and / or page link information of the page associated with the second user. After the related information of the user-associated page, the user may jump to the page associated with the second user. The page associated with the second user may include multiple data objects associated with the second user.

In specific implementation, the data object displayed on the display page may include link information. When the data object is selected, it may jump to the page corresponding to the link information. As an example, the page corresponding to the link information may be the data. The detail page of the object or the page associated with the second user.

Further, in the embodiment of the present application, the second user may also combine the associated multiple data objects into multiple data object sets, and the data object set may be a set composed of multiple data objects having an associated relationship. For example, a second user may associate data objects I1, I2, and I3 in a data object set G1, associate data objects I4, I5 in a data object set G2, and associate data objects I6, I7, I8, and I9 in Data object collection G3.

In the embodiment of the present application, the first user may send an access request to the data object associated with the second user through the client corresponding to the first user. begging. As an example, the client corresponding to the first user may include a client of a designated website or a browser client.

In a specific implementation, the first user may issue an access request in at least one of the following ways:

(1) The first user can load a display page in the client corresponding to the first user. The display page includes multiple data objects including data objects associated with the second user. When the data object associated with the second user is selected in the presentation page, it is determined that the first user has sent an access request for the data object associated with the second user.

(2) If there are many data objects displayed on the display page, the first user can also search for the required data objects associated with the second user in the display page by searching. When the search is successful, click the second The user-related data object may be determined at this time that the first user has sent an access request for the second user-related data object.

(3) The first user may also call the interface provided by the client corresponding to the first user to pass in the link address of the data object associated with the second user to issue an access request for the data object.

(4) If the related information of the page associated with the second user is displayed in the presentation page, if the first user clicks the related information of the page associated with the second user, it can be determined that the first user has issued a Access requests for user-associated data objects.

(5) The first user may also call the interface provided by the client corresponding to the first user to pass in the link address of the page associated with the second user to issue an access request for the data object.

It should be noted that the manner in which the client corresponding to the first user sends an access request to the data object associated with the second user is merely an example of the embodiment of the present application, and those skilled in the art use other methods to pass the first user through other methods. It is possible that the corresponding client sends an access request to the data object associated with the second user, which is not limited in the embodiment of the present application.

The embodiment of the present application can be applied to an e-commerce scenario. The first user can be a buyer user, the second user can be a seller user, and the page associated with the second user can be used by the seller. The page where the user ’s store is located; the data object associated with the second user may be a product sold by the seller user, and the data object set may be a combination of products with an associated relationship in the same store. For example, in an e-commerce website, a seller can combine multiple related products that they sell in advance to obtain a combination of products. For example, a toothbrush, toothpaste, and cup can be combined, or a mobile phone, mobile phone case, Combination of mobile phone related products such as mobile phone film, earphone cable and charger.

Step 102: Determine one or more data objects for the access request, and determine a target data object set related to the first user from the plurality of data object sets. In the embodiment of the present application, After receiving the data object access request associated with the second user sent by the client corresponding to the first user, one or more second users corresponding to the instant access intent of the first user may be determined for the access request. Associated data object.

In a specific implementation, if the first user selects a data object associated with the second user in the presentation page and / or transmits the data object associated with the second user When an access request is issued by the link address of the data object, the corresponding data object can be directly determined according to the access request.

If the first user issues an access request by selecting the relevant information of the page associated with the second user in the presentation page and / or passing in the link address of the page associated with the second user, the second user may determine the second request The user-associated page, thereby obtaining the data object contained in the page associated with the second user as a data object corresponding to the access request.

Applying to the embodiment of the present application, after receiving an access request of a data object associated with a second user sent by a client corresponding to the first user, a plurality of data object sets associated with the second user may also be determined according to the access request And extract a target data object set from the plurality of data object sets, where the target data object set is the most relevant data object set of the plurality of data object sets that is related to the first user.

In a preferred embodiment of the embodiment of the present application, the step of determining a target data object set related to the first user from the plurality of data object sets may include the following sub-steps: sub-step S11, The actual preference values of the first user and the plurality of data object sets are determined separately; the calculation of the actual preference values in the embodiments of the present application needs to consider the instant access behavior of the first user.

In a preferred embodiment of the embodiment of the present application, the sub-step S11 may further include the following sub-steps: a sub-step S111, obtaining the predicted preference values of the multiple data object sets respectively; In the embodiment of the present application, the predicted preference value may be a value calculated offline. The calculation of the value may not take into account the user ’s immediate access behavior, but perform a comprehensive analysis based on the characteristics of the data object collection and the multi-dimensional characteristics in the designated website The calculated value.

In a preferred embodiment of the embodiment of the present application, the sub-step S111 may further include the following sub-steps: sub-step S1111, respectively obtaining a set feature vector of the data object set; in a preferred embodiment of the present application In the embodiment, the sub-step S1111 may further include: obtaining a feature value and a feature vector of an attribute feature of each data object in the data object set; and a feature value and feature of the attribute feature of all the data objects in the data object set The vectors are summarized to obtain a set feature vector of the data object set.

Specifically, each data object has attribute characteristics that reflect the attributes of the data object. For example, if the data object is a product, the corresponding attribute characteristic may include, but is not limited to, the category, price, sales volume, and positioning of the product. , Efficacy, gender distribution, age distribution, etc. of the users who purchased the product.

。 After determining the attribute characteristics of the data object, the feature value v _j (the feature value of the attribute characteristic of the j-th data object) of the data object can be obtained first, and the attribute feature vector of the data object can be calculated according to the feature value v _j

.

The characteristic value of the attribute characteristic can be obtained by calling the specified interface.

Mathematically, the eigenvector (eigenvector) of a linear transformation is a non- A degraded vector whose direction is unchanged by this transformation. The scale at which this vector is scaled under this transformation is called its eigenvalue (eigenvalue). If the vector v and the transformation satisfy Av = λv, then the vector v is said to be a feature vector of the transformation A, and λ is the corresponding feature value.

In the embodiment of the present application, the attribute feature vector may be calculated in a general way of calculating a feature vector based on the feature value. For example, once the feature value λ is determined, the corresponding feature vector may be obtained by solving the equation (A-λ1) v = 0. This application The embodiment does not limit the calculation method of the feature vector.

，以即時將資料物件集合特徵化到資料物件相同的向量空間中。 In the embodiment of the present application, according to the attribute feature values and attribute feature vectors of all data objects included in the data object set, one or more set feature vectors of the data object set may be obtained.

To instantly characterize a collection of data objects into the same vector space as the data objects.

：

In one embodiment, the following formula (1) can be used to calculate the set feature vector of the data object set.

:

In formula (1), a logarithmic function is used to smooth the characteristic values of the attribute characteristics of the data object. It can be seen that the larger the characteristic value of the attribute characteristics of the data object, the greater the proportion of the data object in the data object collection. The larger the ratio. For example, the larger the sales volume of a product, the larger the proportion of the product in the product combination.

Of course, the above formula (1) is only an example of the embodiment of the present application, and those skilled in the art may also use other methods to calculate the set feature vector of the data object set, which is not limited in the embodiment of the present application.

Sub-step S1112, obtaining the characteristic value of the user characteristics of the users on the entire network, and determining the corresponding user characteristic vector according to the characteristic values of the user characteristics; specifically, the users on the entire network refer to each visitor in the specified website user.

As a preferred example of the embodiment of the present application, the user characteristics may include behavior characteristics and / or demographic characteristics. Among them, the behavior characteristics may include, but are not limited to, behaviors such as browsing and collection of data objects by users; demographic attributes Features may include, but are not limited to, the gender, age, occupation, etc. of the user.

In specific implementation, the behavior characteristics of the user can be obtained from the log records of the specified website, and the demographic characteristics of the user can be obtained from the user database of the specified website. The user database records each registered use Demographic characteristics, and more.

，根據特徵值計算特徵向量的方式如上述子步驟S1111所述，本申請案實施例對此不再贅述。 After the behavior characteristics and / or demographic characteristics are determined, the characteristic values of a certain demographic characteristics and / or behavior characteristics of all users can be obtained first, the characteristic values of all users are summarized, and the average characteristic value is obtained. The average eigenvalue calculates the user eigenvector

The method for calculating the feature vector according to the feature value is as described in the above sub-step S1111, which is not described in the embodiment of the present application.

In step S1113, the set feature vector and the user feature vector are respectively input into a preset prediction model to obtain a prediction preference value of the data object set.

The preset prediction model can be a model obtained from offline training. In a preferred embodiment of the application embodiment, the prediction model can be established in the following manner, but it should be understood that the embodiments of the present application are not limited to this, and those skilled in the art use other methods to establish prediction models with the same effect. It is possible: to obtain the visitor feature vectors of all the visitors who access the associated page of the second user in the entire network; to obtain the attribute feature vectors of the data objects associated with the second user; to obtain all accesses to the second use The visit behavior data of the visitor's associated page on the associated page is used as sample information; the sample information, the attribute feature vector, and the visitor feature vector are modeled to generate a prediction model.

Specifically, the calculation method of the visitor feature vector is similar to the calculation method of the above-mentioned user feature vector, except that the data of the user feature vector is derived from the visit behavior of all visitors on the entire network, and the data of the visitor feature vector is derived from all the visits. Visit behavior of visitors of the associated page of the second user.

That is, the visitor feature vector is obtained according to the visitor feature value, and the visitor feature value is the feature value of the visitor feature. The visitor feature may include behavior feature and / or demographic attribute feature, wherein the behavior feature may include but is not limited to: 2. The browsing and collection of data objects associated with users; demographic characteristics may include, but are not limited to, the gender, age, and occupation of visitors.

The attribute feature vector of the data object is as described in the above-mentioned sub-step S1111, and is not repeated here.

All visitors who visit the associated page of the second user The access behavior data in the link page may be a visit behavior of a data object in the link page by a visitor who accesses the link page of the second user, and the access behavior may be expressed as (u, i), where u is a Visitor ID, i is the data object ID visited by the visitor in the associated page.

In a specific implementation, a (u, i) can be used as a sample information. In (u, i), if u performs a specified operation on i, the sample information is a positive sample; otherwise, if u does not perform a specified operation on i, the sample information is a negative sample. For example, in an e-commerce application scenario, if a user browses a product and purchases the product, this sample is a positive sample; otherwise, if the user browses a product but did not purchase the product, this Each sample is a negative sample.

After determining the sample information, the attribute feature vector and the visitor feature vector, it can be used as a modeling parameter and a preset modeling algorithm is used for modeling to obtain a prediction model. As an example, the modeling algorithm may be MPI (Message Passing Interface, which can be used for parallel computing) -GBDT (Gradient Boosting Decision Tree), a machine learning algorithm widely used for classification or regression problems. An iterative decision tree algorithm. The algorithm consists of multiple decision trees, and the conclusions of all the trees are added up as the final answer.) The algorithm is a nonlinear prediction model after modeling.

The function of the prediction model is to estimate the probability that a first user will perform a specified operation on a certain data object in the future based on the historical visit data of the visitors of the related page of the second user and the attribute characteristics of the data objects in the related page. For example, based on historical behavior data such as in-store visitors, merchandise, Estimate the probability that a future user will have a purchasing relationship with a certain product.

,

)。 In a preferred embodiment of the embodiment of the present application, after each set feature vector and user feature vector are input into a preset prediction model, the output user's prediction preference value for each data object set can be expressed as follows : S _L (

,

).

In sub-step S112, the correlation between the first user and the data object set is determined respectively; when applied to the embodiment of the present application, the correlation between the first user and each data object set can also be calculated.

In a preferred embodiment of the embodiment of the present application, the sub-step S112 may further include the following sub-steps: Sub-step S1121, respectively calculating a set word vector of the data object set according to a preset word vector model; Natural language is handed over to algorithms in machine learning for processing. Usually, the language needs to be mathematicalized. Word Embedding (or Word Representation) is a way to mathematicalize words in the language. In the embodiment of the application, the word vector is used to represent the word vector of each data object.

In an implementation manner, the preset word vector model may be generated in the following manner: each data object has title information in a designated website, and the title information of each data object in the designated website is used as a corpus, and a neural network-based The word vector algorithm trains the corpus to obtain a multi-dimensional word vector model. The word vector model includes word vectors of multiple data objects.

In specific implementation, neural network-based word vector algorithms can include recurrent neural network algorithms, recursive neural network algorithms, etc. The following illustrates the principles of one of the neural network algorithms as an example: suppose The n-1 data objects are wt-n + 1, ..., wt-2, wt-1. Now it is necessary to predict the next data object wt based on the known n-1 data objects. C (w) represents the word vector corresponding to the data object w. A unique set of word vectors is used in the entire model and exists in the matrix C (matrix of | V | × m), where | V | represents the size of the vocabulary (Total number of words in the corpus), m represents the dimension of the word vector. The conversion from w to C (w) is to take a row from the matrix. The first layer (input layer) of the network is C (wt-n + 1), ..., C (wt-2), C (wt -1) These n-1 vectors are joined together end-to-end to form a (n-1) m-dimensional vector, hereafter denoted as x. The second layer (hidden layer) of the network is directly calculated using d + Hx, where d is an offset term. After that, use tanh as the startup function. The third layer (output layer) of the network has | V | nodes in total, and each node yi represents the unnormalized log probability of the next word i. Finally, the softmax startup function is used to normalize the output value y into a probability. Finally, the formula for calculating y is: y = b + Wx + Utanh (d + Hx)

Among them, U (| V | × h matrix) is a parameter from the hidden layer to the output layer. Most calculations of the entire model are concentrated on the matrix multiplication of U and the hidden layer; W (| V | × (n-1) m) The matrix contains the directly connected edges from the input layer to the output layer (a directly connected edge is a linear transformation directly from the input layer to the output layer). If no directly connected edges are needed, set W to 0. Finally, the stochastic gradient descent rule can be used to obtain the word vector model.

After the word vector model is obtained, the word vector model can also be synchronized to storage for instant calculation.

。 In the embodiment of the present application, the set word vector of each data object set can be calculated according to the word vector model and the attribute characteristics of the data object.

.

：

In one embodiment, the following formula (2) can be used to calculate the set word vector

:

為第j個資料物件的詞向量。 Among them, v _j is the characteristic value of the attribute characteristic of the j-th data object, and smoothed by a logarithmic function.

Is the word vector of the j-th data object.

Sub-step S1122: Acquire a specified number of data objects recently browsed by the first user, and obtain the intent word vector of the first user based on the specified number of data objects; in the embodiment of the present application, the A user's historical browsing behavior to predict the user's immediate intentions. Specifically, a specified number of data objects recently browsed by the first user (for example, the 10 most recently browsed items on the day) can be obtained, and a word vector of the specified number of data objects can be obtained. According to the words of the specified number of data objects, Vector to obtain the intent word vector of the first user.

In one embodiment, the following formula (3) can be used to calculate the intent word vector of the first user:

Among them, T _j is the number of seconds between the data object browsed by user u and the current time, and α is the attenuation coefficient.

As can be seen from the above formula (3), the more recently viewed data objects, the greater the predicted contribution to the user's current intention.

Sub-step S1123, calculating the similarity of the intent word vector and each set of word vectors as the correlation between the first user and the corresponding set of data objects.

After obtaining the intent word vector of the first user and the set word vector of each data object set, the similarity between the two can be calculated separately, and the similarity reflects the degree of correlation between the first user and the corresponding data object set .

In one embodiment, the following formula (4) can be used to calculate the similarity of the intent word vector and each set of word vectors:

In sub-step S113, for each data object set, the predicted preference value is corrected by using the correlation to obtain the actual preference values of the first user and the data object set.

In the embodiment of the present application, after obtaining the intention word vector of the first user and the correlation degree of each set of word vectors, the predicted preference value can be corrected according to the correlation degree to obtain the first user and each data object set. Actual preference value.

In one embodiment, the following formula (5) can be used to calculate the actual preference value:

Where λ is the fusion parameter.

In step S12, the first N data object sets with the largest actual preference values are selected as the target data object set related to the first user, where N is a positive integer and N is less than the number of data object sets.

In the embodiment of the present application, after the actual preference values of the first user and each data object set are obtained, the first N data object sets with the largest actual preference value may be used as targets related to the first user. Data object set, where N is a positive integer and N is less than the number of data object sets.

Step 103: Choose to send at least one data object to the client from the target data object set and the one or more data objects determined for the access request.

In the embodiment of the present application, after the corresponding one or more data objects and the target data object set are determined according to the access request, at least one data object may be selected therefrom and sent to the client to display the selected data object in the client. .

In a preferred embodiment of this application, step 103 may further include the following sub-steps: sub-step S21, from the target data object set and the one or more data objects determined for the access request Step, selecting at least one data object; sub-step S22, generating a target page according to the selected data object Sub-step S23, returning the target page to the client.

Specifically, after the corresponding one or more data objects are determined according to the access request and the target data object set is determined, from the multiple data objects included in the target data object set and the one or more data objects corresponding to the access request, , Select one or more data objects as the data objects to be displayed to the first user. In a specific implementation, the selection method may be random selection or selection according to a preset priority strategy, for example, preferentially selecting a corresponding data object, and then selecting a data object in a target data set.

Then, a target page can be generated according to the selected data object, and the target page can be returned to the client to display the target page in the client.

For example, referring to the schematic diagram of the target page shown in FIG. 1a, the target page may be displayed in an application app client of a mobile terminal, or the target page may also be displayed in a PC client or a browser client of a mobile terminal. . The target page 10 displayed by the client may include a plurality of data objects 20 corresponding to the access request and a plurality of data object sets 30 related to the first user. Each data object set 30 may include a plurality of data objects. 20, so that the content displayed on the target page conforms to user preferences.

In the embodiment of the present application, the second user may preset one or more data object collections. When receiving a data object access request associated with the second user sent by the client corresponding to the first user, the request is directed to the request. determine One or more data objects, and determining a target data object set related to the first user from the plurality of data object sets, and from the target data object set and one or more data objects determined for the request In selecting, at least one data object is sent to the client, so that the data object obtained by the client corresponding to the first user is more in line with the preference of the first user, and accurate data object push is realized.

It should be noted that, for the method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should know that the embodiments of the present application are not limited by the described action sequence. Because according to the embodiments of the present application, some steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present application.

Referring to FIG. 2, there is shown a structural block diagram of an embodiment of a device for pushing data objects of the present application, which may include the following modules: an access request receiving module 201, configured to receive a message sent by a client corresponding to a first user. A data object access request associated with a second user, wherein the second user has a plurality of data objects associated with it, and a plurality of data object sets determined according to the plurality of data objects; a data object determination module 202 For determining one or more data objects for the access request; a target data object set determining module 203 is configured to determine, from the plurality of data object sets, a target data object related to the first user set; A data object sending module 204 is configured to select to send at least one data object to the client from the target data object set and the one or more data objects determined for the access request.

In a preferred embodiment of the embodiment of the present application, the target data object set determination module 203 may include the following sub-modules: an actual preference determination sub-module for determining the first user and all The actual preference values of the plurality of data object sets are described; the target data object set selection submodule is used to select the first N data object sets with the largest actual preference value as the target data objects related to the first user Set, where N is a positive integer and N is less than the number of data object sets.

In a preferred embodiment of the embodiment of the present application, the actual preference determining sub-module may further include the following unit: a prediction preference obtaining unit configured to obtain the predicted preference values of the multiple data object sets respectively; related A degree calculation unit for determining the correlation between the first user and the data object set, respectively; a correction unit for each data object set, using the correlation to correct the predicted preference value To obtain actual preference values of the first user and the data object set.

In a preferred embodiment of the embodiment of the present application, the prediction preference obtaining unit may further include the following subunits: a collection feature vector obtaining subunit, configured to obtain a collection feature vector of the data object set; The user feature vector acquisition subunit is used to obtain the feature value of the user feature of the entire network user, and the corresponding user feature vector is determined according to the feature value of the user feature; the prediction preference calculation subunit is used to separately The set feature vector and the user feature vector are input into a preset prediction model to obtain a prediction preference value of the data object set.

In a preferred embodiment of the embodiment of the present application, the prediction model is established in the following manner: obtaining all visitor feature vectors of all visitors who access the associated page of the second user in the entire network; obtaining the second Attribute characteristic vectors of user-associated data objects; obtaining access behavior data of all visitors who visited the associated page of the second user in the associated page as sample information; according to the sample information and the attribute characteristics The vector and the visitor feature vector are modeled to generate a prediction model.

In a preferred embodiment of the embodiment of the present application, the set feature vector acquisition subunit is further configured to: obtain a feature value and a feature vector of attribute characteristics of each data object in the data object set; The feature values and feature vectors of the attribute features of all data objects in the data object set are summarized to obtain a set feature vector of the data object set.

In a preferred embodiment of the embodiment of the present application, the data object The file includes title information, and the relevance calculation unit may further include the following subunits: a set word vector calculation subunit, configured to respectively calculate a set word vector of the data object set according to a preset word vector model; an intent word vector A calculation sub-unit for obtaining a specified number of data objects recently viewed by the first user, and obtaining the intent word vector of the first user based on the specified number of data objects; a similarity calculation sub-unit for separately Calculate the similarity of the intent word vector and each set of word vectors as the correlation between the first user and the corresponding set of data objects.

In a preferred embodiment of the embodiment of the present application, the data object sending module 204 may include the following sub-modules: a data object selection sub-module for selecting from the target data object set and the Among the one or more data objects determined by the access request, at least one data object is selected; a target page generation submodule is used to generate a target page according to the selected data object; the target page returns a submodule to be used to The target page is returned to the client.

In a preferred embodiment of the embodiment of the present application, the first user is a buyer user, the second user is a seller user, and the data objects associated with the second user are used by the seller. For the goods sold by the retailer, the data object set is a combination of goods with an associated relationship in the same store.

As for the device embodiment, since it is basically similar to the method embodiment described above, the description is relatively simple. For the relevant part, refer to the description of the method embodiment.

An embodiment of the present application further provides a system for pushing data objects. The system for pushing data objects may include: one or more processors; a memory; and one or more modules that store one or more modules. In the memory and configured to be executed by one or more processors, wherein the one or more modules have the following functions: receiving access to data objects associated with the second user sent by the client corresponding to the first user A request, wherein the second user has an associated plurality of data objects and a plurality of data object sets determined based on the plurality of data objects; determining one or more data objects for the access request, and, Determining a target data object set related to the first user from the plurality of data object sets; from the target data object set and the one or more data objects determined for the access request, Choose to send at least one data item to the client.

Optionally, the one or more modules may have the following functions: determining the actual preference values of the first user and the plurality of data object sets separately; selecting the first N data objects with the largest actual preference value Collection Is the target data object set related to the first user, where N is a positive integer and N is less than the number of data object sets.

Optionally, the one or more modules may have the following functions: obtaining the predicted preference values of the plurality of data object sets separately; separately determining the correlation between the first user and the data object set; For each data object set, the predicted preference value is corrected by using the correlation to obtain the actual preference values of the first user and the data object set.

Optionally, the one or more modules may have the following functions: obtaining a set feature vector of the data object set separately; obtaining a feature value of a user feature of a user of the entire network, and according to the feature of the user feature The value determines the corresponding user feature vector; the set feature vector and the user feature vector are respectively input into a preset prediction model to obtain a prediction preference value of the data object set.

Optionally, the one or more modules may have the following functions: obtaining the visitor feature vectors of all visitors who visit the associated page of the second user on the entire network; obtaining the data objects associated with the second user; Attribute feature vector; obtaining the access behavior data of all visitors who visited the associated page of the second user in the associated page as sample information; according to the sample information, the attribute feature vector, and the visitor The feature vector is used for modeling to generate a prediction model.

Optionally, the one or more modules may have the following functions: obtaining feature values and feature vectors of the attribute features of each data object in the data object set; and attribute characteristics of all data objects in the data object set The eigenvalues and eigenvectors are summarized to obtain a set eigenvector of the data object set.

Optionally, the one or more modules may have the following functions: respectively calculating a set word vector of the data object set according to a preset word vector model; obtaining a specified number of data objects recently browsed by the first user, And obtaining the intent word vector of the first user based on the specified number of data objects; calculating the similarity of the intent word vector and each set of word vectors separately as the first user and the corresponding data object Relevance of the collection.

Optionally, the one or more modules may have the following functions: selecting at least one data object from the target data object set and the one or more data objects determined for the access request; according to the The selected data object generates a target page; the target page is returned to the client.

Optionally, the first user is a buyer user, the second user is a seller user, and the data object associated with the second user is a product sold by the seller user, and the data object Collected in the same store A combination of related products.

FIG. 3 is a schematic structural diagram of a server according to an embodiment of the present application. The server 300 may have a large difference due to different configurations or performance, and may include one or more central processing units (CPU) 322 (for example, one or more processors) and a memory 332, one Or more than one storage medium 330 storing the application 342 or the data 344 (for example, one or one storage device in Shanghai). The memory 332 and the storage medium 330 may be temporarily stored or persistently stored. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of command operations on the server. Furthermore, the central processing unit 322 may be configured to communicate with the storage medium 330 and execute a series of instruction operations in the storage medium 330 on the server 300.

The server 300 may also include one or more power sources 326, one or more wired or wireless network interfaces 350, one or more input-output interfaces 358, one or more keyboards 356, and / or, one or more operations System 341, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, and so on.

In one aspect, the central processing unit 322 may execute an instruction on the server 300 to receive a data object access request associated with a second user sent by a client corresponding to the first user, wherein the second user has A plurality of associated data objects and a plurality of data object sets determined according to the plurality of data objects; Determining one or more data objects for the access request, and determining a target data object set related to the first user from the plurality of data object sets; from the target data object set and the For one or more data objects determined by the access request, selecting to send at least one data object to the client.

Optionally, the central processing unit 322 may further execute instructions on the server 300 to determine the actual preference values of the first user and the plurality of data object sets, respectively; and select the one with the largest actual preference value. The first N data object sets are used as target data object sets related to the first user, where N is a positive integer and N is less than the number of data object sets.

Optionally, the central processing unit 322 may further execute instructions on the server 300 to obtain the predicted preference values of the plurality of data object sets separately; and determine the first user and the data object set respectively. Correlation; For each data object set, the predicted preference value is corrected by using the correlation to obtain the actual preference values of the first user and the data object set.

Optionally, the central processing unit 322 may further execute an instruction on the server 300 to obtain a set feature vector of the data object set, respectively; Obtain the characteristic values of user characteristics of users across the network, and determine corresponding user characteristic vectors according to the characteristic values of the user characteristics; respectively, input the set characteristic vectors and the user characteristic vectors into preset predictions A model to obtain a predicted preference value for the set of data objects.

Optionally, the central processing unit 322 may further execute instructions on the server 300 to obtain the visitor feature vectors of all visitors who access the associated pages of the second user in the entire network; and obtain the second user The attribute feature vector of the associated data object; obtaining the access behavior data of all visitors who visited the associated page of the second user in the associated page as sample information; according to the sample information, the attribute feature vector, and The visitor feature vector is modeled to generate a prediction model.

Optionally, the central processing unit 322 may further execute an instruction on the server 300 to obtain the feature values and feature vectors of the attribute features of each data object in the data object set; The feature values and feature vectors of the attribute features of the data objects are summarized to obtain a set feature vector of the data object set.

Optionally, the central processing unit 322 may also execute instructions on the server 300 for the following operations: According to a preset word vector model, a set word vector of the data object set is calculated respectively; a specified number of data objects recently browsed by the first user are obtained, and the first user is obtained based on the specified number of data objects The intention word vector of each; the similarity of the intention word vector and each set of word vectors is calculated separately as the correlation between the first user and the corresponding set of data objects.

Optionally, the central processing unit 322 may further execute an instruction on the server 300 to select at least one piece of data from the target data item set and the one or more data items determined for the access request. Object; generating a target page according to the selected data object; returning the target page to the client.

Optionally, the first user is a buyer user, the second user is a seller user, and the data object associated with the second user is a product sold by the seller user, and the data object A collection is a combination of related products in the same store.

Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments may refer to each other.

Those skilled in the art should understand that the embodiments of the embodiments of the present application may be provided as a method, a device, or a computer program product. Therefore, the embodiment of the present application may adopt a completely hardware embodiment, a completely software embodiment, or A combination of software and hardware embodiments. Moreover, the embodiments of the present application may adopt computer programs implemented on one or more computer-usable storage media (including but not limited to magnetic disk memory, CD-ROM, optical memory, etc.) containing computer-usable codes. The form of the product.

The embodiments of the present application are described with reference to flowcharts and / or block diagrams of the method, the terminal device (system), and the computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and a combination of the flow and / or block in the flowchart and / or block diagram can be implemented by computer program operation instructions. These computer program operation instructions can be provided to the processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to generate a machine, so that the processor of the terminal device can be processed by the computer or other programmable data The executed operation instruction generates a device for realizing a function specified in one or more flowcharts and / or one or more blocks of the block diagram.

These computer program operation instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing terminal device to work in a specific manner, so that the operation instructions stored in the computer-readable memory generate operation instructions An article of manufacture of a device, the operation instruction device realizing a function specified in one or more flowcharts and / or one or more blocks of a block diagram.

These computer program operation instructions can also be loaded on a computer or other programmable data processing terminal equipment, so that a series of operating steps can be performed on the computer or other programmable terminal equipment to generate computer-implemented processing. Thus, the operation instructions executed on a computer or other programmable terminal device provide steps for realizing the functions specified in one or more flowcharts and / or one or more blocks of the block diagram.

Although the preferred embodiments of the present application have been described, those skilled in the art can make other changes and modifications to these embodiments once they know the basic progressive concepts. Therefore, the scope of the appended application patents is intended to be construed as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of this application.

Finally, it should be noted that in this article, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply these entities. There is any such actual relationship or order between OR operations. Moreover, the terms "including", "comprising", or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or end device that includes a series of elements includes not only those elements but also those that are not explicitly listed Other elements, or elements inherent to such a process, method, article, or terminal. Without more restrictions, the elements defined by the sentence "including a ..." do not exclude that there are other identical elements in the process, method, article, or terminal device including the elements.

The method, device, and system for pushing data objects provided in this application have been described in detail above. Specific examples are used in this article to explain the principle and implementation of this application. The descriptions of the above embodiments are only for help. Understand the methodology of this application and its core ideas; at the same time, For those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation and the scope of application. In summary, the content of this description should not be understood as a limitation on the present application.

Claims (19)

A system for pushing data objects, characterized in that the system includes: one or more processors; a memory; and one or more modules, the one or more modules are stored in the memory and Is configured to be executed by the one or more processors, and the one or more modules have a function of receiving an access request of a data object associated with a second user sent by a client corresponding to the first user, wherein The second user has a plurality of data objects associated with the plurality of data objects determined based on the plurality of data objects; determining one or more data objects for the access request; and Among a plurality of data object sets, a target data object set related to the first user is determined; and from the target data object set and the one or more data objects determined for the access request, selecting a target data object set The client sends at least one data object. A method for pushing data objects, characterized in that the method comprises: receiving a data object access request associated with a second user sent by a client corresponding to a first user, wherein the second user has an associated multiple Data objects, and a plurality of data object collections determined according to the plurality of data objects; Determining one or more data objects for the access request, and determining a target data object set related to the first user from the plurality of data object sets; from the target data object set and the For one or more data objects determined by the access request, selecting to send at least one data object to the client. The method according to item 2 of the scope of patent application, wherein the step of determining a target data object set related to the first user from the plurality of data object sets includes: determining the first Actual preference values of the user and the plurality of data object sets; selecting the first N data object sets with the largest actual preference value as the target data object set related to the first user, where N is positive An integer and N is less than the number of data object collections. The method according to item 3 of the scope of patent application, wherein the step of separately determining actual preference values of the first user and the plurality of data object sets includes: obtaining the Predicting a preference value; determining the correlation between the first user and the data object set separately; and for each data object set, using the correlation to correct the predicted preference value to obtain the first use And the actual preference value of the data object set. The method according to item 4 of the scope of patent application, wherein The steps of separately obtaining the predicted preference values of the plurality of data object sets include: obtaining the set feature vectors of the data object sets separately; obtaining the feature values of user characteristics of users across the network, and according to the user feature The feature value determines the corresponding user feature vector; the set feature vector and the user feature vector are respectively input into a preset prediction model to obtain a prediction preference value of the data object set. The method according to item 5 of the scope of patent application, wherein the prediction model is established in the following manner: obtaining all visitor feature vectors of all visitors who access the associated pages of the second user in the entire network; obtaining the second Attribute characteristic vectors of user-associated data objects; obtaining access behavior data of all visitors who visited the associated page of the second user in the associated page as sample information; according to the sample information and the attribute characteristics The vector and the visitor feature vector are modeled to generate a prediction model. The method according to item 5 or 6 of the scope of the patent application, wherein the step of separately obtaining a set feature vector of the data object set includes: obtaining a feature value of an attribute feature of each data object in the data object set And feature vectors; characteristics of the attribute characteristics of all data objects in the data object set The feature values and feature vectors are summarized to obtain a set feature vector of the data object set. The method according to item 4 of the scope of patent application, wherein the data object includes header information, and the step of separately determining the correlation between the first user and the data object set includes: according to a preset word A vector model, respectively calculating a set word vector of the data object set; obtaining a specified number of data objects recently viewed by the first user, and obtaining the intent word vector of the first user based on the specified number of data objects; Calculate the similarity of the intent word vector and each set of word vectors as the correlation between the first user and the corresponding set of data objects. The method according to item 2 of the scope of the patent application, wherein, from the set of target data items and the one or more data items determined for the access request, selecting to send at least one to the client The step of the data object includes: selecting at least one data object from the target data object set and the one or more data objects determined for the access request; generating a target page according to the selected data object; The target page is returned to the client. The method according to any one of claims 2 to 6, wherein the first user is a buyer user, the second user is a seller user, and the second user is associated with Data object is seller user For the sold products, the data item set is a combination of products with an associated relationship in the same store. A device for pushing data objects, characterized in that the device includes an access request receiving module for receiving a data object access request associated with a second user sent by a client corresponding to a first user, wherein the The second user has a plurality of data objects associated with it, and a plurality of data object sets determined according to the plurality of data objects; a data object determination module for determining one or more data objects for the access request; A target data object set determining module is used to determine a target data object set related to the first user from the plurality of data object sets; a data object sending module is used to determine the target data object set from the target data object set And among the one or more data items determined for the access request, selecting to send at least one data item to the client. The device according to item 11 of the scope of patent application, wherein the target data object set determination module includes: an actual preference determination sub-module for determining the first user and the multiple data object sets respectively Actual preference value; the target data object set selection submodule is used to select the first N data object sets with the largest actual preference value as the target data object set related to the first user, where N is A positive integer and N is less than the number of data object sets. The device according to item 12 of the scope of patent application, wherein the actual preference determination sub-module includes: a prediction preference obtaining unit for obtaining the predicted preference values of the plurality of data object sets respectively; a correlation calculation unit, Configured to determine the correlation between the first user and the data object set separately; a correction unit configured to correct the predicted preference value by using the correlation for each data object set to obtain the The actual preference value of the first user and the set of data objects. The device according to item 13 of the scope of patent application, wherein the prediction preference acquisition unit includes: a collection feature vector acquisition subunit, which is used to obtain a collection feature vector of the data object set; a user feature vector acquisition subunit For obtaining the feature values of the user characteristics of the users on the entire network, and determining the corresponding user feature vectors according to the feature values of the user characteristics; a prediction preference calculation sub-unit for separately combining the set feature vectors and The user feature vector is input into a preset prediction model to obtain a prediction preference value of the data object set. The device according to item 15 of the scope of patent application, wherein the prediction model is established in the following manner: obtaining all visitor feature vectors of all visitors who visit the associated page of the second user in the entire network; obtaining the second User-associated data object attribute characteristics To obtain the access behavior data of all visitors who visited the associated page of the second user in the associated page as sample information; to construct according to the sample information, the attribute feature vector, and the visitor feature vector Model to generate a predictive model. The device according to item 14 or 15 of the scope of the patent application, wherein the set feature vector acquisition subunit is further configured to: obtain a feature value and a feature vector of an attribute feature of each data object in the data object set; The feature values and feature vectors of the attribute features of all data objects in the data object set are summarized to obtain a set feature vector of the data object set. The device according to item 13 of the scope of patent application, wherein the data object includes title information, and the relevance calculation unit includes: a set of word vector calculation subunits, which are respectively used to calculate all the information according to a preset word vector model. A set of word vectors for the set of data objects; an intent word vector calculation subunit for obtaining a specified number of data objects recently viewed by the first user, and obtaining the intention of the first user based on the specified number of data objects Word vector; similarity calculation subunit, configured to calculate the similarity of the intent word vector and each set of word vectors, respectively, as the correlation between the first user and the corresponding set of data objects. The device according to item 11 of the scope of patent application, wherein the data object sending module includes: A data object selection sub-module for selecting at least one data object from the target data object set and the one or more data objects determined for the access request; the target page generation sub-module is used for The selected data object generates a target page; the target page returns a sub-module for returning the target page to the client. The device according to any one of claims 11-15 of the scope of patent application, wherein the first user is a buyer user, the second user is a seller user, and the second user is associated with The data object is a product sold by a seller user, and the data object set is a combination of products with an associated relationship in the same store.