CN115858815A - Method for determining mapping information, advertisement recommendation method, device, equipment and medium - Google Patents

Abstract

The present disclosure provides a method for determining mapping information, an advertisement recommendation method, an advertisement recommendation apparatus, a device and a medium, which relate to the field of artificial intelligence, and in particular to the fields of natural language processing, intelligent search, intelligent recommendation, information flow, etc. The specific implementation scheme is as follows: determining at least one text message related to the historical preference behavior according to the historical preference behavior of the reference user within a predetermined period of time; determining at least one information pair according to the similarity between at least one text message and at least one keyword, wherein each information pair comprises corresponding target text message and target keyword; wherein the at least one keyword corresponds to the at least one original advertisement; and aiming at the same information pair, constructing a relation between the target text information and the target original advertisement corresponding to the target keyword to obtain first mapping information in the mapping information.

Description

Method for determining mapping information, advertisement recommendation method, device, equipment and medium

Technical Field

The present disclosure relates to the field of artificial intelligence technologies, and in particular, to the fields of natural language processing, intelligent search, intelligent recommendation, information flow, and the like, and more particularly, to a method for determining mapping information, an advertisement recommendation method, an advertisement recommendation apparatus, an electronic device, a storage medium, and a computer program product.

Background

The screen-opening advertisement refers to an advertisement appearing when an APP (Application) is started, and the main page is automatically entered after the screen-opening advertisement is displayed. The open screen advertisement is a special advertisement form, belongs to one of brand advertisements, and has the characteristics of short time and strong exposure.

Although the characteristic of full screen exposure of the open screen advertisement can well attract the attention of the user, help the advertiser to obtain the attention of the user and meet the popularization aim, the number of target users interested in the advertisement determined from the user group is small, and the popularization effect is limited.

Disclosure of Invention

The present disclosure provides a method of determining mapping information, an advertisement recommendation method, an apparatus of determining mapping information, an advertisement recommendation apparatus, an electronic device, a storage medium, and a computer program product.

According to an aspect of the present disclosure, there is provided a method of determining mapping information, including: determining at least one text message related to the historical preference behavior according to the historical preference behavior of the reference user within a predetermined period of time; determining at least one information pair according to the similarity between at least one text message and at least one keyword, wherein each information pair comprises corresponding target text message and target keyword; wherein the at least one keyword corresponds to the at least one original advertisement; and aiming at the same information pair, constructing the relation between the target text information and the target original advertisement corresponding to the target key words to obtain first mapping information in the mapping information.

According to another aspect of the present disclosure, there is provided an advertisement recommendation method including: acquiring current user information of a current user; determining candidate advertisements from at least one original advertisement according to the current user information and the mapping information to obtain a candidate advertisement set; and recommending advertisements according to the candidate advertisement set; the mapping information is obtained according to the method for determining the mapping information.

According to another aspect of the present disclosure, there is provided an apparatus for determining mapping information, including: the device comprises a text determining module, an information pair determining module and a first mapping information determining module. The text determination module is used for determining at least one piece of text information related to historical preference behaviors of a reference user in a preset time period. The information pair determining module is used for determining at least one information pair according to the similarity between at least one text message and at least one keyword, and each information pair comprises corresponding target text message and target keyword; wherein the at least one keyword corresponds to the at least one original advertisement. The first mapping information determining module is used for constructing a relation between the target text information and the target original advertisement corresponding to the target keyword aiming at the same information pair to obtain first mapping information in the mapping information.

According to another aspect of the present disclosure, there is provided an advertisement recommendation apparatus including: the system comprises an acquisition module, a set determination module and a first recommendation module, wherein the acquisition module is used for acquiring the current user information of the current user. The set determining module is used for determining candidate advertisements from at least one original advertisement according to the current user information and the mapping information to obtain a candidate advertisement set, and the mapping information is obtained according to the device for determining the mapping information. The first recommending module is used for recommending advertisements according to the candidate advertisement set.

According to another aspect of the present disclosure, there is provided an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the methods provided by the present disclosure.

According to another aspect of the present disclosure, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform a method provided by the present disclosure.

According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the method provided by the present disclosure.

It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.

Drawings

The drawings are included to provide a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:

FIG. 1 is a schematic diagram of an application scenario of a method for determining mapping information, an advertisement recommendation method and an apparatus according to an embodiment of the disclosure;

FIG. 2 is a schematic flow chart diagram of a method of determining mapping information in accordance with an embodiment of the present disclosure;

FIG. 3A is a schematic diagram of determining a user vector-keyword vector combination according to an embodiment of the present disclosure;

FIG. 3B is a schematic diagram of determining textual information-an original advertisement, according to an embodiment of the present disclosure;

FIG. 4 is a schematic diagram of clustering according to an embodiment of the present disclosure;

FIG. 5 is a schematic flow chart diagram of a method of determining mapping information in accordance with another embodiment of the present disclosure;

FIG. 6 is a schematic block diagram of a dotted edge graph according to an embodiment of the disclosure;

FIG. 7 is a schematic diagram of a customized extension according to an embodiment of the present disclosure;

FIG. 8 is a schematic flow chart diagram of an advertisement recommendation method in accordance with an embodiment of the present disclosure;

FIG. 9A is a schematic diagram of a method of advertisement recommendation in accordance with an embodiment of the present disclosure;

FIG. 9B is a schematic diagram of a method of advertisement recommendation in accordance with an embodiment of the present disclosure;

FIG. 9C is a schematic diagram of an intent model in accordance with an embodiment of the present disclosure;

fig. 10 is a schematic block diagram of an apparatus for determining mapping information according to an embodiment of the present disclosure;

FIG. 11 is a block diagram of a schematic structure of an advertisement recommendation device according to an embodiment of the present disclosure; and

fig. 12 is a block diagram of an electronic device for implementing a method of determining mapping information and/or a method of advertisement recommendation according to an embodiment of the present disclosure.

Detailed Description

Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of the embodiments of the disclosure are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

Fig. 1 is a schematic view of an application scenario of a method for determining mapping information, an advertisement recommendation method and an apparatus according to an embodiment of the present disclosure.

It should be noted that fig. 1 is only an example of a system architecture to which the embodiments of the present disclosure may be applied to help those skilled in the art understand the technical content of the present disclosure, and does not mean that the embodiments of the present disclosure may not be applied to other devices, systems, environments or scenarios.

As shown in fig. 1, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 serves as a medium for providing communication links between the terminal devices 101, 102, 103 and the server 105. Network 104 may include various connection types, such as wired and/or wireless communication links, and so forth.

The user may use the terminal devices 101, 102, 103 to interact with the server 105 via the network 104 to receive or send messages or the like. The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.

The server 105 may be a server providing various services, such as a background management server (for example only) providing support for websites browsed by users using the terminal devices 101, 102, 103. The background management server may analyze and process the received data such as the user request, and feed back a processing result (for example, mapping information is needed, an advertisement is needed to be recommended to the user, and the like) to the terminal device.

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

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

The present disclosure provides a method for determining mapping information and an advertisement recommendation method, which can be applied to open screen advertisements and also to other advertisement products having information search capability.

In the off-line phase, the method of determining mapping information may be performed once at intervals to update the mapping information, for example, to update first mapping information, which may characterize a relationship between text information and an advertisement, and second mapping information, which may characterize a relationship between a user and an advertisement. The updated mapping information is then stored in a cache database for use by the online phase.

In the online phase, an advertisement recommendation method may be performed, for example, first obtaining user information of the current user, which may include historical search terms and current user identification of the current user over a past period of time. Next, based on the obtained current user information, a candidate advertisement corresponding to the current user information may be queried in a cache database. For example, the advertisements corresponding to the search terms are queried based on the first mapping information, and the advertisements corresponding to the current user identifier are queried based on the second mapping information, so that a candidate advertisement set is obtained, and then the advertisements in the candidate advertisement set are recommended to the current user.

A method of determining mapping information provided by the present disclosure will be described in detail below, and then an advertisement recommendation method will be described in detail.

Fig. 2 is a schematic flow chart diagram of a method of determining mapping information in accordance with an embodiment of the present disclosure.

As shown in fig. 2, the method 200 of determining mapping information may include operations S210 to S230, and the method mainly finds out a strong intention group based on behaviors of reference users, where the strong intention group may represent: the actively expressed preference behaviors can directly hit target groups of the client intention, wherein the preference behaviors can comprise active searching, clicking, collecting, focusing on, praise and the like. It should be noted that, in the embodiment of the present disclosure, a user may be represented as an object using a target Application (APP), and a client may represent an advertiser that cooperates with the target application and needs the target application to recommend a self advertisement to the user.

In operation S210, at least one text message related to the historical preference behavior is determined according to the historical preference behavior of the reference user within a predetermined period of time.

For example, the method is used for constructing a relationship between user information and advertisements, and then recommending the advertisements to users using the target application, and the reference users may include users using the target application and users using other applications. The number of reference users is at least one.

For example, the predetermined period of time may represent a period of time that has elapsed, such as 30 days, 15 days, 7 days, etc.

For example, the historical preference behavior may include behaviors of searching, clicking, collecting, paying attention, praise, commenting, and the like, and may also include behaviors of watching a certain video for a time period longer than a predetermined time period. It is understood that the user will generate the above-mentioned historical preference behavior for information such as information, advertisement, etc. during the browsing process.

In one example, the text information associated with the historical preference behavior may include text such as title, content, description, etc. of the information, advertisement, etc. described above. For example, a video title that a user likes in the last week is "how to go through college english sixth rating", the text information related to the user's historical preferred behavior may include "how to go through college english sixth rating".

In another example, the text information related to historical preference behavior may also include subfolders obtained by word-cutting the text (e.g., advertisement title clicked, video title collected). For example, a video title that a user likes in the last week is "how to go through college students english sixth level", then textual information relevant for the user's historical preferred behavior may include "english sixth level".

In another example, when a picture is included in the information such as the information, advertisement, etc., the text information related to the historical preference behavior may also include text in the picture.

In operation S220, determining at least one information pair according to a similarity between the at least one text information and the at least one keyword, each information pair including corresponding target text information and target keyword; wherein the at least one keyword corresponds to the at least one original advertisement.

For example, a target application cooperates with several clients and presents advertisements of these clients to users according to actual needs. The customer may provide keywords corresponding to the original advertisement. For example, keywords corresponding to an English education advertisement advl may include "spoken English," "English sixth level," and so on.

Similarity between the text information and the keywords can be calculated, the similarity can include at least one of semantic similarity and text similarity, and then the text information and the keywords with higher similarity are determined as an information pair. For example, a certain text message is "how to pass through the college student english sixth level", a certain keyword is "english sixth level", and the similarity between the two is high, so that the text message "how to pass through the college student english sixth level" and the keyword "english sixth level" can be determined as one information pair.

In operation S230, a relationship between the target text information and the target original advertisement corresponding to the target keyword is constructed for the same information pair, resulting in first mapping information in the mapping information.

For example, for the above information pair, the original advertisement corresponding to the keyword "english six" is a certain english education advertisement adv1, and therefore, a relationship between the text information "how to go through college student english six" and the english education advertisement adv1 can be constructed.

For example, the constructed first mapping information may be stored in the form of a key-value pair, for example, the text information is stored as a key name key, and the identifier of the original advertisement is stored as a key value.

According to the technical scheme provided by the embodiment of the disclosure, the relation between the text information and the original advertisement is constructed according to the historical preference behavior of the reference user, and the first mapping information is obtained. In the subsequent process of recommending advertisements by using the first mapping information, the corresponding original advertisements can be inquired in the first mapping information by using the historical search words of the current user, so that the advertisements can be recommended to interested users.

It should be noted that operation S220 in the foregoing is intended to determine the information pair, and in practical applications, the information pair may be determined by a plurality of schemes, and the schemes of the plurality of determined information pairs may be used in combination. The following describes a scheme of specifying information pairs in detail with reference to fig. 3A to 4.

Fig. 3A is a schematic diagram of determining a user vector-keyword vector combination according to an embodiment of the present disclosure, and fig. 3B is a schematic diagram of determining a textual information-original advertisement 310 according to an embodiment of the present disclosure.

As shown in fig. 3A and 3B, the information pair may be determined by: a first similarity of the at least one text message and the at least one keyword to each other is determined. And then respectively determining the text information and the keywords as the target text information and the target keywords in the same information pair under the condition that the first similarity between the text information and the keywords is greater than or equal to a first similarity threshold value.

As shown in fig. 3A, for example, text information such as a search word, a search click title, an information stream click title, and an over-play video title of a reference user may be acquired by referring to a recommended behavior log 301, a search behavior log 302, and the like of an information stream of the user, where the over-play video represents a video having a play time length exceeding a predetermined time length.

Next, a user vector 304 characterizing the reference user may be determined using the textual information and the deep learning model described above, and a keyword vector 306 may be determined using the keywords 305 of the original advertisement and the deep learning model. For example, the text information described above is input into a deep learning model, which outputs a user vector 304, and the keywords 305 are input into the deep learning model, which outputs a keyword vector 306. The deep learning model may be, for example, the semantic model 303, and the semantic model 303 may adopt ERNIE (Enhanced Representation through Knowledge Integration).

Next, a keyword vector 306 may be retrieved using a retrieval algorithm based on the user vector 304, calculating a first similarity characterizing a distance between the user vector 304 and the keyword vector 306. The first similarity may be a cosine similarity. The above search algorithm may be HNSW (nearest neighbor search algorithm based on graph), and the search time may be approximately shortened to O (N) in large-scale search, where N is the number of nodes of the graph.

Next, filtering may be performed using a first similarity threshold. For example, if the first similarity between the user vector 304 and the keyword vector 306 is less than or equal to the first similarity threshold, the user vector 304 and the keyword vector 306 are filtered, otherwise, the user vector and the keyword vector are retained, and the user vector-keyword vector combination 307 is obtained.

As shown in fig. 3B, the user vector-keyword vector combination 307 may be mapped to a text information-keyword combination 308, and the text information-keyword combination 308 is the above-mentioned information pair. Next, a plurality of bundles of client-provided keywords 305 are retrieved, the original advertisement containing the keywords 305 in the combination 308 is screened out, and the textual information-keyword combination 308 is mapped to a textual information-original advertisement combination 309 for direct online use.

It should be noted that the first similarity threshold may be determined by manual sampling evaluation in combination with a low-flow abest experiment. For example, an initial threshold of the first similarity is preliminarily given, for example, an information pair obtained based on the initial threshold is extracted, and then whether the information pair is accurate is evaluated, so as to evaluate whether the initial threshold is reasonable, and if not, the initial threshold is adjusted. And adjusting the initial threshold value by combining multiple experiments, thereby obtaining a first similarity threshold value.

The information pair is determined by calculating the similarity between the text information and the keywords, and the information pair with similar semantics can be accurately determined from the text information and the keywords, so that the accuracy of the first mapping information is ensured.

Fig. 4 is a schematic diagram of clustering according to an embodiment of the present disclosure.

In some embodiments, the information pairs may also be determined by: clustering a plurality of keywords corresponding to the same original advertisement to obtain at least one synonymous cluster and a centroid corresponding to each synonymous cluster, wherein each synonymous cluster comprises at least one keyword. Next, for a target synonymous cluster of the at least one synonymous cluster, a second similarity between each of the at least one text message and a centroid of the target synonymous cluster is determined. And then determining the text information as the target text information and determining each keyword in the target synonymous cluster as the target keyword corresponding to the target text information under the condition that the second similarity between the text information and the centroid of the target synonymous cluster is greater than or equal to a second similarity threshold value.

As shown in fig. 4, the word packets of the keywords provided by the same client may be clustered, and words of similar subject contents may be collected together to form a synonymous cluster. Clustering is performed on a plurality of keywords in the word package 401 to obtain synonymous clusters 402, 403, 404.

In one example, all of the synonymous clusters may be determined as target synonymous clusters and subsequently processed. In another example, the synonym cluster with the keyword number greater than or equal to the number threshold may be determined as the target synonym cluster, and the number threshold may be 1, 2, 3, and so on. If the number of the keywords in a certain synonymous cluster is smaller than the number threshold, the keywords in the synonymous cluster are shown to be generic words, and the generic words can be removed, so that the influence of the generic words on the accuracy of the first mapping information is avoided.

The target synonym cluster is then recalled based on the user vector. The recall process may include: and carrying out weighted average or other processing on the vectors of the keywords in the synonymous cluster to obtain the centroid of the synonymous cluster. And then calculating a second similarity between the centroid and the user vector, and recalling the target synonymous cluster in the case that the second similarity is greater than or equal to a second similarity threshold, for example, recalling all keywords in the target synonymous cluster.

After a certain target synonymous cluster is recalled, the text information with a second similarity between the target synonymous cluster and a second similarity threshold value is determined as the target text information, and the keywords in the target synonymous cluster are respectively determined as the target keywords, so that a plurality of information pairs are obtained based on the target synonymous cluster.

The embodiment of the disclosure forms the synonymous cluster through clustering, and then recalls based on the synonymous cluster, thereby increasing the triggering efficiency, improving the probability of recalling words with different subjects, and reducing the influence of the similar words occupying a recall channel.

In other examples, the information pairs may also be determined by: an index can be constructed for the text information of the reference user by using a TDM (Tree-based Deep Match) Deep Tree retrieval algorithm, a Tree structure can be obtained after the index is constructed, the nodes of the Tree correspond to the text information, and the father node can correspond to a plurality of text information. And then recalling the text information of the user by using the keywords, for example, calculating the distance between the keywords and each node in the tree, and recalling the node with the smaller distance so as to obtain the keywords corresponding to the node. Next, based on the recalled keywords and textual information, an information pair may be constructed. By constructing the index, the retrieval efficiency can be improved.

It should be noted that, the method for determining mapping information in the foregoing may mine strong-intention people based on the behavior of the reference user, so as to improve the display amount of the advertisement. However, with the increase of customers and the diversification of demands, the strong-intention crowd mined by the method is limited, so that potential user crowds can be mined out of the strong-intention crowd through a similar crowd mining strategy. The potential user population may represent: other behaviors than the strong intention behavior are generated, the preference intention of the user needs to be estimated, and then the people with the expressed intention of the client are hit.

The similar population mining strategy is described below in conjunction with fig. 5 and 6. Fig. 5 is a schematic flowchart of a method of determining mapping information according to another embodiment of the present disclosure, and fig. 6 is a schematic structural diagram of a dot-edge map according to an embodiment of the present disclosure.

As shown in fig. 5, the method 500 for determining mapping information may include operation S510, and may further include operations S540 to S560. The method aims to mine a potential user crowd.

In operation S510, at least one text message related to the historical preference behavior is determined according to the historical preference behavior of the reference user within a predetermined period of time.

For example, the operation may refer to operation S210 above, and is not described herein again. In the present embodiment, the number of reference users may be plural.

In operation S540, a point-edge graph is constructed according to a plurality of reference users and at least one text message.

The point-edge graph comprises a plurality of user nodes corresponding to a plurality of reference users and at least one text node corresponding to at least one text message, wherein the user node corresponding to the target reference user is connected with the text node corresponding to the target text message through an edge, and the target text message is related to the historical preference behavior of the target reference user.

For example, text information may be referred to above, and the text information may be, for example, a title of an advertisement clicked by the user. For example, the text information may be obtained by obtaining advertisement click data referring to different media such as a bar. The text information may also include search terms for reference users.

The point-edge graph can then be constructed using the reference user and the textual information. The process of constructing the point-edge graph is described by taking the example that the reference user u1 clicks a certain advertisement adv 1.

For example, the dotted edge graph may include a user node _ u1 corresponding to the reference user u1, and a text node _ t1 corresponding to the title (i.e., text information) of the advertisement adv1, where the user node _ u1 and the text node _ t1 are connected via one edge. It can be seen that if the text information corresponding to the reference user u1 is the same as the text information corresponding to the reference user u2, a link "reference user u 1-text information-reference user u2" can be constructed in the point-and-edge graph.

In addition, the edges in the point edge graph can be constructed in a word segmentation matching mode. For example, if a certain word segment in the advertisement adv1 title clicked by the reference user u1 is the same as a certain word segment in the advertisement adv3 title clicked by the reference user u3, the edge may be used to connect the user node _ u3 and the text node _ t1 corresponding to the reference user u3, so as to construct a link of "user a-word segment-user B" in the point-edge graph.

As shown in fig. 6, fig. 6 is a schematic block diagram of a dot-edge diagram according to an embodiment of the present disclosure. For example, the example involves user A611, user B612, and user C613, with the advertisement title clicked on by user A611 including the click title 621 and the information searched including the search terms 631, 632. The advertisement title clicked by the user B612 includes the click title 622, or a certain keyword of the advertisement title clicked by the user B612 is the same as a certain keyword in the click title 622, and the searched information includes the search keyword 631. The advertisement title clicked by the user C613 includes the click title 621, or a certain keyword of the advertisement title clicked by the user C613 is the same as a certain keyword in the click title 621, the searched information includes the search keyword 633, and the constructed edge-and-dot graph is shown in fig. 6.

In operation S550, at least one user pair is determined from the plurality of reference users according to the dotted-edge graph.

Illustratively, the constructed point-edge graph may be utilized to generate a user vector. For example, a graph model may be trained in advance, with the input of the model being the graph and the output being the vector for each node. For another example, the nodes in the dotted edge graph may have respective characteristics, and for example, the characteristics of the user nodes may be determined by using text information such as advertisement titles, search terms, etc., clicked by the user. A user vector is then computed based on the features of the user nodes.

Next, a user pair may be determined based on the user vector. Users that click on an advertisement may be referred to as seed users, users that did not click on the advertisement may be referred to as potential users, and seed user-potential user pairs (i.e., user pairs) may be generated using seed user search to recall potential users.

In the recalling process, the similarity between the user vector of the seed user and the user vector of the potential user can be calculated, and the seed user and the potential user are determined as a seed user-potential user pair under the condition that the similarity is larger than or equal to a threshold value.

In operation S560, for the same user pair, a relationship between the user identifier of one reference user and the original advertisement associated with another reference user is constructed, resulting in second mapping information in the mapping information.

For example, in a seed user-potential user pair, when the seed user clicks on a commercial and the potential user does not click on the commercial, the relationship between the user identifier of the potential user and the commercial may be constructed to obtain a second mapping information. The constructed second mapping information may be stored in the form of key-value pairs, for example, with the user identification of the potential user as a key name and the identification of the commercial as a key value. After obtaining the second mapping information, the commercial may be pushed to the potential user during the process of making the advertisement recommendation.

According to the technical scheme provided by the embodiment of the disclosure, the point edge graph is constructed according to the reference user and the text information determined based on the historical preference behavior of the reference user, and the point edge graph can mine potential users, so that interested advertisements are recommended to the users, and meanwhile, the advertisement display amount is improved.

It should be noted that, in practical applications, the method 200 and the method 500 may be performed by any one of them, or may be used in combination.

The method can be used for mining strong intention crowds and potential crowds, but with the increasing growth of information flow and the difference of application attributes among different media, the behaviors of the user on the media side are gradually diversified, the user interest and characterization can be further covered, and customized extension is carried out on different clients. Fig. 7 is a schematic diagram of a customized extent according to an embodiment of the disclosure, and a flow of the customized extent is described below with reference to fig. 7.

As shown in fig. 7, for example, a group of people may be circled using keywords provided by a customer, and a basic positive sample may be determined using characteristics of the group of people. Users who have conversion activity (e.g., filling in forms, etc.) for the ad may also be obtained and their characteristics used to determine a positive sample of deltas. Randomly sampling from the rest users, and determining a negative sample by using the characteristics of the sampled users. The deep learning model is trained based on the positive and negative samples.

When the system is used, the trained deep learning model can be input by using the characteristics of each user in the user group of the target application, the model outputs the information of the delivered population, and then the relationship between each user in the delivered population and the advertisement is established, so that the second mapping information is supplemented, and the effect of expanding the number of users for the client is realized.

The method for determining the mapping information is described in detail above, and the advertisement recommendation method will be described below with reference to fig. 8, where fig. 8 is a schematic flow chart of the advertisement recommendation method according to an embodiment of the disclosure.

As shown in fig. 8, the advertisement recommendation method 800 may include operations S810 to S830.

In operation S810, current user information of a current user is acquired.

For example, a current user opens a target application and may recommend advertisements to the current user before entering the home page of the target application.

For example, the current user information may include historical search terms of the current user over a past period of time, and may also include an identification of the current user.

In operation S820, candidate advertisements are determined from at least one original advertisement according to the current user information and the mapping information, resulting in a candidate advertisement set.

For example, the mapping information is obtained according to the above method of determining mapping information, and the mapping information may include at least one of the first mapping information and the second mapping information. The first mapping information may characterize a relationship between the textual information and the original advertisement, for example, the textual information is a key name key and the original advertisement is identified by a key name value. The second mapping information may characterize a relationship between the user identifier of the reference user and the original advertisement, for example, the identifier of the reference user is a key name key and the identifier of the original advertisement is a key name value.

For example, according to the historical search term of the current user, text information that is the same as or similar to the historical search term is searched for in the first mapping information center, and an advertisement corresponding to the searched text information is determined as a candidate advertisement.

For another example, according to the current user identifier of the current user, a reference user identifier that is the same as the current user identifier may be searched in the second mapping information center, and an advertisement corresponding to the searched reference user identifier may be determined as a candidate advertisement.

In operation S830, advertisement recommendation is made according to the candidate advertisement set.

For example, all advertisements in the candidate advertisement set may be recommended to the current user, or a part of the advertisements may be randomly selected from the candidate advertisement set or selected in combination with a policy as target advertisements, and the target advertisements may be recommended to the current user.

According to the technical scheme provided by the embodiment of the disclosure, the current user information of the current user can be acquired when the target application is opened, and then the advertisement is recommended to the current user based on the current user information and the predetermined mapping information, so that the display quantity of the advertisement is ensured, and the interested advertisement is recommended to the user. In addition, the advertisement can be directly retrieved based on the mapping information without determining the mapping information in real time, so that the data processing efficiency can be improved, and the real-time performance of the open screen advertisement is ensured.

According to another embodiment of the present disclosure, the method for determining a candidate advertisement from at least one original advertisement according to the current user information and the mapping information to obtain a candidate advertisement set may include the following operations:

a current original advertisement is determined from the at least one original advertisement. In the first mapping information, text information corresponding to the current original advertisement is referred to as current text information. In the second mapping information, the reference user corresponding to the current original advertisement is referred to as a current reference user.

Next, in a case where it is determined that the historical search word is not consistent with the keyword corresponding to the at least one original advertisement, it may be determined whether the historical search word is consistent with the current text information, and it may also be determined whether the current user identifier is consistent with the user identifier of the current reference user. If the historical search terms are consistent with the current text information and/or the current user identification is consistent with the user identification of the current reference user, the current original advertisement can be added to the candidate advertisement set as a candidate advertisement, and the current original advertisement is deleted from at least one original advertisement.

And repeating the operation until the number of at least one original advertisement is 0, namely processing the original advertisements in sequence and determining whether to add each original advertisement to the candidate advertisement set.

Fig. 9A is a schematic diagram of an advertisement recommendation method according to an embodiment of the present disclosure, fig. 9B is a schematic diagram of an advertisement recommendation method according to an embodiment of the present disclosure, and fig. 9C is a schematic diagram of an intention model according to an embodiment of the present disclosure.

The following describes an advertisement recommendation method provided in the present disclosure example with reference to fig. 9A to 9C.

The present embodiment is described taking as an example a processing flow for one advertisement (hereinafter referred to as a current original advertisement), and when a target application cooperates with a plurality of advertisers, a processing flow similar to the current original advertisement may be performed sequentially for original advertisements provided by the plurality of advertisers. The process flow for the current original advertisement may include the following stages: a keyword matching stage, a cache database retrieval stage and an intention filtering stage.

In the keyword matching stage, current user information of the current user can be obtained, and the current user information can comprise current user identification and historical search words of the current user in a past period. The keyword corresponding to the current original advertisement may be pulled from a thesaurus of keywords (hereinafter referred to as a keyword thesaurus). It is then determined whether the historical search terms of the current user are consistent with the keywords corresponding to the current original advertisement.

If the historical search term is consistent with the keyword, it may be determined whether an interval duration between the search time of the historical search term and the current time is less than or equal to a predetermined duration, where the predetermined duration may be 5 minutes.

If yes, the interest of the current user to the current original advertisement is high at the current moment, so that the display priority of the current original advertisement can be improved, the current original advertisement is recommended, the interested advertisement is recommended to the user, and the flow quality of the client can be improved.

If not, although the interest of the user in the current advertisement is not as high as that in the case of "yes", the current user still has a high interest in the current original advertisement because the historical search words hit the advertisement keywords, so that the current original advertisement can be added to the candidate advertisement set as a candidate advertisement for subsequent advertisement recommendation. It should be noted that, when advertisement recommendation is performed subsequently, a target advertisement recommended to a current user may be selected from the candidate advertisement set, and the target advertisement may include the current original advertisement or may not include the current original advertisement.

And if the historical search words are not consistent with the keywords, entering a cache database retrieval stage.

It should be noted that, since the keywords are given by the client based on the understanding of the brand target population, when a user searches for the keywords on the APP in the past, the advertisement of the client can be presented to the user, and in this way, the accurate reach of the target population is ensured. But because different customers often have different levels of understanding and word buying ability, keywords provided by the customers are not accurate. For example, the client is conservative or does not know how to choose, providing only few keywords. For another example, the customer wants to increase the exposure rate, and then gives more broad words and long-tail words, which causes difficulty in accurately matching the target population. In addition, search behavior itself is an expression of the user's immediate interest, and there is also a migration or even jump in interest, which presents a challenge to the customer desiring to quantitatively complete the exposure every day. The above situation can cause the problems of insufficient keyword matching display amount and inaccurate exposure crowd positioning, but the customer has certain expectation on the amount and the exposure effect. In the embodiment, through a cache database retrieval stage and an intention filtering stage and in an expansion triggering mode, strong intention target crowds are screened, the display amount and the exposure effect are improved, and the effects of meeting the client display amount and improving the coverage proportion of interested audience crowds are achieved.

In the cache database retrieval stage, the advertisement may be retrieved from the cache database using the current user information, and the cache database stores the first mapping information and the second mapping information. For example, the current user identification is used to search whether corresponding advertisements exist in the first mapping information, and the historical search terms of the current user are used to search whether corresponding advertisements exist in the second mapping information.

And if the current original advertisement and the current user information accord with the first mapping information and/or the second mapping information in the cache database, adding the current original advertisement into the candidate advertisement set.

If the current original advertisement and the current user information do not accord with the first mapping information and the second mapping information in the cache database, an intention filtering stage can be carried out, and whether the current original advertisement is added to the candidate advertisement set or not is determined by using an intention model.

In the intention filtering stage, an external module can be requested, the request can carry current user information, the external module can carry out generalization processing on historical search words of a current user, the words obtained after the generalization processing are used as intention words, and the intention words are returned. Next, it may be determined whether the intent word is consistent with the keywords of the current original advertisement.

If not, the processing flow for the current original advertisement may be ended, and it is determined not to add the current original advertisement to the candidate advertisement set.

If so, the intent model may also be utilized to determine whether to add the current original advertisement to the set of candidate advertisements.

As shown in fig. 9C, the intent model may be trained in advance. Whether the historical search words of the users are consistent with the keywords of the original advertisement or not can be determined, the consistent users are used as positive-case users, and partial users are randomly screened as negative-case users in the inconsistent users. The method comprises the steps of constructing a positive sample by using the characteristics of positive users and the advertisement characteristics, constructing a negative sample by using the characteristics of negative users and the advertisement characteristics, and training an intention model by using the positive sample and the negative sample, thereby establishing the relevance between the crowd intention and the advertisement.

After the intent model is trained, the user characteristics of the current user and the advertisement characteristics of the current advertisement may be input into the trained intent model, the intent model outputs an evaluation value, and if the evaluation value is higher than a threshold, the current original advertisement is added to the candidate advertisement set, otherwise the current original advertisement is not added to the candidate advertisement set. The intention of the user is further mined out by using the intention model, and the discovery of the potential user is supplemented.

According to the method and the device, firstly, keywords provided by the client are matched through the historical search words of the APP user, and a part of brand target users are screened out, wherein the part of clients are strong-intention clients. And then, searching potential crowds by using a cache database, wherein mapping information in the cache database is realized by a behavior mining strategy, a similar crowd mining strategy and a customized model strategy to mine strong intention users and potential users for the client. Next, the intent model is further utilized for filtering.

According to the technical scheme provided by the embodiment of the disclosure, on the basis of the keywords provided by the client, the algorithm strategy is used for further triggering, the target population of the client is fully mined, the exposure of the advertisement is increased, the delivered advertisement can reach the audience population more, and the flow with higher quality is obtained. In addition, advertisements of interest can also be provided to users of the targeted application.

Fig. 10 is a schematic block diagram of an apparatus for determining mapping information according to an embodiment of the present disclosure.

As shown in fig. 10, the apparatus 1000 for determining mapping information may include a text determining module 1010, an information pair determining module 1020, and a first mapping information determining module 1030.

The text determination module 1010 is configured to determine at least one text message related to the historical preference behavior of the reference user within a predetermined period of time.

The information pair determining module 1020 is configured to determine at least one information pair according to a similarity between the at least one text message and the at least one keyword, where each information pair includes a corresponding target text message and a corresponding target keyword; wherein the at least one keyword corresponds to the at least one original advertisement.

The first mapping information determining module 1030 is configured to construct, for the same information pair, a relationship between the target text information and the target original advertisement corresponding to the target keyword, so as to obtain first mapping information in the mapping information.

According to another embodiment of the present disclosure, the information pair determination module includes: a first determination submodule and a second determination submodule. The first determining submodule is used for determining a first similarity between the at least one text message and the at least one keyword. The second determining submodule is used for respectively determining the text information and the keywords as the target text information and the target keywords in the same information pair under the condition that the first similarity between the text information and the keywords is determined to be larger than or equal to the first similarity threshold.

According to another embodiment of the present disclosure, the information pair determination module includes: a clustering submodule, a third determining submodule and a fourth determining submodule. The clustering submodule is used for clustering a plurality of keywords corresponding to the same original advertisement to obtain at least one synonymous cluster and a centroid corresponding to each synonymous cluster, and each synonymous cluster comprises at least one keyword. The third determining submodule is used for determining a second similarity between each of the at least one text message and the centroid of the target synonymous cluster aiming at the target synonymous cluster in the at least one synonymous cluster. And the fourth determining submodule is used for determining the text information as the target text information and determining each keyword in the target synonymous cluster as a target keyword corresponding to the target text information under the condition that the second similarity between the text information and the centroid of the target synonymous cluster is greater than or equal to the second similarity threshold value.

According to another embodiment of the present disclosure, the above apparatus further comprises: and the target cluster determining module is used for determining the synonym clusters with the keyword number being more than or equal to the number threshold as the target synonym clusters after at least one synonym cluster is obtained.

According to another embodiment of the present disclosure, the number of reference users is plural, and the apparatus further includes: the map building module, the user pair determining module and the second mapping information determining module. The graph building module is used for building a point-edge graph according to a plurality of reference users and at least one piece of text information; the point edge graph comprises a plurality of user nodes corresponding to a plurality of reference users and at least one text node corresponding to at least one text message, the user nodes corresponding to the target reference users are connected with the text nodes corresponding to the target text message through edges, and the target text message is related to the historical preference behavior of the target reference users. The user pair determining module is used for determining at least one user pair from a plurality of reference users according to the point edge graph. The second mapping information determining module is used for constructing a relationship between the user identification of one reference user and the original advertisement related to another reference user aiming at the same user pair to obtain second mapping information in the mapping information.

Fig. 11 is a block diagram of a schematic structure of an advertisement recommendation device according to an embodiment of the present disclosure.

As shown in fig. 11, the advertisement recommendation device 1100 may include: an acquisition module 1110, a set determination module 1120, and a first recommendation module 1130.

The obtaining module 1110 is configured to obtain current user information of a current user.

The set determining module 1120 is configured to determine a candidate advertisement from at least one original advertisement according to the current user information and mapping information, and obtain a candidate advertisement set, where the mapping information is obtained according to the mapping information determining apparatus.

The first recommendation module 1130 is configured to make advertisement recommendations based on the set of candidate advertisements.

According to another embodiment of the present disclosure, the current user information includes a historical search term, and the mapping information includes first mapping information representing a relationship between the text information and the original advertisement; the set determination module includes: a first execution sub-module, configured to repeatedly execute the following operations until the number of at least one original advertisement is 0: determining a current original advertisement from at least one original advertisement; the relation between the current original advertisement and the current text information accords with first mapping information; in response to the fact that the historical search terms are inconsistent with the corresponding keywords of the at least one original advertisement and the historical search terms are consistent with the current text information, the current original advertisement is taken as a candidate advertisement and is added to a candidate advertisement set; and deleting the current original advertisement from the at least one original advertisement.

According to another embodiment of the present disclosure, the current user information includes a current user identification; the mapping information comprises second mapping information representing the relationship between the user identification of the reference user and the original advertisement; the set determination module includes: a second execution sub-module for repeatedly executing the following operations until the number of the at least one original advertisement is 0: determining a current original advertisement from at least one original advertisement; the relation between the current original advertisement and the user identification of the current reference user accords with second mapping information; in response to the fact that the historical search words are inconsistent with the keywords corresponding to the at least one original advertisement and the current user identification is consistent with the user identification of the current reference user, adding the current original advertisement serving as a candidate advertisement into the candidate advertisement set; and deleting the current original advertisement from the at least one original advertisement.

According to another embodiment of the present disclosure, the current user information includes historical search terms, and the apparatus further includes: and the adding module is used for responding to the consistency of the historical search words and the keywords corresponding to at least one original advertisement, and adding the original advertisement corresponding to the keywords consistent with the historical search words into the candidate advertisement set as the candidate advertisement.

According to another embodiment of the present disclosure, the current user information includes historical search terms, and the apparatus further includes: and the second recommending module is used for recommending the original advertisement corresponding to the keyword consistent with the historical search word in response to the fact that the historical search word is consistent with the keyword corresponding to at least one original advertisement and the interval duration between the search time of the historical search word and the current time is less than or equal to the preset duration.

In the technical scheme of the disclosure, the collection, storage, use, processing, transmission, provision, disclosure and other processing of the personal information of the related user are all in accordance with the regulations of related laws and regulations and do not violate the good customs of the public order.

In the technical scheme of the disclosure, before the personal information of the user is acquired or collected, the authorization or the consent of the user is acquired.

According to an embodiment of the present disclosure, there is also provided an electronic device, comprising at least one processor; and a memory communicatively coupled to the at least one processor; the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the above-described method of determining mapping information and/or advertisement recommendation method.

According to an embodiment of the present disclosure, there is also provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the above-described method of determining mapping information and/or advertisement recommendation method.

According to an embodiment of the present disclosure, there is also provided a computer program product including a computer program, which when executed by a processor implements the above method of determining mapping information and/or the advertisement recommendation method.

Fig. 12 is a block diagram of an electronic device for implementing a method of determining mapping information and/or an advertisement recommendation method according to an embodiment of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the disclosure described and/or claimed herein.

As shown in fig. 12, the apparatus 1200 includes a computing unit 1201 which can perform various appropriate actions and processes in accordance with a computer program stored in a Read Only Memory (ROM) 1202 or a computer program loaded from a storage unit 1208 into a Random Access Memory (RAM) 1203. In the ram.1203, various programs and data required for the operation of the device 1200 may also be stored. The computing unit 1201, the ROM 1202, and the RAM 1203 are connected to each other by a bus 1204. An input/output (I/O) interface 1205 is also connected to bus 1204.

Various components in the device 1200 are connected to the I/O interface 1205 including: an input unit 1206 such as a keyboard, a mouse, or the like; an output unit 1207 such as various types of displays, speakers, and the like; a storage unit 1208, such as a magnetic disk, optical disk, or the like; and a communication unit 1209 such as a network card, modem, wireless communication transceiver, etc. The communication unit 1209 allows the device 1200 to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.

The computing unit 1201 may be a variety of general purpose and/or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 1201 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The computing unit 1201 performs the various methods and processes described above, such as a method of determining mapping information and/or an advertisement recommendation method. For example, in some embodiments, the method of determining mapping information and/or the advertisement recommendation method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 1208. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 1200 via the ROM 1202 and/or the communication unit 1209. When the computer program is loaded into RAM 1203 and executed by computing unit 1201, one or more steps of the above-described method of determining mapping information and/or advertisement recommendation method may be performed. Alternatively, in other embodiments, the computing unit 1201 may be configured by any other suitable means (e.g., by means of firmware) to perform the method of determining mapping information and/or the advertisement recommendation method.

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), system on a chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the Internet.

The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present disclosure may be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved, and the present disclosure is not limited herein.

The above detailed description should not be construed as limiting the scope of the disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present disclosure should be included in the scope of protection of the present disclosure.

Claims (23)

1. A method of determining mapping information, comprising:

determining at least one text message related to historical preference behaviors of a reference user within a predetermined period of time;

determining at least one information pair according to the similarity between the at least one text message and the at least one keyword, wherein each information pair comprises corresponding target text message and target keyword; wherein the at least one keyword corresponds to at least one original advertisement; and

and aiming at the same information pair, constructing the relation between the target text information and the target original advertisement corresponding to the target key words to obtain first mapping information in the mapping information.

2. The method of claim 1, wherein said determining at least one information pair based on a similarity between said at least one text information and at least one keyword comprises:

determining a first similarity between the at least one text message and the at least one keyword; and

and under the condition that the first similarity between the text information and the keywords is determined to be more than or equal to a first similarity threshold, respectively determining the text information and the keywords as target text information and target keywords in the same information pair.

3. The method of claim 1, wherein said determining at least one information pair based on a similarity between said at least one text information and at least one keyword comprises:

clustering a plurality of keywords corresponding to the same original advertisement to obtain at least one synonymous cluster and a centroid corresponding to each synonymous cluster, wherein each synonymous cluster comprises at least one keyword;

determining, for a target synonymous cluster of the at least one synonymous cluster, a second similarity between each of the at least one text message and a centroid of the target synonymous cluster; and

determining the text information as the target text information and determining each keyword in the target synonymous cluster as a target keyword corresponding to the target text information under the condition that the second similarity between the text information and the centroid of the target synonymous cluster is greater than or equal to a second similarity threshold value.

4. The method of claim 3, further comprising: after at least one of the synonymous clusters is obtained,

and determining the synonymy cluster with the keyword number being greater than or equal to the number threshold value as the target synonymy cluster.

5. The method of claim 1, wherein the number of reference users is plural, the method further comprising:

constructing a point-edge graph according to a plurality of reference users and the at least one piece of text information; the point-edge graph comprises a plurality of user nodes corresponding to the reference users and at least one text node corresponding to the text information, the user nodes corresponding to the target reference users and the text nodes corresponding to the target text information are connected through edges, and the target text information is related to historical preference behaviors of the target reference users;

determining at least one user pair from the plurality of reference users according to the point-edge graph; and

and aiming at the same user pair, constructing a relation between the user identification of one reference user and the original advertisement related to the other reference user to obtain second mapping information in the mapping information.

6. An advertisement recommendation method comprising:

acquiring current user information of a current user;

determining candidate advertisements from at least one original advertisement according to the current user information and the mapping information to obtain a candidate advertisement set; and

performing advertisement recommendation according to the candidate advertisement set;

wherein the mapping information is obtained according to the method of any one of claims 1 to 5.

7. The method of claim 6, wherein the current user information includes historical search terms, the mapping information including first mapping information characterizing a relationship between textual information and an original advertisement;

determining candidate advertisements from at least one original advertisement according to the current user information and the mapping information to obtain a candidate advertisement set, wherein the candidate advertisement set comprises:

repeatedly performing the following operations until the number of the at least one original advertisement is 0:

determining a current original advertisement from the at least one original advertisement; wherein the relationship between the current original advertisement and the current text information conforms to the first mapping information;

in response to the fact that the historical search terms are not consistent with the keywords corresponding to the at least one original advertisement and the historical search terms are consistent with the current text information, adding the current original advertisement as the candidate advertisement to the candidate advertisement set; and

deleting the current original advertisement from the at least one original advertisement.

8. The method of claim 6, wherein the current user information includes a current user identification; the mapping information comprises second mapping information representing the relationship between the user identification of the reference user and the original advertisement;

determining candidate advertisements from at least one original advertisement according to the current user information and the mapping information to obtain a candidate advertisement set, wherein the candidate advertisement set comprises:

repeatedly performing the following operations until the number of the at least one original advertisement is 0:

determining a current original advertisement from the at least one original advertisement; wherein the relationship between the current original advertisement and the user identifier of the current reference user conforms to the second mapping information;

in response to the fact that the historical search words are inconsistent with the keywords corresponding to the at least one original advertisement and the current user identification is consistent with the user identification of the current reference user, adding the current original advertisement as the candidate advertisement to the candidate advertisement set; and

deleting the current original advertisement from the at least one original advertisement.

9. The method of claim 6, wherein the current user information includes historical search terms, the method further comprising:

and in response to the fact that the historical search words are consistent with the keywords corresponding to the at least one original advertisement, adding the original advertisement corresponding to the keywords consistent with the historical search words as the candidate advertisement to the candidate advertisement set.

10. The method of claim 6, wherein the current user information includes historical search terms, the method further comprising:

and recommending the original advertisement corresponding to the keyword consistent with the historical search word in response to that the historical search word is consistent with the keyword corresponding to the at least one original advertisement and the interval duration between the search time of the historical search word and the current time is less than or equal to a preset duration.

11. An apparatus to determine mapping information, comprising:

the text determining module is used for determining at least one piece of text information related to historical preference behaviors of a reference user in a preset time period;

the information pair determining module is used for determining at least one information pair according to the similarity between the at least one text message and the at least one keyword, and each information pair comprises corresponding target text message and target keyword; wherein the at least one keyword corresponds to at least one original advertisement; and

and the first mapping information determining module is used for constructing a relation between the target text information and the target original advertisement corresponding to the target keyword aiming at the same information pair to obtain first mapping information in the mapping information.

12. The apparatus of claim 11, wherein the information pair determination module comprises:

a first determining sub-module, configured to determine a first similarity between the at least one text message and the at least one keyword; and

and the second determining submodule is used for respectively determining the text information and the keywords as the target text information and the target keywords in the same information pair under the condition that the first similarity between the text information and the keywords is determined to be more than or equal to a first similarity threshold value.

13. The apparatus of claim 11, wherein the information pair determination module comprises:

the clustering sub-module is used for clustering a plurality of keywords corresponding to the same original advertisement to obtain at least one synonymous cluster and a centroid corresponding to each synonymous cluster, wherein each synonymous cluster comprises at least one keyword;

a third determining sub-module, configured to determine, for a target synonymous cluster of the at least one synonymous cluster, a second similarity between each of the at least one text message and a centroid of the target synonymous cluster; and

and the fourth determining sub-module is used for determining the text information as the target text information and determining each keyword in the target synonymous cluster as a target keyword corresponding to the target text information under the condition that the second similarity between the text information and the centroid of the target synonymous cluster is greater than or equal to a second similarity threshold value.

14. The apparatus of claim 13, further comprising:

and the target cluster determining module is used for determining the synonym cluster with the keyword number being more than or equal to the number threshold as the target synonym cluster after at least one synonym cluster is obtained.

15. The apparatus of claim 11, wherein the number of reference users is plural, the apparatus further comprising:

the graph building module is used for building a point-edge graph according to a plurality of reference users and the at least one piece of text information; the point edge graph comprises a plurality of user nodes corresponding to the reference users and at least one text node corresponding to the text information, the user node corresponding to the target reference user is connected with the text node corresponding to the target text information through edges, and the target text information is related to historical preference behaviors of the target reference user;

a user pair determining module, configured to determine at least one user pair from the multiple reference users according to the dotted edge graph; and

and the second mapping information determining module is used for constructing a relation between the user identification of one reference user and the original advertisement related to another reference user aiming at the same user pair to obtain second mapping information in the mapping information.

16. An advertisement recommendation device comprising:

the acquisition module is used for acquiring the current user information of the current user;

the set determining module is used for determining candidate advertisements from at least one original advertisement according to the current user information and the mapping information to obtain a candidate advertisement set;

the first recommending module is used for recommending advertisements according to the candidate advertisement set;

wherein the mapping information is obtained by the apparatus according to any one of claims 11 to 15.

17. The apparatus of claim 16, wherein the current user information comprises historical search terms, the mapping information comprising first mapping information characterizing a relationship between textual information and an original advertisement;

the set determination module includes:

a first execution sub-module, configured to repeatedly execute the following operations until the number of the at least one original advertisement is 0:

determining a current original advertisement from the at least one original advertisement; wherein the relationship between the current original advertisement and the current text information conforms to the first mapping information;

in response to that the historical search terms are inconsistent with the keywords corresponding to the at least one original advertisement and the historical search terms are consistent with the current text information, adding the current original advertisement as the candidate advertisement to the candidate advertisement set; and

deleting the current original advertisement from the at least one original advertisement.

18. The apparatus of claim 16, wherein the current user information comprises a current user identification; the mapping information comprises second mapping information representing the relationship between the user identification of the reference user and the original advertisement;

the set determination module includes:

a second execution sub-module, configured to repeatedly execute the following operations until the number of the at least one original advertisement is 0:

determining a current original advertisement from the at least one original advertisement; the relation between the current original advertisement and the user identification of the current reference user accords with the second mapping information;

in response to the fact that the historical search words are inconsistent with the keywords corresponding to the at least one original advertisement and the current user identification is consistent with the user identification of the current reference user, adding the current original advertisement as the candidate advertisement to the candidate advertisement set; and

deleting the current original advertisement from the at least one original advertisement.

19. The apparatus of claim 16, wherein the current user information comprises historical search terms, the apparatus further comprising:

and the adding module is used for responding to the consistency of the historical search words and the keywords corresponding to the at least one original advertisement, and adding the original advertisement corresponding to the keywords consistent with the historical search words into the candidate advertisement set as the candidate advertisement.

20. The apparatus of claim 16, wherein the current user information comprises historical search terms, the apparatus further comprising:

and the second recommending module is used for recommending the original advertisement corresponding to the keyword consistent with the historical search word in response to the fact that the historical search word is consistent with the keyword corresponding to the at least one original advertisement and the interval duration between the search time of the historical search word and the current time is less than or equal to the preset duration.

21. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,

the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 10.

22. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1 to 10.

23. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 10.

Images