CN115829653A - Method, device, equipment and medium for determining relevancy of advertisement text - Google Patents

Method, device, equipment and medium for determining relevancy of advertisement text Download PDF

Info

Publication number
CN115829653A
CN115829653A CN202211541802.4A CN202211541802A CN115829653A CN 115829653 A CN115829653 A CN 115829653A CN 202211541802 A CN202211541802 A CN 202211541802A CN 115829653 A CN115829653 A CN 115829653A
Authority
CN
China
Prior art keywords
advertisement
text
texts
determining
edge
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211541802.4A
Other languages
Chinese (zh)
Inventor
谭云飞
刘晓庆
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Baidu Netcom Science and Technology Co Ltd
Original Assignee
Beijing Baidu Netcom Science and Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Baidu Netcom Science and Technology Co Ltd filed Critical Beijing Baidu Netcom Science and Technology Co Ltd
Priority to CN202211541802.4A priority Critical patent/CN115829653A/en
Publication of CN115829653A publication Critical patent/CN115829653A/en
Pending legal-status Critical Current

Links

Images

Abstract

The disclosure provides a method, a device, equipment and a medium for determining relevancy of advertisement texts, and relates to the technical field of artificial intelligence, in particular to the technical field of natural language processing. The implementation scheme is as follows: acquiring first and second feature vector representations corresponding to the first and second advertisement texts, respectively; and determining a degree of correlation between the first and second advertisement texts based on the first and second feature vector representations, wherein the feature vector representation of the advertisement text is obtained by using a determination process as follows: acquiring historical advertisement data; acquiring revenue information of the advertisement text; determining at least one associated text corresponding to each advertisement text based on a co-occurrence relation among the plurality of advertisement texts in the historical advertisement data; and determining a feature vector representation for each advertisement text based on semantic and revenue information for each advertisement text and its corresponding associated text.

Description

Method, device, equipment and medium for determining relevancy of advertisement text
Technical Field
The present disclosure relates to the field of artificial intelligence technologies, and in particular, to a method and an apparatus for determining relevancy of advertisement text, an electronic device, a computer-readable storage medium, and a computer program product.
Background
Artificial intelligence is the subject of research that makes computers simulate some human mental processes and intelligent behaviors (such as learning, reasoning, thinking, planning, etc.), both at the hardware level and at the software level. Artificial intelligence hardware technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing, and the like; the artificial intelligence software technology mainly comprises a computer vision technology, a voice recognition technology, a natural language processing technology, machine learning/deep learning, a big data processing technology, a knowledge map technology and the like.
In an advertising marketing scenario, it is often desirable to actively recommend advertisements to users, particularly advertisements related to particular content based on the particular content. Therefore, it is necessary to decide the recommendation policy based on the degree of correlation of the candidate content with the specific content.
The approaches described in this section are not necessarily approaches that have been previously conceived or pursued. Unless otherwise indicated, it should not be assumed that any of the approaches described in this section qualify as prior art merely by virtue of their inclusion in this section. Similarly, unless otherwise indicated, the problems mentioned in this section should not be considered as having been acknowledged in any prior art.
Disclosure of Invention
The present disclosure provides a method, an apparatus, an electronic device, a computer-readable storage medium, and a computer program product for determining relevancy of advertisement text.
According to an aspect of the present disclosure, there is provided a method for determining relevancy of advertisement text, including: acquiring a first feature vector representation corresponding to a first advertisement text and a second feature vector representation corresponding to a second advertisement text; and determining a correlation degree between the first advertisement text and the second advertisement text based on the first feature vector representation and the second feature vector representation, wherein the first feature vector representation and the second feature vector representation are obtained by using a determination process as follows: obtaining historical advertisement data containing a plurality of advertisement texts, wherein the plurality of advertisement texts comprise the first advertisement text and the second advertisement text; acquiring revenue information of each advertisement text in the plurality of advertisement texts; for each advertisement text in the plurality of advertisement texts, determining at least one associated text corresponding to the advertisement text from at least one other advertisement text except the advertisement text based on a co-occurrence relationship among the plurality of advertisement texts in the historical advertisement data; and for each advertisement text in the plurality of advertisement texts, determining a feature vector representation corresponding to the advertisement text based on the semantic and revenue information of the advertisement text and the semantic and revenue information corresponding to the corresponding at least one associated text.
According to another aspect of the present disclosure, there is also provided an advertisement text recommendation method including: acquiring a target advertisement text and a plurality of candidate advertisement texts; determining the relevancy between the target advertisement text and the candidate advertisement texts by using the relevancy determination method of the advertisement texts; and determining at least one advertisement text to be recommended from the candidate advertisement texts based on the correlation degree between the target advertisement text and the candidate advertisement texts.
According to another aspect of the present disclosure, there is provided an advertisement text relevancy determination apparatus including: a first acquisition unit configured to acquire a first feature vector representation corresponding to a first advertisement text and a second feature vector representation corresponding to a second advertisement text; and a first determination unit configured to determine a degree of correlation between the first advertisement text and the second advertisement text based on the first feature vector representation and a second feature vector representation, wherein the first feature vector representation and the second feature vector representation are obtained by a feature vector determination unit, the feature vector determination unit including: a first acquisition subunit configured to acquire historical advertisement data containing a plurality of advertisement texts, the plurality of advertisement texts including the first advertisement text and a second advertisement text; a second acquisition subunit configured to acquire revenue information of each of the plurality of advertisement texts; a first determining subunit configured to determine, for each of the plurality of advertisement texts, at least one associated text corresponding to the advertisement text from at least one other advertisement text other than the advertisement text based on a co-occurrence relationship between the plurality of advertisement texts in the historical advertisement data; and a second determining subunit configured to determine, for each of the plurality of advertisement texts, a feature vector representation corresponding to the advertisement text based on the semantic and revenue information of the advertisement text and the semantic and revenue information corresponding to the respective at least one associated text.
According to another aspect of the present disclosure, there is also provided an advertisement text recommending apparatus including: a second acquisition unit configured to acquire a target advertisement text and a plurality of candidate advertisement texts; the relevancy determination apparatus of advertisement texts as described above, configured to determine relevancy between the target advertisement text and the plurality of candidate advertisement texts; and a second determining unit configured to determine at least one advertisement text to be recommended from the plurality of candidate advertisement texts based on the correlation between the target advertisement text and the plurality of candidate advertisement texts.
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 method of relevancy determination of advertisement text described above.
According to another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the above-described advertisement text relevance determination method.
According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program is capable of implementing the above method for determining relevancy of advertisement text when executed by a processor.
According to one or more embodiments of the present disclosure, the degree of correlation between advertisement texts can be determined more accurately.
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 accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the embodiments and, together with the description, serve to explain the exemplary implementations of the embodiments. The illustrated embodiments are for purposes of illustration only and do not limit the scope of the claims. Throughout the drawings, identical reference numbers designate similar, but not necessarily identical, elements.
FIG. 1 shows a schematic diagram of an exemplary system in which various methods described herein may be implemented, according to an exemplary embodiment of the present disclosure;
FIG. 2 shows a flow chart of a process of determining a feature vector representation of advertisement text according to an example embodiment of the present disclosure;
FIG. 3 illustrates a flowchart of a relevancy determination method of advertisement text according to an exemplary embodiment of the present disclosure;
FIG. 4 shows a schematic diagram of a text diagram according to an example embodiment of the present disclosure;
fig. 5 illustrates a block diagram of a feature vector determination unit according to an exemplary embodiment of the present disclosure;
fig. 6 is a block diagram illustrating a structure of an advertisement text relevancy determination apparatus according to an exemplary embodiment of the present disclosure;
FIG. 7 illustrates a block diagram of an exemplary electronic device that can be used to implement embodiments 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 embodiments of the present 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 of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
In the present disclosure, unless otherwise specified, the use of the terms "first", "second", etc. to describe various elements is not intended to limit the positional relationship, the timing relationship, or the importance relationship of the elements, and such terms are used only to distinguish one element from another. In some examples, a first element and a second element may refer to the same instance of the element, and in some cases, based on the context, they may also refer to different instances.
The terminology used in the description of the various described examples in this disclosure is for the purpose of describing particular examples only and is not intended to be limiting. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the elements may be one or more. Furthermore, the term "and/or" as used in this disclosure is intended to encompass any and all possible combinations of the listed items.
In the related art, one implementation manner is to determine the relevance of advertisement texts by using semantic similarity between the advertisement texts, or alternatively, determine the relevance of the advertisement texts by using co-occurrence relationship between the advertisement texts in historical data. However, these two methods have limited expression capability for the correlation characteristics between a plurality of advertisement texts, and cannot take into account the revenue information corresponding to the advertisement texts.
Based on the above, the disclosure provides a method for determining relevancy of advertisement texts, which is implemented by constructing a text graph by using co-occurrence relations among a plurality of advertisement texts in historical advertisement data, determining feature vector representations of the advertisement texts corresponding to each node by using semantics and revenue information of each node and associated nodes in the text graph, and further determining relevancy among the advertisement texts by using the feature vector representations.
Embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.
Fig. 1 illustrates a schematic diagram of an exemplary system 100 in which various methods and apparatus described herein may be implemented in accordance with embodiments of the present disclosure. Referring to fig. 1, the system 100 includes one or more client devices 101, 102, 103, 104, 105, and 106, a server 120, and one or more communication networks 110 coupling the one or more client devices to the server 120. Client devices 101, 102, 103, 104, 105, and 106 may be configured to execute one or more applications.
In embodiments of the present disclosure, server 120 may run one or more services or software applications that enable the relevancy determination methods for ad text to be performed.
In some embodiments, the server 120 may also provide other services or software applications, which may include non-virtual environments and virtual environments. In certain embodiments, these services may be provided as web-based services or cloud services, for example, provided to users of client devices 101, 102, 103, 104, 105, and/or 106 under a software as a service (SaaS) model.
In the configuration shown in fig. 1, server 120 may include one or more components that implement the functions performed by server 120. These components may include software components, hardware components, or a combination thereof, which may be executed by one or more processors. A user operating a client device 101, 102, 103, 104, 105, and/or 106 may, in turn, utilize one or more client applications to interact with the server 120 to take advantage of the services provided by these components. It should be understood that a variety of different system configurations are possible, which may differ from system 100. Accordingly, fig. 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.
The user may use client devices 101, 102, 103, 104, 105, and/or 106 to send advertisement text. The client device may provide an interface that enables a user of the client device to interact with the client device. The client device may also output information to the user via the interface. Although fig. 1 depicts only six client devices, those skilled in the art will appreciate that any number of client devices may be supported by the present disclosure.
Client devices 101, 102, 103, 104, 105, and/or 106 may include various classes of computer devices, such as portable handheld devices, general purpose computers (such as personal computers and laptops), workstation computers, wearable devices, smart screen devices, self-service terminal devices, service robots, gaming systems, thin clients, various messaging devices, sensors, or other sensing devices, and so forth. These computer devices may run various classes and versions of software applications and operating systems, such as MICROSOFT Windows, APPLE iOS, UNIX-like operating systems, linux, or Linux-like operating systems (e.g., GOOGLE Chrome OS); or include various Mobile operating systems such as MICROSOFT Windows Mobile OS, iOS, windows Phone, android. Portable handheld devices may include cellular telephones, smart phones, tablets, personal Digital Assistants (PDAs), and the like. Wearable devices may include head-mounted displays (such as smart glasses) and other devices. The gaming system may include a variety of handheld gaming devices, internet-enabled gaming devices, and the like. The client device is capable of executing a variety of different applications, such as various Internet-related applications, communication applications (e.g., email applications), short Message Service (SMS) applications, and may use a variety of communication protocols.
Network 110 may be any type of network known to those skilled in the art that may support data communications using any of a variety of available protocols, including but not limited to TCP/IP, SNA, IPX, etc. Merely by way of example, one or more networks 110 may be a Local Area Network (LAN), an ethernet-based network, a token ring, a Wide Area Network (WAN), the internet, a virtual network, a Virtual Private Network (VPN), an intranet, an extranet, a blockchain network, a Public Switched Telephone Network (PSTN), an infrared network, a wireless network (e.g., bluetooth, WIFI), and/or any combination of these and/or other networks.
The server 120 may include one or more general purpose computers, special purpose server computers (e.g., PC (personal computer) servers, UNIX servers, mid-end servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and/or combination. The server 120 may include one or more virtual machines running a virtual operating system, or other computing architecture involving virtualization (e.g., one or more flexible pools of logical storage that may be virtualized to maintain virtual storage for the server). In various embodiments, the server 120 may run one or more services or software applications that provide the functionality described below.
The computing units in server 120 may run one or more operating systems including any of the operating systems described above, as well as any commercially available server operating systems. The server 120 may also run any of a variety of additional server applications and/or middle tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, and the like.
In some implementations, the server 120 may include one or more applications to analyze and consolidate data feeds and/or event updates received from users of the client devices 101, 102, 103, 104, 105, and 106. Server 120 may also include one or more applications to display data feeds and/or real-time events via one or more display devices of client devices 101, 102, 103, 104, 105, and 106.
In some embodiments, the server 120 may be a server of a distributed system, or a server incorporating a blockchain. The server 120 may also be a cloud server, or a smart cloud computing server or a smart cloud host with artificial intelligence technology. The cloud Server is a host product in a cloud computing service system, and is used for solving the defects of high management difficulty and weak service expansibility in the traditional physical host and Virtual Private Server (VPS) service.
The system 100 may also include one or more databases 130. In some embodiments, these databases may be used to store data and other information. For example, one or more of the databases 130 may be used to store information such as audio files and video files. The database 130 may reside in various locations. For example, the database used by the server 120 may be local to the server 120, or may be remote from the server 120 and may communicate with the server 120 via a network-based or dedicated connection. The database 130 may be of different categories. In certain embodiments, the database used by the server 120 may be, for example, a relational database. One or more of these databases may store, update, and retrieve data to and from the database in response to the command.
In some embodiments, one or more of the databases 130 may also be used by applications to store application data. The databases used by the application may be different classes of databases, such as key-value stores, object stores, or conventional stores supported by a file system.
The system 100 of fig. 1 may be configured and operated in various ways to enable application of the various methods and apparatus described in accordance with the present disclosure.
Fig. 2 shows a flow diagram of a process 200 for determining a feature vector representation of advertisement text according to an example embodiment of the present disclosure. As shown in fig. 2, process 200 includes:
step S201, obtaining historical advertisement data containing a plurality of advertisement texts, wherein the plurality of advertisement texts comprise a first advertisement text and a second advertisement text;
step S202, obtaining revenue information of each advertisement text in the advertisement texts;
step S203, aiming at each advertisement text in the plurality of advertisement texts, determining at least one associated text corresponding to the advertisement text from at least one other advertisement text except the advertisement text based on the co-occurrence relationship among the plurality of advertisement texts in the historical advertisement data; and
step S204, aiming at each advertisement text in the plurality of advertisement texts, determining a feature vector representation corresponding to the advertisement text based on the semantic and income information of the advertisement text and the semantic and income information corresponding to the corresponding at least one associated text.
Fig. 3 illustrates a flowchart of a relevancy determination method 300 of advertisement text according to an exemplary embodiment of the present disclosure. As shown in fig. 3, the method 300 includes:
step S301, acquiring a first feature vector representation corresponding to a first advertisement text and a second feature vector representation corresponding to a second advertisement text, wherein the first feature vector representation and the second feature vector representation are obtained by using the process 200; and
step S302, based on the first feature vector representation and the second feature vector representation, determining the correlation degree between the first advertisement text and the second advertisement text.
Thus, based on the above determination process 200, the associated text corresponding to each advertisement text can be determined by using the co-occurrence relationship among a plurality of advertisement texts in the historical advertisement data, and further, based on the semantics and revenue information of each advertisement text and its associated text, the feature vector representation of each advertisement text is determined, so that the feature vector representation of the advertisement text can more accurately represent the correlation features of the advertisement text and other advertisement texts, and at the same time, can represent the revenue information of the advertisement text. Furthermore, the relevance among the advertisement texts is determined by utilizing the feature vector representation, so that the relevance among the advertisement texts can be further determined by combining the profit information on the basis of the association relationship among the advertisement texts indicated by the text co-occurrence relationship, and the accuracy is improved.
In some examples, the advertisement text may be, for example, advertisement name text recommended to the user or commodity name text actively queried by the user, as long as the relevant information of the advertisement marketing object can be characterized, which is not limited by the present disclosure.
In some examples, when the advertisement text is an advertisement title text recommended to the user, the revenue information for the advertisement text may be determined based on a fee for the advertisement. It should be understood that the revenue information of the advertisement text may also include other content, for example, revenue information corresponding to the marketing object represented by the advertisement text, such as sales amount of goods, browsing amount of pages, playing amount of videos, etc., which is not limited by the present disclosure.
In some examples, determining the degree of correlation between the first advertisement text and the second advertisement text based on the first feature vector representation and the second feature vector representation in step S302 may include: determining a degree of correlation between the first advertisement text and the second advertisement text based on a degree of similarity between the first feature vector representation and the second feature vector representation. The similarity may be determined by calculating a euclidean distance between the first eigenvector representation and the second eigenvector representation, or may be determined by calculating indexes such as cosine similarity and manhattan distance, for example.
In some examples, after determining the feature vector representations of the plurality of advertisement texts by using the process 200, the plurality of advertisement texts and the corresponding plurality of feature vector representations may be stored in a database, so that the corresponding first feature vector representation and second feature vector representation can be queried from the database based on the first advertisement text and the second advertisement text in step S301, thereby improving the efficiency of relevance determination of the advertisement texts.
According to some embodiments, the determining, in step S203, for each of the plurality of advertisement texts, at least one associated text corresponding to the advertisement text from at least one other advertisement text except the advertisement text based on a co-occurrence relationship between the plurality of advertisement texts in the historical advertisement data includes: constructing a text graph containing a plurality of nodes corresponding to the plurality of advertisement texts one to one based on the co-occurrence relationship among the plurality of advertisement texts in the historical advertisement data, wherein aiming at any two advertisement texts of the plurality of texts, in response to the fact that the co-occurrence relationship between the two advertisement texts meets a preset condition, a connecting edge is established based on the nodes corresponding to the two advertisement texts; and for each node in the plurality of nodes, determining at least one associated node corresponding to the node from at least one other node based on the connection relation among the plurality of nodes so as to obtain at least one associated text corresponding to the corresponding advertisement text.
Therefore, based on the above determination process 200, a text graph can be constructed by using the co-occurrence relationship among a plurality of advertisement texts in the historical advertisement data, and the feature vector representation of the advertisement text corresponding to each node is determined by using the semantic and revenue information of each node and the associated node in the text graph, so that the feature vector representation of the advertisement text can more accurately represent the correlation features of the advertisement text and other advertisement texts, and simultaneously can represent the revenue information of the advertisement text. Furthermore, the relevance among the advertisement texts is determined by utilizing the characteristic vector representation, so that the relevance among the advertisement texts can be further determined by combining the profit information on the basis of the incidence relation among the texts represented by the graph structure of the text graph, and the accuracy is improved.
According to some embodiments, determining in step 204, for each advertisement text of the plurality of advertisement texts, a feature vector representation of the advertisement text based on the semantic and revenue information of the advertisement text and the semantic and revenue information of the corresponding at least one associated text comprises: determining an edge vector representation of each connecting edge in at least one connecting edge between a node corresponding to the advertisement text and at least one corresponding associated node based on semantic and revenue information of the advertisement text respectively corresponding to two endpoints of the connecting edge; and determining a feature vector representation of the advertisement text based on the edge vector representation of the at least one connecting edge. Therefore, the relevance characteristics of the advertisement texts corresponding to the two end points of each edge can be represented by using the edge vector representation, and then a plurality of edge vectors between each node and the associated node are aggregated to obtain more accurate characteristic vector representation of the texts corresponding to each node.
In some examples, the determining, based on the edge vector representation of the at least one connecting edge, the feature vector representation of the advertisement text corresponding to the node may include: and calculating the average value of the edge vector representation of the at least one connecting edge to obtain the characteristic vector representation of the advertisement text corresponding to the node. This step may also be performed in other manners, for example, it may be calculated based on the edge vector representation of the at least one connecting edge and a preset formula to obtain a feature vector representation of the advertisement text corresponding to the node.
According to some embodiments, the determining, based on semantic and revenue information of the advertisement text respectively corresponding to the two endpoints of the connected edge, the edge vector representation of the connected edge includes: determining semantic similarity between the advertisement texts respectively corresponding to the two endpoints; and determining the edge vector representation of the connecting edge based on the semantic similarity and the revenue information of the advertisement texts respectively corresponding to the two endpoints. Therefore, the semantic similarity can be used for representing the correlation between the advertisement texts respectively corresponding to the two endpoints, and the method is more convenient and accurate.
In some examples, semantic feature vector representations of the advertisement texts corresponding to the two endpoints respectively are determined, and the semantic similarity is determined based on the similarity between the two semantic feature vector representations. The semantic feature vector representation may be implemented by inputting advertisement text into a language model, or may be implemented by querying a database storing a plurality of texts and corresponding semantic feature vectors, which is not limited by the disclosure.
According to some embodiments, the determining semantic similarity between the advertisement texts respectively corresponding to the two endpoints includes: and inputting the advertisement texts respectively corresponding to the two end points of the connecting edge into a pre-training language model to obtain the semantic similarity output by the pre-training language model, wherein the pre-training language model is obtained by training by using labeled corpus data. Therefore, the semantic similarity between the two advertisement texts can be obtained by utilizing the pre-training language model, and the efficiency and the accuracy are improved.
In some examples, the pre-trained language model may be, for example, an ernie model.
According to some embodiments, the determination process 200 further comprises: and for each connecting edge in at least one connecting edge between the node corresponding to the advertisement text and the corresponding at least one associated node, determining the co-occurrence frequency between the advertisement texts respectively corresponding to two end points of the connecting edge, and determining the edge vector representation of the connecting edge based on the semantic and revenue information of the advertisement texts respectively corresponding to the two end points of the connecting edge and the co-occurrence frequency. Therefore, the correlation between the two endpoints can be accurately indicated by combining the co-occurrence frequency of the advertisement texts corresponding to the two endpoints respectively.
In some examples, the co-occurrence frequency may be a number of co-occurrences cnt. Thus, the fusion information S of the number of co-occurrences cnt and the profit information acp can be determined based on the following formula:
S＝a*cnt+b*acp
in this example, a and b in the formula may be weight values set according to actual demands.
According to some embodiments, the determination process 200 further comprises: and respectively executing normalization processing on the co-occurrence frequency and the profit information to obtain normalized co-occurrence frequency and normalized profit information, and determining edge vector representation of the connecting edge based on the semantic meaning and normalized profit information of the advertisement text respectively corresponding to the two end points of the connecting edge and the normalized co-occurrence frequency. Therefore, the numerical ranges of the co-occurrence frequency and the income information can be scaled to the same interval, the calculation process is simplified, and more accurate edge vector representation is obtained.
In some examples, normalization may be performed using the following formula:
Figure BDA0003978024380000111
where x is the initial revenue information or co-occurrence frequency, x min As a minimum value, x, of the frequency of co-occurrence or gain information in the overall data max Is the maximum value of the benefit information or the co-occurrence frequency in the whole data, and X' is normalized benefit information or normalized co-occurrence frequency.
According to some embodiments, the determining, based on semantic and revenue information of the advertisement text respectively corresponding to the two endpoints of the connected edge, the edge vector representation of the connected edge includes: inputting the advertisement texts and the revenue information thereof corresponding to the two endpoints of the connection edge into an edge vector coding model to obtain the edge vector representation output by the edge vector coding model, wherein the edge vector coding model is obtained by training in the following way: obtaining a sample text graph containing a plurality of nodes in one-to-one correspondence with a plurality of sample texts and profit information of each sample text in the plurality of sample texts, wherein the sample text graph comprises a plurality of connecting edges for connecting the plurality of nodes; for each connecting edge in a plurality of connecting edges included in a sample text graph, inputting sample texts and income information thereof corresponding to two end points of the connecting edge into the edge vector coding model to obtain an edge vector representation of the connecting plate output by the edge vector coding model; determining feature vector representations of a plurality of sample texts corresponding to the plurality of nodes based on the edge vector representations of the plurality of connecting edges; acquiring a real correlation degree between a first sample text and a second sample text in the plurality of sample texts; determining a prediction correlation between the first sample text and the second sample text based on a first feature vector representation and a second feature vector representation corresponding to the first sample text and the second sample text, respectively; and adjusting parameters of the edge vector coding model based on the true correlation and the prediction correlation. Therefore, the edge vector representation can be obtained by using the edge vector coding model, the relevance prediction task between texts can be executed by using the edge vector representation output by the edge vector coding model, and the model training is carried out based on the relevance prediction task, so that the efficiency and the accuracy are improved.
In some examples, the co-occurrence frequency of the advertisement texts corresponding to the two end points of the connection edge may also be input into the edge vector coding model at the same time, for example, the above-described fusion information S may be input into the edge vector coding model to obtain an edge vector representation capable of more accurately characterizing the relevancy between the two advertisement texts.
According to some embodiments, the plurality of advertisement texts comprise a plurality of historical query texts and a plurality of historical recommended texts, the historical advertisement data comprises a plurality of text pairs, each text pair in the plurality of text pairs comprises one historical query text and a historical recommended text recommended to the user based on the historical query text, and the establishing, in response to determining that the co-occurrence relationship between the two advertisement texts satisfies a preset condition, an edge of the text graph with a node corresponding to the two advertisement texts as a vertex comprises: in response to determining that the historical advertisement data includes a text pair consisting of the two advertisement texts, edges of a text graph are established with nodes corresponding to the two advertisement texts as vertices. Therefore, the relevance represented by the mapping relation between the historical query text and the historical recommendation text in the historical advertisement data can be fully utilized, and the edge of the text graph is constructed on the basis, so that the relevance between two endpoint nodes can be accurately indicated by the edge in the text graph.
In some examples, a frequency of co-occurrence between respective two advertisement texts may be determined based on a frequency of each text pair in the historical advertisement data.
According to some embodiments, the determining, for each node in the text graph, at least one associated node corresponding to the node from a plurality of other nodes based on the connection relationships between the plurality of nodes comprises: for each other node in the plurality of other nodes, determining that the other node is the associated node in response to the number of connection hops between the other node and the node not being greater than a preset threshold. Therefore, other nodes closer to the node can be determined as the associated nodes, the influence of the long-distance node on the feature vector representation of the advertisement text of the node is avoided, and the accuracy is improved.
According to some embodiments, the determining, for each node in the text graph, at least one associated node corresponding to the node from a plurality of other nodes based on the connection relationships between the plurality of nodes comprises: determining respective sampling probabilities of the other nodes based on a connection relation between the nodes and a preset rule, wherein according to the preset rule, the sampling probability of a node with a smaller connection hop count with the node is larger than the sampling probability of a node with a larger connection hop count with the node; and randomly sampling the plurality of other nodes based on the sampling probability to obtain the at least one associated node. Therefore, the associated nodes corresponding to each node can be obtained by utilizing layered random sampling, the sampling probability corresponding to the nodes with closer distances is higher, the number of the associated nodes is reduced, the vector calculation process is simplified, and meanwhile, the accuracy is guaranteed.
Fig. 4 shows a schematic diagram of a text diagram according to an exemplary embodiment of the present disclosure. In this example, for node a at the center, two layers of neighbor nodes associated therewith may be sampled, specifically, a node b, c, d with a connection hop count of one is a first layer node, and a node e, f, g with a connection hop count of two is a second layer node. The sampling frequency corresponding to each layer of nodes may decrease as the number of layers increases, and the sampling frequency corresponding to each layer of nodes may be determined based on an exponential decay function, for example.
In this example, an edge vector representation of edges a-b, a-c, a-d, b-e, d-f, d-g, respectively, may be obtained by utilizing the steps described above. In this way, the nodes of each layer can be aggregated, that is, the feature vector representation corresponding to the node of the layer can be determined based on the edge vector representation of the connecting edge between the node of the layer and the node of the next layer and the feature vector representation corresponding to the node of the next layer. For example, the feature vector representation corresponding to the node d may be determined based on the edge vector representations of the edges d-f and d-g, and then the feature vector representation of the advertisement text corresponding to the node a may be determined based on the edge vector representations of the edges a-b, a-c and a-d and the feature vector representations of the nodes b, c and d.
According to another aspect of the present disclosure, there is also provided an advertisement text recommendation method including: acquiring a target advertisement text and a plurality of candidate advertisement texts; determining the relevancy between the target advertisement text and the candidate advertisement texts by using the relevancy determination method of the advertisement texts; and determining at least one advertisement text to be recommended from the candidate advertisement texts based on the correlation degree between the target advertisement text and the candidate advertisement texts.
In some examples, the plurality of candidate advertisement texts may be ranked based on a correlation between the target advertisement text and the plurality of candidate advertisement texts, and at least one advertisement text to be recommended may be determined based on a ranking result.
According to another aspect of the present disclosure, an apparatus for determining relevancy of advertisement text is also provided. Fig. 5 illustrates a block diagram of a feature vector determination unit 500 according to an exemplary embodiment of the present disclosure. Fig. 6 shows a block diagram of a relevancy determination apparatus 600 for advertisement text according to an exemplary embodiment of the present disclosure.
As shown in fig. 5, the feature vector determination unit 500 includes:
a first obtaining subunit 501 configured to obtain historical advertisement data including a plurality of advertisement texts, the plurality of advertisement texts including the first advertisement text and a second advertisement text;
a second obtaining subunit 502 configured to obtain revenue information of each of the plurality of advertisement texts;
a first determining subunit 503 configured to determine, for each of the plurality of advertisement texts, at least one associated text corresponding to the advertisement text from at least one other advertisement text except the advertisement text based on a co-occurrence relationship between the plurality of advertisement texts in the historical advertisement data; and
a second determining subunit 504 configured to determine, for each advertisement text of the plurality of advertisement texts, a feature vector representation corresponding to the advertisement text based on the semantic and revenue information of the advertisement text and the semantic and revenue information corresponding to the respective at least one associated text.
The operations of the units 501 to 504 of the unit 500 are similar to those of the steps S201 to S204 described above, and are not described herein again.
As shown in fig. 6, the apparatus 600 for determining relevancy of advertisement text includes:
a first obtaining unit 601 configured to obtain a first feature vector representation corresponding to a first advertisement text and a second feature vector representation corresponding to a second advertisement text; and
a first determining unit 602 configured to determine a degree of correlation between the first advertisement text and the second advertisement text based on the first feature vector representation and the second feature vector representation, wherein the first feature vector representation and the second feature vector representation are obtained by means of unit 200.
The operations of the units 601-602 of the apparatus 600 are similar to the operations of the steps S301-S302 described above, and are not described herein again.
According to some embodiments, the first determining subunit comprises: a construction module configured to construct a text graph including a plurality of nodes corresponding to the plurality of advertisement texts one to one based on a co-occurrence relationship between the plurality of advertisement texts in the historical advertisement data, wherein for any two advertisement texts of the plurality of texts, in response to determining that the co-occurrence relationship between the two advertisement texts satisfies a preset condition, a connecting edge is established based on the nodes corresponding to the two advertisement texts; and a first determining module configured to determine, for each of the plurality of nodes, at least one associated node corresponding to the node from at least one other node based on the connection relationship between the plurality of nodes, to obtain at least one associated text corresponding to the corresponding advertisement text.
According to some embodiments, the second determining subunit comprises: a second determining module configured to determine, for each connecting edge of at least one connecting edge between a node corresponding to the advertisement text and a corresponding at least one associated node, an edge vector representation of the connecting edge based on semantic and revenue information of the advertisement text respectively corresponding to two endpoints of the connecting edge; and a third determination module configured to determine a feature vector representation of the advertisement text based on the edge vector representation of the at least one connecting edge.
According to some embodiments, the second determination module is configured to: determining semantic similarity between advertisement texts respectively corresponding to the two endpoints; and determining the edge vector representation of the connecting edge based on the semantic similarity and the revenue information of the advertisement texts respectively corresponding to the two endpoints.
According to some embodiments, the second determination module is configured to: and inputting the advertisement texts respectively corresponding to the two endpoints of the connecting edge into a pre-training language model to acquire the semantic similarity output by the pre-training language model, wherein the pre-training language model is obtained by training by using labeled corpus data.
According to some embodiments, the feature vector determination unit further comprises: a third determining subunit configured to determine, for each of at least one connection edge between the node corresponding to the advertisement text and the corresponding at least one associated node, a co-occurrence frequency between the advertisement texts corresponding to the two endpoints of the connection edge, respectively, and wherein the second determining module is configured to determine the edge vector representation of the connection edge based on the semantic and profit information of the advertisement texts corresponding to the two endpoints of the connection edge, respectively, and the co-occurrence frequency.
According to some embodiments, the feature vector determination unit further comprises: a processing subunit configured to perform normalization processing on the co-occurrence frequency and the benefit information respectively to obtain a normalized co-occurrence frequency and normalized benefit information, and wherein the first determining module is configured to determine an edge vector representation of the connecting edge based on the normalized co-occurrence frequency and the semantic and normalized benefit information of the advertisement text respectively corresponding to the two endpoints of the connecting edge.
According to some embodiments, the second determination module is configured to: inputting the advertisement texts and the revenue information thereof corresponding to the two endpoints of the connection edge into an edge vector coding model to obtain the edge vector representation output by the edge vector coding model, wherein the edge vector coding model is obtained by training in the following way: obtaining a sample text graph containing a plurality of nodes in one-to-one correspondence with a plurality of sample texts and profit information of each sample text in the plurality of sample texts, wherein the sample text graph comprises a plurality of connecting edges for connecting the plurality of nodes; for each connecting side in a plurality of connecting sides included in a sample text graph, inputting sample texts and income information thereof corresponding to two end points of the connecting side into the edge vector coding model to obtain an edge vector representation of the connecting plate output by the edge vector coding model; determining feature vector representations of a plurality of sample texts corresponding to the plurality of nodes based on the edge vector representations of the plurality of connecting edges; acquiring a real correlation degree between a first sample text and a second sample text in the plurality of sample texts; determining a prediction correlation between the first sample text and the second sample text based on a first feature vector representation and a second feature vector representation corresponding to the first sample text and the second sample text, respectively; and adjusting parameters of the edge vector coding model based on the true correlation and the prediction correlation.
According to some embodiments, the plurality of advertisement texts comprises a plurality of historical query texts and a plurality of historical recommendation texts, the historical advertisement data comprises a plurality of text pairs, each text pair of the plurality of text pairs comprises one historical query text and a historical recommendation text recommended to the user based on the historical query text, and wherein the construction module is configured to: in response to determining that the historical advertisement data includes a text pair consisting of the two advertisement texts, edges of a text graph are established with nodes corresponding to the two advertisement texts as vertices.
According to some embodiments, the first determination module is configured to: for each other node in the plurality of other nodes, determining that the other node is the associated node in response to the number of connection hops between the other node and the node not being greater than a preset threshold.
According to some embodiments, the first determination module is configured to: determining respective sampling probabilities of the other nodes based on a connection relation between the nodes and a preset rule, wherein according to the preset rule, the sampling probability of a node with a smaller connection hop count with the node is larger than the sampling probability of a node with a larger connection hop count with the node; and randomly sampling the plurality of other nodes based on the sampling probability to obtain the at least one associated node.
According to another aspect of the present disclosure, there is also provided an advertisement text recommendation apparatus including: a second acquisition unit configured to acquire a target advertisement text and a plurality of candidate advertisement texts; the relevancy determination apparatus of advertisement texts as described above, configured to determine relevancy between the target advertisement text and the plurality of candidate advertisement texts; and a second determining unit configured to determine at least one advertisement text to be recommended from the plurality of candidate advertisement texts based on the correlation between the target advertisement text and the plurality of candidate advertisement texts.
According to another aspect of the present disclosure, there is also 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 method of relevancy determination of advertisement text described above.
According to another aspect of the present disclosure, there is also provided a non-transitory computer readable storage medium storing computer instructions for causing the computer to execute the method for determining relevancy of advertisement text described above.
According to another aspect of the present disclosure, there is also provided a computer program product comprising a computer program, wherein the computer program when executed by a processor implements the method for determining relevancy of advertisement text as described above.
Referring to fig. 7, a block diagram of a structure of an electronic device 700, which may be a server or a client of the present disclosure, which is an example of a hardware device that may be applied to aspects of the present disclosure, will now be described. Electronic device is intended to represent various forms of digital electronic computer devices, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, 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. 7, the device 700 comprises a computing unit 701, which may perform various suitable actions and processes according to a computer program stored in a Read Only Memory (ROM) 702 or a computer program loaded from a storage unit 708 into a Random Access Memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other by a bus 704. An input/output (I/O) interface 705 is also connected to bus 704.
Various components in the device 700 are connected to the I/O interface 705, including: an input unit 706, an output unit 707, a storage unit 708, and a communication unit 709. The input unit 706 may be any type of device capable of inputting information to the device 700, and the input unit 706 may receive input numeric or character information and generate key signal inputs related to user settings and/or function controls of the electronic device, and may include, but is not limited to, a mouse, a keyboard, a touch screen, a track pad, a track ball, a joystick, a microphone, and/or a remote control. Output unit 707 may be any type of device capable of presenting information and may include, but is not limited to, a display, speakers, a video/audio output terminal, a vibrator, and/or a printer. Storage unit 708 may include, but is not limited to, magnetic or optical disks. The communication unit 709 allows the device 700 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and/or chipsets, such as bluetooth (TM) devices, 802.11 devices, wiFi devices, wiMax devices, cellular communication devices, and/or the like.
Computing unit 701 may be a variety of general and/or special purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated 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 calculation unit 701 executes the respective methods and processes described above, such as the relevancy determination method of advertisement text. For example, in some embodiments, the relevancy determination method for advertisement text may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of a computer program may be loaded onto and/or installed onto device 700 via ROM 702 and/or communications unit 709. When loaded into RAM 703 and executed by the computing unit 701, may perform one or more of the steps of the method for determining relevance of advertisement text described above. Alternatively, in other embodiments, the computing unit 701 may be configured by any other suitable means (e.g., by means of firmware) to perform the relevancy determination method for advertisement text.
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 may 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), the internet, and blockchain networks.
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. The server may be a cloud server, a server of a distributed system, or a server with a combined blockchain.
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 performed in parallel, sequentially or in different orders, and are not limited herein as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved.
Although embodiments or examples of the present disclosure have been described with reference to the accompanying drawings, it is to be understood that the above-described methods, systems and apparatus are merely exemplary embodiments or examples and that the scope of the present invention is not limited by these embodiments or examples, but only by the claims as issued and their equivalents. Various elements in the embodiments or examples may be omitted or may be replaced with equivalents thereof. Further, the steps may be performed in an order different from that described in the present disclosure. Further, various elements in the embodiments or examples may be combined in various ways. It is important that as technology evolves, many of the elements described herein may be replaced with equivalent elements that appear after the present disclosure.

Claims (27)

1. A method for determining relevancy of advertisement text comprises the following steps:
acquiring a first feature vector representation corresponding to a first advertisement text and a second feature vector representation corresponding to a second advertisement text; and
determining a degree of correlation between the first advertisement text and the second advertisement text based on the first feature vector representation and the second feature vector representation,
wherein the first feature vector representation and the second feature vector representation are obtained using a determination process as follows:
obtaining historical advertisement data containing a plurality of advertisement texts, wherein the plurality of advertisement texts comprise the first advertisement text and the second advertisement text;
acquiring revenue information of each advertisement text in the plurality of advertisement texts;
for each advertisement text in the plurality of advertisement texts, determining at least one associated text corresponding to the advertisement text from at least one other advertisement text except the advertisement text based on a co-occurrence relationship among the plurality of advertisement texts in the historical advertisement data; and
for each advertisement text in the plurality of advertisement texts, determining a feature vector representation corresponding to the advertisement text based on the semantic and revenue information of the advertisement text and the semantic and revenue information corresponding to the corresponding at least one associated text.
2. The method of claim 1 wherein, for each of the plurality of advertisement texts, determining at least one associated text corresponding to the advertisement text from at least one other advertisement text other than the advertisement text based on a co-occurrence relationship between the plurality of advertisement texts in the historical advertisement data comprises:
constructing a text graph containing a plurality of nodes corresponding to the plurality of advertisement texts one to one based on the co-occurrence relationship among the plurality of advertisement texts in the historical advertisement data, wherein aiming at any two advertisement texts of the plurality of texts, in response to the fact that the co-occurrence relationship between the two advertisement texts meets a preset condition, a connecting edge is established based on the nodes corresponding to the two advertisement texts; and
for each node in the plurality of nodes, determining at least one associated node corresponding to the node from at least one other node based on the connection relation among the plurality of nodes to obtain at least one associated text corresponding to the corresponding advertisement text.
3. The method of claim 2, wherein the determining, for each of the plurality of advertisement texts, a feature vector representation of the advertisement text based on the semantic and revenue information of the advertisement text and the semantic and revenue information of the corresponding at least one associated text comprises:
determining an edge vector representation of each connecting edge in at least one connecting edge between a node corresponding to the advertisement text and at least one corresponding associated node based on semantic and revenue information of the advertisement text respectively corresponding to two endpoints of the connecting edge; and
based on the edge vector representation of the at least one connecting edge, a feature vector representation of the advertisement text is determined.
4. The method of claim 3, wherein the determining the edge vector representation of the connected edge based on semantic and revenue information of the advertisement text corresponding to the two endpoints of the connected edge respectively comprises:
determining semantic similarity between advertisement texts respectively corresponding to the two endpoints; and
and determining the edge vector representation of the connecting edge based on the semantic similarity and the revenue information of the advertisement texts respectively corresponding to the two endpoints.
5. The method of claim 4, wherein the determining semantic similarity between the advertisement texts respectively corresponding to the two endpoints comprises:
and inputting the advertisement texts respectively corresponding to the two end points of the connecting edge into a pre-training language model to obtain the semantic similarity output by the pre-training language model, wherein the pre-training language model is obtained by training by using labeled corpus data.
6. The method of any of claims 3-5, wherein the determining further comprises:
determining, for each of at least one connecting edge between the node corresponding to the advertisement text and the corresponding at least one associated node, a co-occurrence frequency between the advertisement texts respectively corresponding to two endpoints of the connecting edge,
and determining the edge vector representation of the connecting edge based on the semantic and profit information of the advertisement texts respectively corresponding to the two endpoints of the connecting edge and the co-occurrence frequency.
7. The method of claim 6, wherein the determining further comprises:
respectively performing normalization processing on the co-occurrence frequency and the profit information to obtain normalized co-occurrence frequency and normalized profit information,
and determining the edge vector representation of the connecting edge based on the semantic and normalized profit information of the advertisement texts respectively corresponding to the two endpoints of the connecting edge and the normalized co-occurrence frequency.
8. The method of any one of claims 2-7, wherein the determining of the edge vector representation of the connecting edge based on semantic and revenue information of the advertisement text corresponding to the two endpoints of the connecting edge respectively comprises:
inputting the advertisement texts and the income information corresponding to the two end points of the connecting edge into an edge vector coding model to obtain the edge vector representation output by the edge vector coding model,
wherein, the edge vector coding model is obtained by training in the following way:
obtaining a sample text graph containing a plurality of nodes in one-to-one correspondence with a plurality of sample texts and profit information of each sample text in the plurality of sample texts, wherein the sample text graph comprises a plurality of connecting edges for connecting the plurality of nodes;
for each connecting side in a plurality of connecting sides included in a sample text graph, inputting sample texts and income information thereof corresponding to two end points of the connecting side into the edge vector coding model to obtain an edge vector representation of the connecting plate output by the edge vector coding model;
determining feature vector representations of a plurality of sample texts corresponding to the plurality of nodes based on the edge vector representations of the plurality of connecting edges;
acquiring a real correlation degree between a first sample text and a second sample text in the plurality of sample texts;
determining a prediction correlation degree between the first sample text and the second sample text based on a first feature vector representation and a second feature vector representation respectively corresponding to the first sample text and the second sample text; and
and adjusting parameters of the edge vector coding model based on the real correlation degree and the prediction correlation degree.
9. The method of any of claims 2-8, wherein the plurality of advertisement texts comprises a plurality of historical query texts and a plurality of historical recommendation texts, the historical advertisement data comprises a plurality of text pairs, each text pair of the plurality of text pairs comprises one historical query text and a historical recommendation text that is recommended to a user based on the historical query text,
and wherein, in response to determining that the co-occurrence relationship between the two advertisement texts meets a preset condition, establishing an edge of the text graph with the nodes corresponding to the two advertisement texts as vertices comprises:
in response to determining that the historical advertisement data includes a text pair consisting of the two advertisement texts, edges of a text graph are established with nodes corresponding to the two advertisement texts as vertices.
10. The method of any one of claims 2-9, wherein the determining, for each node in the text graph, at least one associated node from a plurality of other nodes corresponding to the node based on connection relationships between the plurality of nodes comprises:
for each other node in the plurality of other nodes, determining that the other node is the associated node in response to the number of connection hops between the other node and the node not being greater than a preset threshold.
11. The method of any one of claims 2-9, wherein the determining, for each node in the text graph, at least one associated node from a plurality of other nodes corresponding to the node based on connection relationships between the plurality of nodes comprises:
determining respective sampling probabilities of the other nodes based on a connection relation between the nodes and a preset rule, wherein according to the preset rule, the sampling probability of a node with a smaller connection hop count with the node is larger than the sampling probability of a node with a larger connection hop count with the node; and
randomly sampling the plurality of other nodes based on the sampling probability to obtain the at least one associated node.
12. An advertisement text recommendation method comprising:
acquiring a target advertisement text and a plurality of candidate advertisement texts;
determining a degree of correlation between the target advertisement text and the plurality of candidate advertisement texts using the method of any one of claims 1-11; and
and determining at least one advertisement text to be recommended from the candidate advertisement texts based on the correlation degree between the target advertisement text and the candidate advertisement texts.
13. An apparatus for determining relevancy of advertisement text, comprising:
a first acquisition unit configured to acquire a first feature vector representation corresponding to a first advertisement text and a second feature vector representation corresponding to a second advertisement text; and
a first determination unit configured to determine a degree of correlation between the first advertisement text and the second advertisement text based on the first feature vector representation and the second feature vector representation,
wherein the first feature vector representation and the second feature vector representation are obtained using a feature vector determination unit, the feature vector determination unit comprising:
a first acquisition subunit configured to acquire historical advertisement data containing a plurality of advertisement texts, the plurality of advertisement texts including the first advertisement text and a second advertisement text;
a second acquiring subunit configured to acquire revenue information of each of the plurality of advertisement texts;
a first determining subunit configured to determine, for each of the plurality of advertisement texts, at least one associated text corresponding to the advertisement text from at least one other advertisement text other than the advertisement text based on a co-occurrence relationship between the plurality of advertisement texts in the historical advertisement data; and
a second determining subunit configured to determine, for each of the plurality of advertisement texts, a feature vector representation corresponding to the advertisement text based on the semantic and revenue information of the advertisement text and the semantic and revenue information corresponding to the respective at least one associated text.
14. The apparatus of claim 13, wherein the first determining subunit comprises:
a construction module configured to construct a text graph including a plurality of nodes corresponding to the plurality of advertisement texts one to one based on a co-occurrence relationship between the plurality of advertisement texts in the historical advertisement data, wherein for any two advertisement texts of the plurality of texts, in response to determining that the co-occurrence relationship between the two advertisement texts satisfies a preset condition, a connecting edge is established based on the nodes corresponding to the two advertisement texts; and
a first determining module configured to determine, for each of the plurality of nodes, at least one associated node corresponding to the node from at least one other node based on a connection relationship between the plurality of nodes, to obtain at least one associated text corresponding to the corresponding advertisement text.
15. The apparatus of claim 14, wherein the second determining subunit comprises:
a second determining module configured to determine, for each connecting edge of at least one connecting edge between a node corresponding to the advertisement text and a corresponding at least one associated node, an edge vector representation of the connecting edge based on semantic and revenue information of the advertisement text respectively corresponding to two endpoints of the connecting edge; and
a third determination module configured to determine a feature vector representation of the advertisement text based on the edge vector representation of the at least one connecting edge.
16. The apparatus of claim 15, wherein the second determining module is configured to:
determining semantic similarity between advertisement texts respectively corresponding to the two endpoints; and
and determining the edge vector representation of the connecting edge based on the semantic similarity and the revenue information of the advertisement texts respectively corresponding to the two endpoints.
17. The apparatus of claim 16, wherein the second determining module is configured to:
and inputting the advertisement texts respectively corresponding to the two end points of the connecting edge into a pre-training language model to obtain the semantic similarity output by the pre-training language model, wherein the pre-training language model is obtained by training by using labeled corpus data.
18. The apparatus of any one of claims 15-17, wherein the feature vector determination unit further comprises:
a third determining subunit configured to determine, for each of at least one connecting edge between the node corresponding to the advertisement text and the corresponding at least one associated node, a co-occurrence frequency between the advertisement texts respectively corresponding to two end points of the connecting edge,
and wherein the second determination module is configured to determine the edge vector representation of the connecting edge based on semantic and revenue information of the advertisement text respectively corresponding to the two endpoints of the connecting edge and the co-occurrence frequency.
19. The apparatus of claim 18, wherein the feature vector determination unit further comprises:
a processing subunit configured to perform a normalization process on the co-occurrence frequency and the benefit information, respectively, to obtain a normalized co-occurrence frequency and normalized benefit information,
and wherein the first determination module is configured to determine an edge vector representation of the connecting edge based on semantic and normalized revenue information of the advertisement text respectively corresponding to the two endpoints of the connecting edge and the normalized co-occurrence frequency.
20. The apparatus of any one of claims 15-19, wherein the second determining module is configured to:
inputting the advertisement texts and the income information corresponding to the two end points of the connecting edge into an edge vector coding model to obtain the edge vector representation output by the edge vector coding model,
wherein, the edge vector coding model is obtained by training in the following way:
obtaining a sample text graph containing a plurality of nodes in one-to-one correspondence with a plurality of sample texts and profit information of each sample text in the plurality of sample texts, wherein the sample text graph comprises a plurality of connecting edges for connecting the plurality of nodes;
for each connecting side in a plurality of connecting sides included in a sample text graph, inputting sample texts and income information thereof corresponding to two end points of the connecting side into the edge vector coding model to obtain an edge vector representation of the connecting plate output by the edge vector coding model;
determining feature vector representations of a plurality of sample texts corresponding to the plurality of nodes based on the edge vector representations of the plurality of connecting edges;
acquiring a real correlation degree between a first sample text and a second sample text in the plurality of sample texts;
determining a prediction correlation between the first sample text and the second sample text based on a first feature vector representation and a second feature vector representation corresponding to the first sample text and the second sample text, respectively; and
and adjusting parameters of the edge vector coding model based on the real correlation degree and the prediction correlation degree.
21. The apparatus of any of claims 14-20, wherein the plurality of advertisement texts comprises a plurality of historical query texts and a plurality of historical recommendation texts, the historical advertisement data comprises a plurality of text pairs, each text pair of the plurality of text pairs comprises one historical query text and a historical recommendation text that is recommended to a user based on the historical query text,
and wherein the build module is configured to:
in response to determining that the historical advertisement data includes a text pair consisting of the two advertisement texts, edges of a text graph are established with nodes corresponding to the two advertisement texts as vertices.
22. The apparatus of any one of claims 14-21, wherein the first determining module is configured to:
for each other node in the plurality of other nodes, determining the other node as the associated node in response to the number of connection hops between the other node and the node not being greater than a preset threshold.
23. The apparatus of any one of claims 14-21, wherein the first determining module is configured to:
determining respective sampling probabilities of the other nodes based on a connection relation between the nodes and a preset rule, wherein according to the preset rule, the sampling probability of a node with a smaller connection hop count with the node is larger than the sampling probability of a node with a larger connection hop count with the node; and
randomly sampling the plurality of other nodes based on the sampling probability to obtain the at least one associated node.
24. An advertisement text recommending apparatus comprising:
a second acquisition unit configured to acquire a target advertisement text and a plurality of candidate advertisement texts;
the apparatus of any of claims 13-23, configured to determine a degree of correlation between the target advertisement text and the plurality of candidate advertisement texts; and
a second determining unit configured to determine at least one advertisement text to be recommended from the plurality of candidate advertisement texts based on the correlation between the target advertisement text and the plurality of candidate advertisement texts.
25. An electronic device, comprising:
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 method of any one of claims 1-12.
26. A non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method of any one of claims 1-12.
27. A computer program product comprising a computer program, wherein the computer program realizes the method according to any one of claims 1-12 when executed by a processor.
CN202211541802.4A 2022-12-02 2022-12-02 Method, device, equipment and medium for determining relevancy of advertisement text Pending CN115829653A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211541802.4A CN115829653A (en) 2022-12-02 2022-12-02 Method, device, equipment and medium for determining relevancy of advertisement text

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211541802.4A CN115829653A (en) 2022-12-02 2022-12-02 Method, device, equipment and medium for determining relevancy of advertisement text

Publications (1)

Publication Number Publication Date
CN115829653A true CN115829653A (en) 2023-03-21

Family

ID=85543869

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211541802.4A Pending CN115829653A (en) 2022-12-02 2022-12-02 Method, device, equipment and medium for determining relevancy of advertisement text

Country Status (1)

Country Link
CN (1) CN115829653A (en)

Similar Documents

Publication Publication Date Title
CN112579909A (en) Object recommendation method and device, computer equipment and medium
CN114612749B (en) Neural network model training method and device, electronic device and medium
CN114791982B (en) Object recommendation method and device
CN114443989B (en) Ranking method, training method and device of ranking model, electronic equipment and medium
CN115114424A (en) Response method and device for query request
CN115168545A (en) Group searching method, device, electronic equipment and medium
CN115269989B (en) Object recommendation method, device, electronic equipment and storage medium
CN116304335A (en) Object recommendation method, user preference identification method and device
CN113722594B (en) Training method and device of recommendation model, electronic equipment and medium
CN115600646A (en) Language model training method, device, medium and equipment
CN114445147A (en) Electronic ticket issuing method, electronic ticket issuing device, electronic ticket issuing apparatus, and electronic ticket issuing medium
CN114281990A (en) Document classification method and device, electronic equipment and medium
CN114445190A (en) Commodity recommendation method and device
CN114048383A (en) Information recommendation method and device, electronic equipment and medium
CN115829653A (en) Method, device, equipment and medium for determining relevancy of advertisement text
CN113312554A (en) Method and device for evaluating recommendation system, electronic equipment and medium
CN112860681A (en) Data cleaning method and device, computer equipment and medium
CN115809364B (en) Object recommendation method and model training method
CN114861071B (en) Object recommendation method and device
CN115033782B (en) Object recommendation method, training method, device and equipment of machine learning model
CN114861658B (en) Address information analysis method and device, equipment and medium
CN116383372B (en) Data analysis method and system based on artificial intelligence
CN112765975B (en) Word segmentation disambiguation processing method, device, equipment and medium
CN115292608A (en) Content resource recommendation method, neural network training method, device and equipment
CN114169440A (en) Model training method, data processing method, device, electronic device and medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination