Information pushing method, apparatus, device, storage medium and program product

Abstract

The application provides an information pushing method and device, equipment, storage medium and program product, which can be applied to the field of big data technology. The method comprises the following steps: based on object portrait data of a target object and content attribute data of to-be-pushed content, determining a first conversion probability of the target object based on each content component in a plurality of content components, wherein the conversion probability indicates a predicted probability of the target object performing a conversion operation on the to-be-pushed content; based on the first conversion probability of each content component, determining a second conversion probability of each content component combination in at least one content component combination, wherein each content component combination comprises at least part of the plurality of content components; according to the second conversion probability of each content component combination, determining a target content component combination for information pushing; and based on the to-be-pushed content and the target content component combination, generating target pushing information, and sending the target pushing information to the target object.

Description

Technical Field

[0001] This application relates to the field of big data, specifically to an information push method, apparatus, device, storage medium, and program product. Background Technology

[0002] With the development of internet technology and the widespread adoption of smart terminals, various applications can push information to target devices to achieve functions such as information delivery and service notification. However, in these technologies, information push typically relies on fixed, single patterns, resulting in low adaptability to the target audience and consequently low efficiency. Summary of the Invention

[0003] In view of the above problems, embodiments of this application provide an information push method, apparatus, device, storage medium, and program product.

[0004] According to a first aspect of this application, an information push method is provided, comprising: determining a first conversion probability of the target object based on object profile data of a target object and content attribute data of content to be pushed, wherein the conversion probability indicates the predicted probability that the target object will perform a conversion operation on the content to be pushed; determining a second conversion probability of each combination of at least one combination of content components based on the first conversion probability of each content component, wherein each combination of content components includes at least some of the content components among the multiple content components; determining a target combination of content components for information push based on the second conversion probability of each combination of content components; and generating target push information based on the content to be pushed and the target combination of content components, and sending the target push information to the target object.

[0005] According to embodiments of this application, determining a second conversion probability of each content component combination in at least one content component combination based on a first conversion probability of each content component includes: determining a synergy coefficient among at least some content components among a plurality of content components based on the association relationship between different content components, wherein the synergy coefficient indicates the degree of influence of any content component among the at least some content components on the first conversion probability of other content components when the at least some content components are used in combination; and for any content component combination, determining a second conversion probability of the content component combination based on the first conversion probability of each content component in the content component combination and the synergy coefficient among the plurality of content components in the content component combination.

[0006] According to an embodiment of this application, determining a second conversion probability of each content component combination in at least one content component combination based on a first conversion probability of each content component includes: determining a conversion correction factor matching each content component based on the degree of influence of operating environment parameters on the first conversion probability of each content component; for any content component, determining a corrected first conversion probability corresponding to the content component based on the first conversion probability of the content component and the conversion correction factor; and determining a second conversion probability of each content component combination based on the corrected first conversion probability corresponding to each content component.

[0007] According to an embodiment of this application, based on the object profile data of the target object and the content attribute data of the content to be pushed, the first conversion probability of the target object based on each of the multiple content components is determined, including: based on the component attribute data of each of the multiple content components, determining the degree of matching between each content component and the object profile data of the target object; and based on the degree of matching corresponding to each content component, determining the first conversion probability of the corresponding content component.

[0008] According to an embodiment of this application, the object profile data further includes object behavior data of the target object, which indicates that the target object has active object behavior based on multiple historical moments; sending target push information to the target object includes: determining the time distribution characteristics of multiple historical moments based on at least one candidate time period; determining a third conversion probability matching each candidate time period based on the time distribution characteristics; using the candidate time period with the highest corresponding third conversion probability as the target time period for information push; and sending target push information to the target object based on the target time period.

[0009] According to embodiments of this application, determining the temporal distribution characteristics of multiple historical moments based on at least one candidate time period includes: mapping multiple historical moments to multiple predicted active moments based on a first mapping scale, and determining the multiple predicted active moments based on a first distribution characteristic of at least one candidate time period, wherein the first mapping scale indicates that the m-th historical moment adjacent to any target historical moment is taken as the predicted active moment, where m is an integer greater than 0; and / or mapping multiple historical moments to multiple predicted active time periods based on a second mapping scale, and determining the multiple predicted active time periods based on a second distribution characteristic of at least one candidate time period, wherein the second mapping scale indicates that any target historical moment and the n-th historical moment adjacent to the target historical moment are taken as boundary points to obtain the corresponding predicted active time period, where n is an integer greater than 0; wherein the first distribution characteristic and / or the second distribution characteristic constitute the temporal distribution characteristics.

[0010] According to an embodiment of this application, the first distribution feature includes at least one of the following features: the number of active moments, the dispersion of active moments, and the time period in which active moments are concentrated; the second distribution feature includes at least one of the following features: the number of active time periods, the proportion of active time periods covered by the duration, and the distribution of active time periods across candidate time periods.

[0011] According to a second aspect of this application, an information push device is provided, comprising: a first conversion probability determination module, configured to determine a first conversion probability of the target object based on object profile data of the target object and content attribute data of the content to be pushed, wherein the conversion probability indicates the predicted probability that the target object will perform a conversion operation on the content to be pushed; a second conversion probability determination module, configured to determine a second conversion probability of each combination of at least one combination of content components based on the first conversion probability of each content component, wherein each combination of content components includes at least some of the content components among the plurality of content components; a target content component combination determination module, configured to determine a target content component combination for information push based on the second conversion probability of each combination of content components; a target push information determination module, configured to generate target push information based on the content to be pushed and the target content component combination; and a target push information sending module, configured to send the target push information to the target object.

[0012] According to a third aspect of this application, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method described above.

[0013] According to a fourth aspect of this application, a computer-readable storage medium is also provided, on which a computer program or instructions are stored, wherein the computer program or instructions, when executed by a processor, implement the steps of the above-described method.

[0014] According to a fifth aspect of this application, a computer program product is also provided, including a computer program or instructions that, when executed by a processor, implement the steps of the above-described method. Attached Figure Description

[0015] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0016] Figure 1 The illustrations depict application scenarios of information push methods, apparatus, devices, storage media, and program products according to embodiments of this application.

[0017] Figure 2A flowchart illustrating an embodiment of the information push method according to this application is shown schematically;

[0018] Figure 3 A flowchart illustrating an object profile data processing method according to an embodiment of this application is shown schematically.

[0019] Figure 4 A flowchart illustrating a method for determining a combination of target content components according to one embodiment of this application is shown schematically;

[0020] Figure 5 A flowchart illustrating an information push method according to a second embodiment of this application is shown schematically.

[0021] Figure 6 A flowchart illustrating a method for determining a target push time according to an embodiment of this application is shown schematically.

[0022] Figure 7 This schematic diagram illustrates the structure of an information push device according to an embodiment of the present application;

[0023] Figure 8 A block diagram schematically illustrates an electronic device suitable for implementing an information push method according to an embodiment of this application. Detailed Implementation

[0024] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0025] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0026] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0027] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0028] In the technical solution of this application, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with relevant laws, regulations, and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0029] In scenarios where personal information is used for automated decision-making, the methods, devices, and systems provided in this application all provide users with corresponding operation entry points for users to choose to agree to or reject the automated decision results; if the user chooses to reject, the process enters the expert decision-making process.

[0030] With the development of internet technology and the widespread adoption of smart terminals, various applications deliver information and provide service notifications by pushing information to the target devices. However, in these technologies, information push often relies on pre-set rules or a single pattern, lacking the flexibility to adjust push strategies to different target objects or scenarios. Under this strategy, the content of the pushed information is prone to deviating from the actual state of the target object. Target objects often fail to respond further after receiving information with low relevance, resulting in poor information push effectiveness and low efficiency.

[0031] To address the aforementioned issues, embodiments of this application provide an information push method. The method includes: determining a first conversion probability of the target object based on object profile data of a target object and content attribute data of the content to be pushed, wherein the conversion probability indicates the predicted probability that the target object will perform a conversion operation on the content to be pushed; determining a second conversion probability of each combination of at least one content component based on the first conversion probability of each content component, wherein each content component combination includes at least some of the content components from the plurality of content components; determining a target content component combination for information push based on the second conversion probability of each content component combination; and generating target push information based on the content to be pushed and the target content component combination, and sending the target push information to the target object.

[0032] Figure 1 The illustrations depict application scenarios of information push methods, apparatuses, devices, storage media, and program products according to embodiments of this application.

[0033] like Figure 1 As shown, application scenario 100 according to an embodiment of this application may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables. For example, a user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send information, etc.

[0034] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be electronic devices such as smartphones, wearable devices, personal computers, intelligent voice interaction devices, smart home appliances, intelligent vehicles, in-vehicle terminals, aircraft, unmanned vending terminals, and extended reality devices. Extended reality devices can include virtual reality devices, augmented reality devices, and mixed reality devices. A client application for the target application can be installed and run on the terminal devices. This target application can include, but is not limited to, financial transaction applications, payment applications, shopping applications, web browser applications, search applications, instant messaging tools, email clients, and social media platform software (these are just examples). Furthermore, this application embodiment does not limit the form of the target application, and it can include, but is not limited to, applications, mini-programs, etc., installed on the terminal devices, and can also be in the form of web pages.

[0035] Server 105 can be a server providing various services, such as a backend management server supporting websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process received user requests and other data, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services such as cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks, and basic cloud computing services such as big data. The server can be the backend server of the aforementioned target application, used to provide backend services to the clients of the target application.

[0036] It should be noted that the information push method provided in this application embodiment can generally be executed by server 105 and / or terminal devices 101-103. Accordingly, the information push device provided in this application embodiment can generally be set in server 105 and / or terminal devices 101-103.

[0037] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0038] Figure 2 A flowchart illustrating an embodiment of an information push method according to this application is shown. Figure 2 As shown, the information push method according to the embodiments of this application may include steps S210 to S240.

[0039] In step S210, based on the object profile data of the target object and the content attribute data of the content to be pushed, the first conversion probability of the target object based on each of the multiple content components is determined, wherein the conversion probability indicates the predicted probability of the target object performing a conversion operation on the content to be pushed.

[0040] In embodiments of this application, the target object's consent or authorization can be obtained before acquiring its information. For example, before step S210, a request to acquire its object profile data can be sent to the target object. If the target object consents or authorizes the acquisition of the object profile data, step S210 is executed.

[0041] In the embodiments of this application, a corresponding operation entry point can be provided for the target object, allowing the target object to choose to agree to or reject the automated decision result. That is, before processing the target object's profile data, the target object can be provided with an instruction to agree to or reject the processing via the corresponding operation entry point. If the target object agrees to the processing, the profile data is processed, i.e., step S210 is executed. If the target object refuses to perform the processing, the expert decision-making process is initiated.

[0042] In the embodiments of this application, the target object is the recipient of the pushed information. In specific application scenarios, the target object can be an application user, website visitor, device user, etc. The target object's profile data can be a collection of various target object feature data. For example, the target object's profile data can include the target object's historical behavior data, such as browsing history, click history, etc. The target object's profile data can also include the target object's static attribute data, such as user personal information, device information, etc. The target object profile data can be derived from server-side logs, client-side event tracking data, or external data interfaces, etc.

[0043] In the embodiments of this application, the content to be pushed can be the main body of the information push. In different application scenarios, the content to be pushed can be different information entities. For example, in a financial information push scenario, the content to be pushed can be financial product information, market news, or risk warnings. In a content recommendation scenario, the content to be pushed can be news, short videos, or text and image content. In a service notification scenario, the content to be pushed can be order status, activity notifications, or system messages.

[0044] Content attribute data can be the feature data of the content to be pushed, such as content type, content theme, content format, and the attribute dimensions involved in the content. For example, in a financial push scenario, content attribute data may include return attributes, risk attributes, and term attributes. In a content recommendation scenario, content attribute data may include content category, content length, and content popularity.

[0045] Content components can be the basic units that make up the content to be pushed out. Each content component can carry at least one part of the content to be pushed out, and one or more content components can be combined to form a complete push message. For example, in a financial push scenario, content components may include profit description components, risk warning components, and term description components. In other push scenarios, content components may include title components, body text components, image components, and interactive components.

[0046] In the embodiments of this application, a first conversion probability of the target object based on the object profile data of the target object and the content attribute data of the content to be pushed can be determined. The first conversion probability refers to the likelihood that the target object will perform a conversion operation for the content to be pushed containing a certain content component. A conversion operation refers to the response operation performed by the target object to the received push information. For example, in a financial push scenario, the conversion operation could be clicking on details, browsing the complete content, or filling out a form. In a content recommendation scenario, the conversion operation could be clicking to read, completing viewing, or sharing. In a service notification scenario, the conversion operation could be confirming viewing or performing subsequent operations.

[0047] The first conversion probability can be determined in several ways. For example, the degree of matching between the object profile data and the content attribute data corresponding to each content component can be calculated, and this matching degree can be mapped to the first conversion probability. A higher matching degree indicates that the content component matches the characteristics of the target object more closely, and the higher the first conversion probability. The first conversion probability can also be obtained based on the response rate statistics of the target object to each content component in historical data. Alternatively, the first conversion probability can be based on the output of a prediction model, which takes object profile data and content attribute data as input and outputs the conversion probability of each content component.

[0048] Since content components can carry at least some attributes of the content to be pushed, target audiences may have different acceptance preferences for the same content presented in different content components. For example, target audiences may have a higher acceptance of benefit description components but a lower acceptance of risk warning components. By determining the first conversion probability of each content component, we can obtain the target audience's differentiated acceptance preferences for each content component, providing a basis for the selection of content component combinations.

[0049] In step S220, based on the first conversion probability of each content component, a second conversion probability of each content component combination in at least one content component combination is determined. Each content component combination includes at least a portion of the multiple content components.

[0050] A content component combination is a collection of at least some of the content components. A combination may contain two or more content components. For each candidate content component combination, a second conversion probability needs to be determined. The second conversion probability indicates the likelihood that the target audience will perform a conversion operation on the content to be pushed when the multiple content components in the combination are combined to form the content to be pushed.

[0051] The second conversion probability of a content component combination can be determined in various ways based on the first conversion probabilities of each content component in the combination, such as summing, weighted summing, averaging, or using other calculation methods. By determining the second conversion probability, the overall conversion effect of different content component combinations can be compared and optimized.

[0052] In step S230, the target content component combination for information push is determined based on the second conversion probability of each content component combination.

[0053] Specifically, after obtaining the second conversion probabilities corresponding to each of the multiple candidate content component combinations, selection can be made based on the second conversion probabilities to determine the optimal content component combination as the target content component combination.

[0054] The selection method for target content component combinations can be determined based on specific needs. For example, the content component combination with the highest second conversion probability can be selected as the target content component combination; alternatively, a probability threshold can be set, selecting content component combinations whose second conversion probability exceeds the threshold and meets other constraints, such as restrictions on the number of content components and the overall matching degree between the combination and the content to be pushed. When the second conversion probabilities of multiple content component combinations are similar, other factors can be considered to determine the target content component combination. These other factors may include the types of content components included in the combination and the target audience's historical preferences. Selecting the target content component combination for push notifications based on the second conversion probability ensures that the pushed content is suitable for the target audience, increasing the likelihood of the information being effectively received.

[0055] In step S240, target push information is generated based on the combination of the content to be pushed and the target content component, and the target push information is sent to the target object.

[0056] Specifically, the content to be pushed can be combined according to the content component structure indicated by the target content component combination to form a complete push information content, which is then used as the target push information and sent to the target object's terminal device through a push channel. The push channel can be in-app push, message push service, or SMS, etc.

[0057] By determining the first conversion probability of each content component through the embodiments of this application, the differentiated acceptance tendency of the target object to different content components can be obtained; by determining the second conversion probability of each content component combination based on the first conversion probability, the overall conversion effect of the content component combination can be evaluated; by determining the target content component combination for push based on the second conversion probability, the content format of the push information can be adjusted according to the actual situation of the target object, improving the matching degree between the push content and the target object, reducing invalid pushes, and improving the efficiency and accuracy of information push.

[0058] In step S210 of some embodiments of this application, before determining the first conversion probability of each content component based on the object profile data of the target object and the content attribute data of the content to be pushed, feature fusion and / or desensitization processing can be performed on the object profile data.

[0059] Figure 3 The flowchart illustrating an object profile data processing method according to an embodiment of this application is shown schematically.

[0060] In some embodiments of this application, the original information of the target object can be fused and / or desensitized using quantum feature mapping, such as... Figure 3 As shown, the operation may include steps S310 to S350.

[0061] In step S310, the original object profile data of the target object is obtained. The original object profile data may include original information of the target object in multiple dimensions, such as the target object's static attribute data, risk assessment data, and historical behavior data.

[0062] In step S320, at least a portion of the original object profile data is encoded into a quantum state representation.

[0063] For example, numerical information such as age can be encoded using a quantum rotation gate, mapping the age value to the rotation angle of the quantum rotation gate, thus encoding the age value as a single-qubit superposition state. A quantum rotation gate is a single-qubit gate whose function is to perform a rotation operation on a quantum state; the rotation angle determines the probability amplitude distribution ratio of the two ground states in the resulting quantum state.

[0064] For example, if the target object is 35 years old, the age value 35 can be multiplied by π / 180 to obtain the rotation angle θ = 35π / 180, thus establishing the correspondence between the age value and the rotation angle.

[0065] Then, a quantum rotation gate can be applied to the initial state of the qubit. Above, the encoded quantum state is obtained:

[0066]

[0067] That is, the age value, after being encoded by quantum states, can be represented as a superposition state of a single quantum bit:

[0068]

[0069] Among them, in the superposition state of the encoded single quantum bit, The probability amplitude of the state is 0.9537. The probability amplitude of the state is 0.3007, which can be respectively associated with the projection components of the age value in the quantum state space. Different age values ​​can correspond to different rotation angles, thus producing different probability amplitude allocation ratios, so that age information is encoded in the form of quantum state probability amplitudes.

[0070] For example, numerical information such as asset values ​​can be encoded using amplitude encoding, which maps asset values ​​to probability amplitudes of quantum states. Amplitude encoding is an encoding method that maps numerical information to probability amplitudes of quantum states. By normalizing the numerical value and using it as the probability amplitude of the ground state in a quantum state, numerical information can be encoded into quantum states in the form of probability amplitudes.

[0071] For example, if the asset value of a target object is 1.2 million, we can assume that the upper limit of the asset value is 5 million. Then the asset normalization value can be p = 120 / 500 = 0.24. This value can be encoded as the probability amplitude of a quantum state, and the superposition state representation of the qubit is obtained as follows:

[0072]

[0073] In other words, asset value, after being encoded in quantum states, can be represented as a superposition of single qubits:

[0074]

[0075] For example, discrete category information such as risk preference levels can be encoded using computational ground-state encoding. Computational ground-state encoding is an encoding method that maps discrete category information to computational ground states in quantum state space; different categories of discrete information can correspond to different computational ground states.

[0076] For example, if there are three levels of risk preference, the initial quantum state can be represented as a three-qubit ground state. If a target's risk preference level is 3, then the three qubits can be processed separately through a Pauli X-gate. The Pauli X-gate is one of the logic gates in quantum computing; its function is to... state flip to attitude, will state flip to The state is equivalent to the classical NOT gate. Applying a Pauli X gate to the first qubit will... Flip to ,get Applying a Pauli X-gate to the second qubit will... Flip to ,get Applying an X gate to the third qubit will Flip to ,get The calculated ground state This is the coding result corresponding to risk preference level 3.

[0077] In step S330, the quantum state representations corresponding to each data point of the original object portrait data are combined to obtain a combined quantum state. This combined quantum state can represent the fusion result of each data point of the original object portrait data in quantum space.

[0078] For example, the age code, asset code, and risk preference code mentioned above can be combined, and their combined quantum state can be represented as:

[0079]

[0080] In step S340, a quantum noise perturbation is applied to the combined quantum state to obtain the perturbed combined quantum state. This quantum noise perturbation can be achieved by applying a noise operator based on the Pauli matrix to the combined quantum state. The probability amplitude and phase of each ground state in the perturbed quantum state are changed, which can increase the difficulty of deducing the original information from the quantum state, thereby further protecting the privacy and security of the target object.

[0081] For example, a noisy Hamiltonian can be constructed based on the Pauli matrix, which describes the noise perturbation applied to the quantum state. The noisy Hamiltonian can take the form of a combination of a bit-flipping matrix and a phase-flipping matrix:

[0082]

[0083] Noise operators can be constructed based on the noisy Hamiltonian:

[0084]

[0085] in Since the noise intensity parameter is used, the perturbed quantum state output is:

[0086]

[0087] For example, sensitive fields in object profile data can also be isolated. Sensitive fields such as device unique identifiers or ID card numbers can be processed using encryption to de-identify them. For instance, a quantum key distribution protocol can be used to generate encryption keys. The quantum key distribution protocol generates unconditionally secure keys based on the principles of quantum mechanics, and the communicating parties negotiate and share the key through the transmission and measurement of quantum states. The encryption key can be used to encrypt and store sensitive fields, and a hash algorithm can be used to convert the sensitive fields into fixed-length hash values. In subsequent processing, the hash value can be used as an identifier for the target object to participate in data processing, thus achieving the isolation of sensitive fields.

[0088] In step S350, the perturbed quantum state is measured to obtain the probability distribution corresponding to each data measurement ground state. Based on the probability distribution, the desensitized object profile data is determined. During the measurement process, the squared modulus of the probability amplitude corresponding to each data measurement ground state can be calculated to obtain the probability value of each data measurement ground state being measured.

[0089] After obtaining the perturbed quantum state, it can be measured to obtain the probability distribution corresponding to each data measurement ground state of the object profile data. Since the perturbed quantum state is a superposition state, it will collapse to each data measurement ground state with a certain probability during measurement. For any data measurement ground state among the perturbed quantum states, its corresponding probability value is equal to the square of the modulus of the probability amplitude of that data measurement ground state, and thus the probability distribution corresponding to each data measurement ground state can be obtained.

[0090] Based on the probability distribution corresponding to the ground state of each data measurement, the desensitized object profile data can be determined. Since the perturbed quantum state is a superposition state, the measurement results are presented in probability form. The probability values ​​corresponding to different data measurement ground states reflect the statistical characteristics of the original object profile information after perturbation. The data measurement ground state with the highest probability value can be selected as the output, and the feature corresponding to the data measurement ground state can be used as the desensitized feature value.

[0091] For example, if the target's original age is 35, after the aforementioned feature mapping, noise perturbation, and probability measurement, the ground state with the highest probability value can be selected, resulting in an anonymized age information that could be the age range "30-40". If the original asset value is 1.2 million, the anonymized asset information could be asset level "2". If the original risk preference level is 3, the anonymized risk preference category could be "B". For sensitive fields, such as the device's unique identifier, the hash value after hashing is output. The final anonymized object profile data can be represented as a combination of age range, asset level, risk category, and anonymous identifier.

[0092] The anonymized object profile data described above can be used as input data to determine the first conversion probability of each content component. By outputting the anonymized features of the original object profile data, the risk of the original information being restored can be reduced while preserving the statistical characteristics of the data, thus protecting the privacy of the target object.

[0093] In step S220 of some embodiments of this application, determining the second conversion probability of each content component combination in at least one content component combination based on the first conversion probability of each content component may include: determining the synergy coefficient between at least some content components among a plurality of content components based on the association relationship between different content components, wherein the synergy coefficient indicates the degree of influence of any content component among at least some content components on the first conversion probability of other content components when the at least some content components are used in combination; and determining the second conversion probability of the content component combination for any content component combination based on the first conversion probability of each content component in the content component combination and the synergy coefficient between the plurality of content components in the content component combination.

[0094] Specifically, based on the relationships between different content components, a synergy coefficient can be determined among at least some of the content components. The synergy coefficient characterizes the mutual influence that occurs when content components are used in combination. When two or more content components are combined into content to be pushed to a target audience, there may be mutual influences between these content components on the conversion probability.

[0095] The synergy coefficient can be determined in several ways. For example, it can be determined based on historical push data. The conversion rates of each content component pushed individually and when multiple content components are pushed together are statistically analyzed and compared. The synergy coefficient is determined based on the difference between the two. For instance, if the first conversion probability of content component A pushed alone is 0.3, and the first conversion probability of content component B pushed alone is 0.4, and the actual conversion probability of their combined push is 0.55, which is higher than a certain percentage of the sum of their individual conversion probabilities, then the synergy coefficient between A and B is positive, indicating a synergistic enhancement effect when used together. Conversely, the synergy coefficient is negative, indicating a synergistic inhibition effect when used together.

[0096] For example, the synergy coefficient can be set based on the semantic or business association between content components. For instance, in a financial push scenario, when the benefit description component and the risk warning component are presented together, they can provide more complete information, which helps the target audience make a judgment. In this case, the benefit description component and the risk warning component can be set to have a positive association and a positive synergy coefficient. Conversely, if the attributes of the two content components are duplicated or conflict with each other, they can be set to have a negative association.

[0097] For example, the synergy coefficient can also be determined based on the differentiated behavioral data of different target objects, and the synergy coefficient of different target objects for the same content component combination can be different.

[0098] After obtaining the synergy coefficient, for any combination of content components, the second conversion probability of the combination of content components can be determined based on the first conversion probability of each content component in the combination of content components and the synergy coefficient between each content component in the combination of content components.

[0099] Specifically, the first conversion probability of each content component can be used as a base value. This base value is then processed using the synergy coefficients between the components within the combination to obtain the second conversion probability of the content component combination. For example, for a combination containing content components A, B, and C, the first conversion probabilities of A, B, and C individually, as well as the synergy coefficients between A and B, A and C, and B and C, can be obtained. These data are then calculated using a preset algorithm to yield the second conversion probability of the content component combination. The synergy coefficients, when determining the second conversion probability, can quantify and correct the mutual influence between the content components within the combination. This ensures that the second conversion probability reflects not only the individual conversion effect of each content component but also the synergistic or inhibitory effects produced after the combination, thus making the evaluation of the conversion effect of the content component combination more accurate.

[0100] For example, suppose a content component combination includes a benefit description component and a risk warning component. The first conversion probability of the benefit description component is 0.4, and the first conversion probability of the risk warning component is 0.3. The synergy coefficient between the two is determined to be a positive value of 0.15. When determining the second conversion probability of this combination, it can be calculated based on 0.4, 0.3, and 0.15. The resulting second conversion probability can be higher than 0.4 or 0.3 alone, reflecting the synergistic enhancement effect when the two components are used together. Conversely, if the synergy coefficient between the two content components is negative, the second conversion probability will decrease accordingly, reflecting the synergistic inhibition effect when used together. For example, the synergy coefficient can also be used only to correct the relationship between some content components in the content component combination. For instance, when there is a strong correlation between some content components in the content component combination, only the synergy coefficient between the strongly correlated content components can be corrected, while the synergy coefficient between other content components can use default values ​​(such as 0 or 1) to reduce computational complexity.

[0101] Through the embodiments of this application, the synergy coefficient can serve as a quantitative descriptive parameter for the mutual influence between content components, reflecting the strength and direction of the correlation between content components. When determining the second conversion probability, it can accurately assess the overall conversion effect of each content component combination, providing a basis for determining the target content component combination. The synergy coefficient can also be determined separately for different target objects, achieving personalized adaptation of the content component combination pushed, reducing invalid pushes, and improving push effectiveness.

[0102] In step S220 of some embodiments of this application, determining the second conversion probability of each content component combination in at least one content component combination based on the first conversion probability of each content component may include: determining a conversion correction factor matching each content component based on the degree of influence of the operating environment parameters on the first conversion probability of each content component; for any content component, determining the corrected first conversion probability corresponding to the content component based on the first conversion probability and the conversion correction factor; and determining the second conversion probability of each content component combination based on the corrected first conversion probability corresponding to each content component.

[0103] Specifically, runtime environment parameters can be various environmental information that, independent of the target profile data, can affect the effectiveness of information delivery when making push notification decisions. Runtime environment parameters can include parameters reflecting business status, device status, and network status, etc.

[0104] For example, parameters reflecting business status can be used to indicate dynamic changes in the current business operation, such as market fluctuations and changes in business metrics. These parameters can influence the target audience's acceptance of different types of content. For instance, risk warning components may be more readily accepted during periods of market volatility.

[0105] For example, parameters reflecting device status can be used to indicate the current operating status of the target terminal device, such as remaining battery power and storage space. These parameters may affect the display performance of content components. For instance, the acceptance of video components may decrease when the battery is low.

[0106] For example, parameters reflecting network status can be used to indicate the network connectivity of the target terminal device, such as signal strength and network type. These parameters can affect the transmission and loading of content components. For instance, in environments with poor network signal, the loading of image or video components may be hindered, while text components are less affected.

[0107] The aforementioned runtime environment parameters can be real-time environmental information independent of the target object's attributes, which can be obtained through the corresponding system interface during push decision-making. Adjusting the conversion probability based on runtime environment parameters can improve the adaptability of push content under different conditions.

[0108] Based on the influence of runtime environment parameters on the initial conversion probability of each content component, a conversion correction factor matching each content component can be determined. The conversion correction factor indicates the direction and degree of influence of the current runtime environment parameters on the initial conversion probability of a particular content component. The correction factor can be positive, negative, or zero. A positive correction factor indicates that the initial conversion probability of the content component is increased under the current runtime environment; a negative correction factor indicates that the initial conversion probability of the content component is decreased under the current runtime environment; and a correction factor of zero indicates that the initial conversion probability of the content component is unaffected under the current runtime environment.

[0109] Conversion correction factors can be determined in several ways. For example, an influence mapping rule between runtime environment parameters and each content component can be pre-established, and the conversion correction factor for each content component can be determined based on the current runtime environment parameters and this mapping rule. Alternatively, the correction factor can be determined based on the conversion performance statistics of each content component under different runtime environment conditions in historical push data. For instance, the deviation between the actual conversion rate and the benchmark conversion rate of each content component under a certain type of runtime environment condition can be statistically analyzed in past pushes, and this deviation can be used as the conversion correction factor for that content component under that type of runtime environment condition. Another example is the determination of conversion correction factors based on a predictive model, which can take runtime environment parameters and content component attributes as input and output the conversion correction factor.

[0110] After obtaining the conversion correction factors for each content component based on runtime environment parameters, for any given content component, the corrected first conversion probability can be determined based on its first conversion probability and the corresponding conversion correction factor. Specifically, the first conversion probability can be corrected using the conversion correction factor, for example, by multiplying the first conversion probability by the conversion correction factor, adding the first conversion probability to the conversion correction factor, or using other calculation methods. The corrected first conversion probability reflects the predicted probability that the target object will perform a conversion operation on the content to be pushed based on the content component under the current runtime environment conditions. Compared to the uncorrected first conversion probability, the corrected first conversion probability can accurately reflect the conversion probability of the content component under the current runtime environment.

[0111] After obtaining the corrected first conversion probability for each content component, the second conversion probability for each combination of content components can be determined based on these corrected first conversion probabilities. Specifically, for any candidate combination of content components, the corrected first conversion probabilities for each content component in the combination can be obtained, and the second conversion probability of the combination can be obtained based on these corrected first conversion probabilities. Since the corrected first conversion probabilities reflect the degree of influence of the current operating environment on the conversion probabilities of each content component, the second conversion probability of the combination of content components determined based on the corrected first conversion probabilities also reflects the influence of the current operating environment on the overall conversion effect of the combination of content components, making the evaluation of the conversion effect of the combination of content components in the current operating environment more accurate.

[0112] In some embodiments of this application, the first conversion probability of each content component can be corrected based on the runtime environment parameters to obtain the corrected first conversion probability, and then the second conversion probability of the combination of each content component can be determined based on the corrected first conversion probability and the synergy coefficient between each content component; alternatively, the first conversion probability of each content component can be obtained based on the synergy coefficient, and then the first conversion probability can be corrected based on the runtime environment parameters.

[0113] Through the embodiments of this application, the conversion correction factor can correct the first conversion probability so that the corrected first conversion probability is adapted to the current operating environment. The second conversion probability of the content component combination determined based on the corrected first conversion probability can reflect the impact of the current operating environment on the conversion effect of the content component combination. In this way, a push content component combination that is more suitable for the current environmental conditions can be determined under different operating environments, thereby improving the adaptability of push content under different conditions and reducing invalid pushes caused by changes in the operating environment.

[0114] Figure 4 A flowchart illustrating a method for determining a combination of target content components according to one embodiment of this application is shown schematically.

[0115] In some embodiments of this application, after obtaining the first conversion probability of each content component and the synergy coefficient between each content component, a target content component combination can be determined from multiple candidate content component combinations. When the number of content components is large, the number of all content component combinations may grow exponentially. This combinatorial optimization problem can be solved using quantum annealing, such as... Figure 4 As shown, the operation may include steps S410 to S430.

[0116] In step S410, the selection status of each content component is mapped to the state of a qubit. For each candidate content component, it can be mapped to a qubit in a first state (e.g., ...). When the state is (e.g., state 1), it indicates that the content component is not selected and is in the second state (e.g., state 2). The state indicates that the content component is selected.

[0117] In step S420, an objective function is constructed to describe the compositional relationships of the content components. This objective function can be constructed based on the first conversion probability corresponding to each content component and the synergy coefficient between the content components.

[0118] For example, the first conversion probability corresponding to each content component can be mapped to the weight parameter of each qubit. For any content component, the higher its first conversion probability, the better the conversion effect when the component is used alone, and the larger its corresponding weight in the objective function can be, making the component more likely to be selected.

[0119] The synergy coefficients between the various content components can be mapped to coupling parameters between qubits. For example, for any two content components, if the synergy coefficient is positive, it indicates a synergistic enhancement effect when used together. The corresponding coupling parameter in the objective function can encourage both components to be selected simultaneously or not selected at the same time. If the synergy coefficient is negative, it indicates a synergistic inhibition effect when used together. The corresponding coupling parameter can encourage both components to be selected differently.

[0120] The objective function can be constructed to be negatively correlated with the conversion probability of the content component combination. Thus, the problem of determining the optimal content component combination can be transformed into the problem of determining the qubit state corresponding to the minimum value of the objective function.

[0121] In step S430, the minimum value of the objective function is obtained through quantum annealing to obtain the optimal combination of content components for conversion. During quantum annealing, the system initially exists in a high-energy state, where each qubit is in a superposition state, meaning multiple possible combinations of content components can coexist. By gradually adjusting the transverse magnetic field strength of the system, which can slowly decrease from a high value to a low value over time according to a predetermined protocol, the system slowly evolves from a high-energy state to a low-energy state. As the evolution progresses, the system eventually converges to the state with the lowest energy. At the end of the evolution, each qubit collapses to a defined state, and the combination of content components corresponding to this state can be considered the optimal target combination of content components, with the lowest objective function value, i.e., the highest conversion probability.

[0122] For example, assume that the first conversion probability of the profit description component corresponding to qubit q1 is 0.187, the first conversion probability of the term description component corresponding to qubit q2 is 0.152, and the first conversion probability of the risk warning component corresponding to qubit q3 is 0.121. Based on historical data, the synergy coefficient between the profit description component and the risk warning component is positive (0.0378), indicating a synergistic enhancement effect when used together; the synergy coefficient between the risk warning component and the term description component is negative (-0.0095), indicating a synergistic inhibition effect when used together; and there is no significant synergy between the profit description component and the term description component, with a synergy coefficient of 0. The aforementioned first conversion probabilities and synergy coefficients can be mapped to the weight parameters and coupling parameters of the qubits, respectively, to construct an objective function. The minimum value of the objective function is obtained by quantum annealing. The states of each qubit in the annealing result are q1=+1 (selected), q2=-1 (not selected), and q3=+1 (selected). That is, the optimal combination of objective content components is the benefit description component and the risk warning component. The second conversion probability corresponding to this combination of content components is 0.3658.

[0123] In other embodiments of this application, the influence of runtime environment parameters on the first conversion probability of each content component can be introduced when constructing the objective function. For example, the current runtime environment parameters can be obtained, and a correction factor corresponding to each content component can be determined based on the environment parameters. The first conversion probability can be corrected using the correction factor to obtain the corrected first conversion probability. Using the corrected first conversion probability as a weight parameter of the qubit in the construction of the objective function can improve the adaptability of the target content component combination under different environmental conditions.

[0124] After determining the target content component combination, target push information can be generated based on the target content component combination and the content to be pushed.

[0125] Through the embodiments of this application, quantum annealing can be used to determine the optimal combination of content components from a large number of candidate combinations. Because quantum annealing can perform a global search in the solution space using the quantum tunneling effect, it avoids the problem of the algorithm easily getting trapped in local optima, making the target combination of content components closer to the global optimum and improving the conversion effect of the target combination of content components.

[0126] In step S210 of some embodiments of this application, determining the first conversion probability of the target object based on the object profile data of the target object and the content attribute data of the content to be pushed may include: determining the degree of matching between each content component and the object profile data of the target object based on the component attribute data of each content component; and determining the first conversion probability of the corresponding content component based on the degree of matching corresponding to each content component.

[0127] Specifically, each content component can have its own component attribute data, which can be used to describe the content characteristics carried by that component. Component attribute data can be associated with the content attribute data of the content to be pushed; that is, the component attribute data of each content component can correspond to at least a portion of the content attribute data characteristics in the content to be pushed. For example, in a financial push scenario, the component attribute data of the profit description component can include profit attribute information and its presentation method, while the component attribute data of the risk warning component can include risk attribute information and its presentation method. Both can describe different dimensions of content characteristics in the content to be pushed.

[0128] Target profile data is a collection of various information describing the characteristics of a target target, reflecting their acceptance preferences for different types of content. For example, target profile data might include the target target's responses to content related to benefits or risks in historical push notifications. By comparing the target target's target profile data with the component attribute data of each content component, the suitability of each content component to the target target can be determined. A higher degree of matching indicates that the content characteristics carried by the content component are more consistent with the characteristics of the target target, and the target target's acceptance preference for that content component is higher; a lower degree of matching indicates a lower suitability of the content component to the target target, and the target target's acceptance preference for it is lower.

[0129] The degree of matching between each content component and the target object's profile data can be determined in several ways. For example, the similarity between the object profile data and component attribute data can be calculated, and this similarity can be used as the degree of matching. The similarity calculation can employ a vector space model, where the object profile data and component attribute data are represented as feature vectors, and the similarity is measured by calculating the distance between the two vectors (such as cosine distance, Euclidean distance, etc.). The closer the distance, the higher the degree of matching. Alternatively, the degree of matching can be determined based on the target object's historical response rate to each content component in historical push data. If the target object shows a high response rate to a certain type of content component in historical pushes, the content component has a high degree of matching with the target object; conversely, the degree of matching is low. Another approach is to input the object profile data and component attribute data into a pre-trained matching model, which can be trained based on historical data to learn the matching relationship between the object profile data and component attribute features.

[0130] After obtaining the matching degree between each content component and the target audience profile data, the first conversion probability of each content component can be determined based on its matching degree. There can be a positive correlation between the matching degree and the first conversion probability. A higher matching degree indicates a higher fit between the content component and the target audience, making it more likely that the target audience will perform a conversion operation based on the content component being pushed to them, thus resulting in a higher first conversion probability; conversely, a lower matching degree corresponds to a lower first conversion probability.

[0131] Determining the first conversion probability based on the matching degree can be achieved in several ways. For example, the matching degree can be directly used as the first conversion probability, meaning the first conversion probability equals the numerical value of the matching degree. Alternatively, a mapping function can be established between the matching degree and the first conversion probability, inputting the matching degree into this function and outputting the corresponding first conversion probability. This mapping function can be a linear function, a piecewise function, or a nonlinear function. When the matching degree is determined based on historical response rates, the historical response rates can also be smoothed or normalized and then used as the first conversion probability.

[0132] Because different content components carry different content characteristics within the content to be pushed, the target audience may have varying degrees of matching with different content components under the same content to be pushed. By determining the degree of matching between each content component and the target audience's profile data, and then determining the first conversion probability of each content component based on the degree of matching, the differentiated acceptance tendency of the target audience towards different content components can be obtained, providing a basis for determining and selecting content component combinations. In some embodiments of this application, the first conversion probability of each content component can be determined first by the degree of matching, and then the second conversion probability of each content component combination can be determined based on the aforementioned synergy coefficient and / or correction factor.

[0133] Through the embodiments of this application, the first conversion probability of each content component can be determined based on the degree of matching between the component attribute data of each content component and the object profile data of the target object, thereby clarifying the differentiated acceptance tendency of the target object to different content components and improving the adaptability of the target push information to the target object.

[0134] In some embodiments of this application, before sending the target push information to the target object, the target time period for sending the target push information can be determined based on the object behavior data of the target object.

[0135] Figure 5 A flowchart illustrating an information push method according to a second embodiment of this application is shown.

[0136] In embodiments of this application, the object profile data may further include object behavior data of the target object, which indicates that the target object has active behavior based on multiple historical time points. Sending target push information based on the target time period may include steps S510-S540.

[0137] In step S510, multiple historical moments are determined based on the time distribution characteristics of at least one candidate time period.

[0138] Historical moments can be the points in time when a target object exhibits active behavior within a historical time period. For example, in a push notification scenario, active behavior can include the target object's response to push notifications, proactive data requests, and content access. The timestamps of these behaviors constitute historical moments. Object behavior data can originate from server-side logs, client-side event tracking data, etc., recording the specific time information of the target object's active behavior within a historical time period.

[0139] Candidate time periods can be multiple time periods obtained by dividing a certain time range into certain time intervals. For example, dividing a day into hours can yield 24 candidate time periods, each corresponding to a time interval. Candidate time periods can correspond to a certain time period in the future and serve as the target time period for potential push notifications.

[0140] After obtaining multiple historical moments, the temporal distribution characteristics of these moments over at least one candidate time period can be determined. These temporal distribution characteristics describe the distribution of historical moments across the candidate time periods. The number and distribution patterns of historical moments may differ across different candidate time periods. Historical moments may be concentrated in some candidate time periods, indicating high activity levels of the target object during those periods; conversely, historical moments may be sparse or absent in others, indicating low activity levels. Since candidate time periods are divisions of future time intervals, the temporal distribution characteristics essentially reflect the projection of the target object's active behavior in historical time periods onto the candidate time periods. In other words, the distribution of the target object's activity levels across historical time periods is used to predict its activity levels in future time periods.

[0141] Determining the temporal distribution characteristics can be achieved in several ways. For example, one can obtain the historical moments corresponding to all active behaviors of the target object within a preset historical time period (such as the past 7 or 14 days), and map each historical moment to a corresponding candidate time period. For instance, if a historical moment is 14:35, it could belong to the candidate time period of 14:00-15:00. By counting the number of historical moments within each candidate time period, the activity frequency corresponding to each candidate time period can be obtained. The higher the frequency, the higher the activity level of the target object within that candidate time period in its historical data. Furthermore, the frequency of each candidate time period can be normalized to obtain the frequency proportion or probability distribution corresponding to each candidate time period.

[0142] In step S520, the third conversion probability matching each candidate time period is determined based on the time distribution characteristics.

[0143] After determining the temporal distribution characteristics, a third conversion probability matching each candidate time period can be determined based on these characteristics. The third conversion probability characterizes the likelihood that a target object will perform a conversion operation when a push notification is sent to it during each candidate time period. The third conversion probability can be positively correlated with the target object's historical activity level within each candidate time period. The higher the target object's historical activity level during a particular candidate time period, the more closely that time period matches the target object's activity pattern, the greater the likelihood that the push notification sent during that time period will be effectively received, and the higher the corresponding third conversion probability.

[0144] The third conversion probability can be determined in several ways. For example, when the time distribution feature is the activity frequency of each candidate time period, the activity frequency of each candidate time period can be normalized, and the normalized value can be used as the third conversion probability of that candidate time period. For instance, if the activity frequency of a candidate time period is 35 times, and the total activity frequency of all candidate time periods is 200 times, then the third conversion probability of that candidate time period is 35 / 200 = 0.175. Alternatively, the proportion or probability distribution of the activity frequency of each candidate time period in the time distribution feature can be directly used as the third conversion probability. The third conversion probability can also be determined comprehensively based on multiple statistical indicators in the time distribution feature, such as simultaneously considering activity frequency, activity concentration, and activity trend, and using a weighted combination or comprehensive scoring method to obtain the third conversion probability of each candidate time period.

[0145] In step S530, the candidate time period with the highest third conversion probability is selected as the target time period for information push.

[0146] In step S540, the target push information is sent to the target object based on the target time period.

[0147] After obtaining the third conversion probability corresponding to each candidate time period, the candidate time period with the highest third conversion probability can be used as the target time period for information push. The target time period is used as the time basis for sending the target push information, and the target push information is sent to the target object within the target time period.

[0148] Through the embodiments of this application, the temporal distribution characteristics of multiple historical moments in candidate time periods can be determined based on the object behavior data of the target object. This allows for the determination of the third conversion probability of each candidate time period. The candidate time period with the highest third conversion probability is then selected as the target time period, and target push information is sent within the target time period. Since the target time period can be determined based on the actual active behavior distribution of the target object and is compatible with the target object's activity patterns, push information sent within the target time period is more likely to be received and processed by the target object. This increases the probability of effective reception of push information, reduces invalid pushes caused by improper timing, and improves the effectiveness of information push.

[0149] In some embodiments of this application, step S510 may include: mapping multiple historical moments to multiple predicted active moments based on a first mapping scale, and determining a first distribution feature of the multiple predicted active moments based on at least one candidate time period; and / or, mapping multiple historical moments to multiple predicted active time periods based on a second mapping scale, and determining a second distribution feature of the multiple predicted active time periods based on at least one candidate time period. The first distribution feature and / or the second distribution feature constitute a time distribution feature. The first mapping scale indicates that the m-th historical moment adjacent to any target historical moment is taken as the predicted active moment, and the second mapping scale indicates that any target historical moment and the n-th historical moment adjacent to the target historical moment are taken as boundary points to obtain the corresponding predicted active time period, where m and n are both integers greater than 0.

[0150] In embodiments of this application, the method for determining the temporal distribution characteristics may include: mapping multiple historical moments to multiple predicted active moments based on a first mapping scale. The first mapping scale indicates that the m-th historical moment adjacent to any target historical moment is also considered a predicted active moment.

[0151] The target historical moment can be any one of multiple historical moments. The m-th historical moment adjacent to the target historical moment refers to the m-th adjacent historical moment in the time series, counting forward or backward from the target historical moment.

[0152] For example, when m=1, for any target historical moment, the previous or next historical moment adjacent to it in the time series is included in the range of predicted active moments, which is equivalent to taking the previous or next historical moment adjacent to the target historical moment as the predicted active moment. When m=2, the target historical moment is paired with the second historical moment before or after it.

[0153] The first mapping scale establishes a correlation between discrete historical moments, representing the dependencies between historical active moments. The predicted active moments obtained based on the first mapping scale can also contain these correlations. After obtaining multiple predicted active moments, a first distribution feature of these predicted active moments over at least one candidate time period can be determined. The first distribution feature describes the distribution of predicted active moments across each candidate time period.

[0154] The first distribution feature may include at least one of the following features: the number of active moments, the dispersion of active moments, and the time period in which active moments are concentrated. The number of active moments refers to the cumulative number of predicted active moments within each candidate time period, reflecting the frequency of the target object's activity within each candidate time period. The dispersion of active moments refers to the degree of dispersion of the predicted active moments within each candidate time period; the lower the dispersion, the more concentrated the target object's active moments. The time period in which active moments are concentrated refers to the candidate time period where the predicted active moments are most concentrated, reflecting the location of the target object's active moments clustering. The first distribution feature can reflect the distribution pattern of the target object's active moments.

[0155] In embodiments of this application, the method for determining the time distribution characteristics may further include: mapping multiple historical moments into multiple predicted active time periods based on a second mapping scale. The second mapping scale indicates that any target historical moment and the nth historical moment adjacent to the target historical moment are used as boundary points to obtain the corresponding predicted active time periods.

[0156] Specifically, for any target historical moment, it can be used as the starting boundary point, and the nth historical moment adjacent to the target historical moment can be used as the ending boundary point. A time period is formed with these two historical moments as endpoints, and this time period is the predicted active time period. For example, when n=2, for any target historical moment t, with t as the starting point and the second historical moment after t (t+2) as the ending point, a time period from t to t+2 is formed, and this time period is the predicted active time period.

[0157] The second mapping scale transforms discrete historical moments into continuous predicted active time periods, reflecting the potential duration of the target object's active behavior. After obtaining multiple predicted active time periods, a second distribution characteristic can be determined for these periods across at least one candidate time period. This second distribution characteristic describes the distribution of predicted active time periods within the candidate time periods.

[0158] The second distribution feature may include at least one of the following features: the number of active time periods, the percentage of active time periods covered, and the distribution of active time periods across candidate time periods. The number of active time periods refers to the cumulative number of predicted active time periods within each candidate time period; the percentage of active time periods covered refers to the proportion of the coverage time of a predicted active time period within each candidate time period to the total duration of that candidate time period; and the distribution of active time periods across candidate time periods refers to the distribution of predicted active time periods across multiple candidate time periods. The second distribution feature can reflect the persistence and temporal distribution patterns of the target object's active behavior.

[0159] In the embodiments of this application, the time distribution feature can be composed of a first distribution feature, a second distribution feature, or a combination of both. The first and second distribution features can describe the activity patterns of the target object from different perspectives. The first distribution feature focuses on the distribution of active moments, reflecting when the target object is most active; the second distribution feature focuses on the distribution of active time periods, reflecting within which time intervals the target object remains continuously active. When the first and second distribution features together constitute the time distribution feature, they can corroborate and complement each other, jointly forming a more comprehensive time distribution feature.

[0160] Through the embodiments of this application, historical moments can be mapped to predicted active moments or predicted active time periods based on a first mapping scale or a second mapping scale, extracting the activity patterns of the target object from different perspectives. The first mapping scale enables the time distribution features to reflect the distribution patterns of the target object's active moments, while the second mapping scale enables the time distribution features to reflect the persistence and time range of the target object's active behavior. Using these two mapping scales provides accurate time distribution features for determining the third transformation probability, improving the accuracy of target time period determination.

[0161] Figure 6 A flowchart illustrating a method for determining a target push time according to an embodiment of this application is shown schematically.

[0162] In some embodiments of this application, the conversion probability corresponding to each candidate time period can be determined by quantum computing based on the historical time distribution of the active behavior of the target object. This operation may include steps S610 to S630.

[0163] In step S610, multiple historical moments of the target object are encoded into multiple quantum superposition states to obtain an initial quantum state used to characterize the temporal behavior pattern of the target object.

[0164] Specifically, it is possible to obtain object behavior data of the target object within a certain historical time period, such as the interaction time sequence of the target object in the past 7 days (e.g., 9:30, 12:15, 17:45, etc.). For any historical moment, it can be encoded as a single-qubit superposition state.

[0165] For example, a historical moment can be normalized to a preset time period (such as 24 hours), and the normalized value can be used as the rotation angle of a quantum rotation gate. This historical moment is then encoded into a single-qubit superposition state through the quantum rotation gate. The probability amplitudes of the two ground states of this single-qubit superposition state are determined by the sine and cosine values ​​of the rotation angle, respectively.

[0166] For example, the historical moment t can be normalized according to the length T of a preset time period, and the rotation angle θ = π·t / T can be used to encode the quantum state:

[0167]

[0168] Because the positions of different historical moments within a preset time period are different, their corresponding rotation angles are different, resulting in different encoded quantum superposition states. In the single-qubit superposition states corresponding to each historical moment, state and The states can correspond to the starting phase and ending phase of the time axis within a preset time period, respectively, representing the phase position of that time point within the preset time period.

[0169] After encoding each historical moment as a superposition of single qubit states, a tensor product operation can be performed on each single qubit superposition state to obtain an initial quantum state. This initial quantum state is a combination of tensor products of multiple single qubit superposition states, used to characterize the overall phase distribution of the historical moments of the target object within a preset time period.

[0170] In step S620, a time correlation operator representing the correlation between historical moments is constructed, and the time correlation operator is applied to the initial quantum state to obtain the output quantum state.

[0171] Specifically, the time correlation operator can be in the form of a unitary operator, used to establish correlations between adjacent qubits to introduce temporal information. For example, the time correlation operator can be represented as:

[0172]

[0173] in It can describe the interaction between adjacent qubits. This can be the coupling strength between adjacent qubits; It can be an associated parameter.

[0174] The coupling strength can be determined based on the interaction correlation between adjacent time points in historical data. For example, if two adjacent historical moments tend to occur simultaneously in history, the corresponding coupling strength is high; if they tend to occur mutually exclusively, the coupling strength is low.

[0175] Associated parameters It can be used to control the association range of the time correlation operator, that is, to control the scale range of extracted temporal features. When When the entanglement is small, it only covers a small number of adjacent qubits, and the extraction mainly focuses on local correlation features between adjacent historical moments, corresponding to activity patterns on a smaller timescale. When When the entanglement is large, it can propagate to more distant qubits, expanding the correlation range and enabling the extraction of long-range correlation features spanning multiple historical moments, corresponding to periodic patterns over larger timescales. In some embodiments, the correlation parameters... Smaller temporal correlation operators can serve as the first mapping scale to extract the first distribution features; while correlation parameters... Larger time-correlation operators can be used as a second mapping scale to extract second distribution features.

[0176] For example, when the correlation parameter takes a first value (e.g., π / 4), the time correlation operator can extract hourly-level temporal correlation features, reflecting the activity patterns of the target object between adjacent time points. When the correlation parameter takes a second value (e.g., π / 2), the time correlation operator can extract day-level temporal correlation features, reflecting the activity patterns of the target object across day cycles. By adjusting the correlation parameter, multi-scale temporal feature extraction can be achieved within the same quantum circuit.

[0177] Applying the time correlation operator to the initial quantum state yields the output quantum state, which carries temporal correlation information between historical moments and can be used for subsequent comparison and measurement with each candidate time period.

[0178] In step S630, the output quantum state is projected based on the ground state corresponding to each candidate time period to obtain the third transformation probability corresponding to each candidate time period, and the candidate time period with the highest third transformation probability is taken as the target push time.

[0179] Specifically, the preset time period can be divided into multiple candidate time periods (e.g., divided by hours, resulting in 24 candidate time periods). For each candidate time period, its position within the preset time period is encoded as the corresponding time measurement ground state. The output quantum state is then projected onto the time measurement ground state corresponding to each candidate time period: for any candidate time period, the squared modulus of the inner product of its corresponding time measurement ground state and the output quantum state can be calculated to obtain the third transformation probability corresponding to that candidate time period. This third transformation probability can be used to characterize the probability that the target object will perform a transformation operation on the push information when the target push information is sent during that candidate time period.

[0180] For example, if the squared inner product modulus of the ground state and output quantum state corresponding to the first candidate time period (14:00-15:00) is 0.1666, the squared inner product modulus of the ground state and output quantum state corresponding to the second candidate time period (15:00-16:00) is 0.3785, and the squared inner product modulus of the ground state and output quantum state corresponding to the third candidate time period (16:00-17:00) is 0.0369, it indicates that the conversion probability of sending push information is highest in the second candidate time period, followed by the first candidate time period, and lowest in the third candidate time period. The candidate time period with the highest conversion probability can be used as the target push time, and the target push information can be sent to the target object within the target push time.

[0181] In other embodiments of this application, the time correlation parameter can be adjusted based on the activity characteristics of the target object. For example, for target objects with relatively regular activity behavior, a smaller correlation parameter value can be used to extract fine features over a short period; for target objects with relatively dispersed activity behavior, a larger correlation parameter value can be used to extract long-term periodic features. The specific selection of the correlation parameter value can be preset according to the actual application scenario or determined based on statistical analysis of historical data.

[0182] Through the embodiments of this application, based on the historical time distribution of the target object, the conversion probability corresponding to each candidate time period can be determined using quantum computing, and the candidate time period with the highest conversion probability is taken as the target push time. By extracting the temporal correlation features between historical moments through quantum computing, the target push time can be predicted without a large amount of training data, improving the timing accuracy of information push, reducing invalid pushes caused by improper sending timing, and further improving the push effect.

[0183] Based on the above information push method, embodiments of this application also provide an information push device. The following will be combined with... Figure 7 The device is described in detail.

[0184] Figure 7 A schematic block diagram of an information push device according to an embodiment of this application is shown.

[0185] like Figure 7 As shown, the information push device 700 of this embodiment includes a first conversion probability determination module 710, a second conversion probability determination module 720, a target content component combination determination module 730, a target push information determination module 740, and a target push information sending module 750.

[0186] The first conversion probability determination module 710 is used to determine the first conversion probability of the target object based on the object profile data of the target object and the content attribute data of the content to be pushed, among multiple content components. In one embodiment, the first conversion probability determination module 710 can be used to execute the step S210 described above, which will not be repeated here.

[0187] The second conversion probability determination module 720 is used to determine the second conversion probability of each content component combination in at least one content component combination based on the first conversion probability of each content component. In one embodiment, the second conversion probability determination module 720 may be used to perform step S220 described above, which will not be repeated here.

[0188] The target content component combination determination module 730 is used to determine the target content component combination for information push based on the second conversion probability of each content component combination. In one embodiment, the target content component combination determination module 730 can be used to execute the step S230 described above, which will not be repeated here.

[0189] The target push information determination module 740 is used to generate target push information based on the combination of the content to be pushed and the target content component. In one embodiment, the target push information determination module 740 can be used to execute the step S240 described above, which will not be repeated here.

[0190] The target push information sending module 750 is used to send target push information to the target object. In one embodiment, the target push information sending module 750 can be used to perform step S250 described above, which will not be repeated here.

[0191] According to an embodiment of this application, the second conversion probability determination module 720 is further configured to determine the synergy coefficient between at least some of the multiple content components based on the association relationship between different content components; and for any combination of content components, to determine the second conversion probability of the combination of content components based on the first conversion probability of each content component in the combination of content components and the synergy coefficient between the multiple content components in the combination of content components.

[0192] According to an embodiment of this application, the second conversion probability determination module 720 is further configured to determine a conversion correction factor matching each content component based on the degree of influence of the operating environment parameters on the first conversion probability of each content component; for any content component, based on the first conversion probability of the content component and the conversion correction factor, determine the corrected first conversion probability corresponding to the content component; and based on the corrected first conversion probability corresponding to each content component, determine the second conversion probability of each combination of content components.

[0193] According to an embodiment of this application, the first conversion probability determination module 710 is further configured to determine the degree of matching between each content component and the object profile data of the target object based on the component attribute data of each content component among multiple content components; and to determine the first conversion probability of the corresponding content component based on the degree of matching corresponding to each content component.

[0194] According to an embodiment of this application, the information push device 700 further includes a target push time determination module, which is used to determine the time distribution characteristics of multiple historical moments based on at least one candidate time period; determine the third conversion probability matching each candidate time period according to the time distribution characteristics; and take the candidate time period with the highest corresponding third conversion probability as the target time period for information push.

[0195] According to an embodiment of this application, the target push information sending module 750 is further configured to send target push information to the target object based on a target time period.

[0196] According to an embodiment of this application, the target push time determination module is further configured to map multiple historical moments to multiple predicted active moments based on a first mapping scale, and determine the multiple predicted active moments based on a first distribution feature of at least one candidate time period; and / or map multiple historical moments to multiple predicted active time periods based on a second mapping scale, and determine the multiple predicted active time periods based on a second distribution feature of at least one candidate time period; the first distribution feature and / or the second distribution feature constitute a time distribution feature.

[0197] According to embodiments of this application, any multiple modules among the first conversion probability determination module 710, the second conversion probability determination module 720, the target content component combination determination module 730, the target push information determination module 740, and the target push information sending module 750 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the first conversion probability determination module 710, the second conversion probability determination module 720, the target content component combination determination module 730, the target push information determination module 740, and the target push information sending module 750 can be at least partially implemented as hardware circuits, such as field-programmable gate arrays, programmable logic arrays, systems-on-a-chip, systems-on-a-substrate, systems-on-package, application-specific integrated circuits, or other reasonable means of integrating or packaging circuits, or implemented in any one of software, hardware, and firmware methods, or in a suitable combination of any of these methods. Alternatively, at least one of the first conversion probability determination module 710, the second conversion probability determination module 720, the target content component combination determination module 730, the target push information determination module 740, and the target push information sending module 750 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0198] Figure 8 A block diagram schematically illustrates an electronic device suitable for implementing an information push method according to an embodiment of this application.

[0199] like Figure 8 As shown, an electronic device 800 according to an embodiment of this application includes a processor 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory 802 or a program loaded from a storage portion 808 into a random access memory 803. The processor 801 may include, for example, a general-purpose microprocessor, an instruction set processor and / or an associated chipset and / or a dedicated microprocessor. The processor 801 may also include onboard memory for caching purposes. The processor 801 may include a single processing unit or multiple processing units for executing different steps of the method flow according to an embodiment of this application.

[0200] Random access memory 803 stores various programs and data required for the operation of electronic device 800. Processor 801, read-only memory 802, and random access memory 803 are interconnected via bus 804. Processor 801 executes various steps of the method flow according to embodiments of this application by executing programs stored in read-only memory 802 and / or random access memory 803. It should be noted that the programs may also be stored in one or more memories other than read-only memory 802 and random access memory 803. Processor 801 may also execute various steps of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0201] According to embodiments of this application, the electronic device 800 may further include an input / output interface 805, which is also connected to a bus 804. The electronic device 800 may also include one or more of the following components connected to the input / output interface 805: an input section 806 including a keyboard, mouse, etc.; an output section 807 including a cathode ray tube, liquid crystal display, etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the input / output interface 805 as needed. A removable medium 811, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 810 as needed so that computer programs read from it can be installed into the storage section 808 as needed.

[0202] Embodiments of this application also provide a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0203] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof. In embodiments of this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include the read-only memory 802 described above, and / or random access memory 803, and / or one or more memories other than read-only memory 802 and random access memory 803.

[0204] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to cause the computer system to implement the methods provided in the embodiments of this application.

[0205] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via the communication section 809, and / or installed from a removable medium 811. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0206] In embodiments of this application, the computer program can be downloaded and installed from a network via communication section 809, and / or installed from removable medium 811. When the computer program is executed by processor 801, it performs the functions defined in the system of embodiments of this application. According to embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0207] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0208] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0209] Those skilled in the art will understand that the features described in the various embodiments of this application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments of this application can be combined and / or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

Claims

1. An information push method, characterized in that, include: Based on the object profile data of the target object and the content attribute data of the content to be pushed, the first conversion probability of the target object based on each of the multiple content components is determined, wherein the conversion probability indicates the predicted probability that the target object will perform a conversion operation on the content to be pushed; Based on the first conversion probability of each content component, a second conversion probability of each content component combination in at least one content component combination is determined, wherein each content component combination includes at least some of the content components among the plurality of content components. Based on the second conversion probability of each of the aforementioned content component combinations, a target content component combination for information push is determined; and Based on the combination of the content to be pushed and the target content component, target push information is generated and sent to the target object.

2. The information push method according to claim 1, characterized in that, Determining the second conversion probability of each content component combination in at least one content component combination based on the first conversion probability of each content component includes: Based on the relationships between different content components, a synergy coefficient is determined among at least some of the content components. The synergy coefficient indicates the degree of influence of any one of the at least some content components on the first conversion probability of other content components when the at least some content components are used in combination. For any combination of content components, the second conversion probability of the combination of content components is determined based on the first conversion probability of each content component in the combination of content components and the synergy coefficient between the multiple content components in the combination of content components.

3. The information push method according to claim 1, characterized in that, Determining the second conversion probability of each content component combination in at least one content component combination based on the first conversion probability of each content component includes: Based on the degree of influence of the operating environment parameters on the first conversion probability of each content component, a conversion correction factor matching each content component is determined; For any content component, the corrected first conversion probability corresponding to the content component is determined based on the first conversion probability of the content component and the conversion correction factor; The second conversion probability of each combination of the content components is determined based on the modified first conversion probability corresponding to each of the content components.

4. The information push method according to claim 1, characterized in that, The determination of the first conversion probability of the target object based on the object profile data of the target object and the content attribute data of the content to be pushed, including: Based on the component attribute data of each of the multiple content components, determine the degree of matching between each content component and the object profile data of the target object; Based on the matching degree corresponding to each content component, the first conversion probability of the corresponding content component is determined.

5. The information push method according to any one of claims 1-4, characterized in that, The object profile data also includes object behavior data of the target object, which indicates that the target object has active behavior based on multiple historical moments; sending the target push information to the target object includes: The multiple historical moments are determined based on the temporal distribution characteristics of at least one candidate time period; Based on the time distribution characteristics, a third conversion probability matching each of the candidate time periods is determined; The candidate time period with the highest probability of the third conversion is selected as the target time period for information push; and Based on the target time period, the target push information is sent to the target object.

6. The information push method according to claim 5, characterized in that, The determination of the plurality of historical moments based on the temporal distribution characteristics of at least one candidate time period includes: Based on a first mapping scale, the plurality of historical moments are mapped to a plurality of predicted active moments, and the plurality of predicted active moments are determined based on a first distribution feature of at least one candidate time period, wherein the first mapping scale indicates that the m-th historical moment adjacent to any target historical moment is taken as the predicted active moment, where m is an integer greater than 0; and / or Based on the second mapping scale, the multiple historical moments are mapped to multiple predicted active time periods, and the multiple predicted active time periods are determined based on the second distribution feature of at least one candidate time period. The second mapping scale indicates that any target historical moment and the nth historical moment adjacent to the target historical moment are used as boundary points to obtain the corresponding predicted active time period, where n is an integer greater than 0. The first distribution feature and / or the second distribution feature constitute the time distribution feature.

7. The information push method according to claim 6, characterized in that, The first distribution feature includes at least one of the following features: the number of active moments, the dispersion of active moments, and the time period in which active moments are concentrated; The second distribution feature includes at least one of the following features: the number of active time periods, the percentage of active time periods covered, and the distribution of active time periods across candidate time periods.

8. An information push device, characterized in that, The device includes: The first conversion probability determination module is used to determine the first conversion probability of the target object based on the object profile data of the target object and the content attribute data of the content to be pushed, wherein the conversion probability indicates the predicted probability that the target object will perform a conversion operation on the content to be pushed; The second conversion probability determination module is used to determine the second conversion probability of each content component combination in at least one content component combination based on the first conversion probability of each content component, wherein each content component combination includes at least some content components among the plurality of content components. The target content component combination determination module is used to determine the target content component combination for information push based on the second conversion probability of each of the content component combinations. The target push information determination module is used to generate target push information based on the combination of the content to be pushed and the target content component; and The target push information sending module is used to send the target push information to the target object.

9. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 7.

Images