WO2025103131A1 - Product strategy delivery - Google Patents

Abstract

Disclosed in the embodiments of the present description are a product strategy delivery method and apparatus, and a storage medium and a terminal. The method comprises: on the basis of a user action sequence, determining at least one target professional factor that meets a preset popularity condition, wherein the professional factor is a factor that provides a professional explanation for a product; determining at least one candidate product strategy corresponding to each target professional factor, wherein the candidate product strategy at least comprises a target professional factor and a strategy element corresponding to the target professional factor; and determining, from among all candidate product strategies, a candidate product strategy that meets a preset delivery condition as a target product strategy to be delivered. Since a professional factor can provide a professional explanation for a product strategy, a corresponding product strategy determined on the basis of a target professional factor can explicitly present an intrinsic professional concept of a product to a user.

Description

Product strategy launch Technical Field

The embodiments of this specification relate to the field of computer Internet technology, and in particular, to a product strategy delivery method, device, storage medium, and terminal.

Background Art

With the development of the Internet industry, more and more users conduct financial transactions through online trading platforms, such as buying daily necessities on online shopping platforms, buying and selling funds and stocks on financial institution platforms, etc. Taking financial institution platforms as an example, they usually recommend some financial products to users to provide users with professional and reliable purchase guidance, thereby improving user experience and user satisfaction. Therefore, a product strategy delivery method is needed that can make decisions based on a professional perspective to optimize the recommendation strategy, so as to help users quickly obtain high-quality product information.

Summary of the invention

The embodiments of this specification provide a product strategy delivery method, device, storage medium, and terminal, which can solve the technical problem in related technologies that users cannot obtain professional information about recommended products.

In a first aspect, an embodiment of the present specification provides a product strategy delivery method, the method comprising: based on a user behavior sequence of at least one user for a historical product strategy, determining at least one target professional factor that meets a preset popularity condition from at least one preset professional factor, the professional factor being a factor for professionally interpreting a product; respectively determining at least one candidate product strategy corresponding to each target professional factor, the candidate product strategy comprising at least a target professional factor and a strategy element corresponding to the target professional factor; and determining, from all candidate product strategies, a candidate product strategy that meets the preset delivery condition as the target product strategy to be delivered.

In a second aspect, an embodiment of the present specification provides a product strategy delivery device, which includes: a professional factor selection module, which is used to determine at least one target professional factor that meets a preset popular condition from at least one preset professional factor based on a user behavior sequence of at least one user for a historical product strategy, wherein the professional factor is a factor for professionally interpreting a product; a candidate strategy determination module, which is used to respectively determine at least one candidate product strategy corresponding to each target professional factor, wherein the candidate product strategy includes at least a target professional factor and a strategy element corresponding to the target professional factor; a product strategy delivery module, which is used to determine, from all candidate product strategies, a candidate product strategy that meets the preset delivery condition as the target product strategy to be delivered.

In a third aspect, an embodiment of the present specification provides a computer program product comprising instructions, which, when executed on a computer or a processor, enables the computer or the processor to execute the steps of the above method.

In a fourth aspect, an embodiment of the present specification provides a computer storage medium, wherein the computer storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing the steps of the above-mentioned method.

In a fifth aspect, an embodiment of the present specification provides a terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is suitable for being loaded by the processor and executing the steps of the above method.

The beneficial effects of the technical solutions provided by some embodiments of this specification include at least: the embodiments of this specification provide a product strategy delivery method, based on the user behavior sequence of at least one user for the historical product strategy, At least one target professional factor that meets the preset popular conditions is determined from at least one preset professional factor, and the professional factor is a factor that professionally explains the product; at least one candidate product strategy corresponding to each target professional factor is determined, and the candidate product strategy at least includes the target professional factor and the strategy element corresponding to the target professional factor; from all candidate product strategies, the candidate product strategy that meets the preset launch conditions is determined as the target product strategy to be launched. Since professional factors can provide professional explanations for product strategies, when the popular target professional factors determined according to the user behavior sequence are used as the core of the product strategy, the corresponding product strategy determined according to the target professional factor can explicitly display the inherent professional concept of the product to the user, so that the user can quickly obtain product information according to the product strategy, which can solve the user's urgent need for explainable product professional information.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

FIG1 is an exemplary system architecture diagram of a product strategy delivery method provided in an embodiment of this specification.

FIG2 is a flow chart of a product strategy delivery method provided in an embodiment of this specification.

FIG3 is a flowchart of a product strategy delivery method provided in an embodiment of this specification.

FIG. 4 is a schematic diagram of a tree structure of a policy element assembly rule provided in an embodiment of this specification.

FIG. 5 is a schematic diagram of a display interface for preselecting strategy specifications provided in an embodiment of this specification.

FIG6 is a logic flow diagram of a product strategy delivery method provided in an embodiment of this specification.

FIG. 7 is a diagram showing a behavior link modeling in a strategy delivery model provided in an embodiment of this specification.

FIG8 is a model structure diagram of a strategy delivery model provided in an embodiment of this specification.

FIG. 9 is a structural block diagram of a product strategy delivery device provided in an embodiment of this specification.

FIG. 10 is a schematic diagram of the structure of a terminal provided in an embodiment of this specification.

DETAILED DESCRIPTION

In order to make the features and advantages of the embodiments of this specification more obvious and easy to understand, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the embodiments of this specification.

When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of this specification. Instead, they are only examples of devices and methods consistent with some aspects of the embodiments of this specification as detailed in the attached claims.

With the development of the Internet industry, more and more users conduct financial transactions through online trading platforms, such as buying daily necessities on online shopping platforms, buying and selling funds and stocks on financial institution platforms, etc. Taking financial institution platforms as an example, they usually recommend some financial products to users to provide users with professional and reliable purchase guidance, thereby improving user experience. User experience and user satisfaction. A common recommendation guidance method is to place a configured product strategy on the user browsing page. The product strategy is an instantiation of the strategy consisting of text, charts and action points used to introduce product conditions and information in the marketing recommendation and advertising delivery system. The product strategy in the financial scenario can also be called the financial strategy to attract users to click and make subsequent purchases.

Currently, in some common product strategy delivery plans, manual configuration and input are usually performed by operators during the strategy material configuration stage. During the manual configuration process, operators manually select a combination of strategy elements such as text and charts to be displayed to users based on their own prior experience. When configuring the strategy, the main considerations are display effect, delivery location, etc.

However, for some highly professional scenarios, such as financial scenarios, most users who purchase financial fund products usually pay attention, click, and purchase based on some basic knowledge, such as returns, risks, and recent trends. That is, in such scenarios, users pay more attention to the intrinsic attributes and professional analysis of the product itself, and hope to intuitively see the intrinsic professional attributes of the product in the strategy. However, the existing strategy delivery solutions do not explicitly display the intrinsic professional attributes that are strongly related to the product itself to users, resulting in the inability of existing strategy delivery solutions to meet users' urgent needs for product explainability.

Therefore, an embodiment of the present specification provides a product strategy delivery method, which determines at least one target professional factor that meets preset popularity conditions based on user behavior sequences, where the professional factor is a factor that provides professional interpretation of the product; determines at least one candidate product strategy corresponding to each target professional factor, where the candidate product strategy includes at least the target professional factor and the strategy element corresponding to the target professional factor; and from all candidate product strategies, determines the candidate product strategy that meets the preset delivery conditions as the target product strategy to be delivered, so as to solve the technical problem that the above-mentioned users cannot obtain the professional information of the recommended products.

Please refer to FIG. 1 , which is an exemplary system architecture diagram of a product strategy delivery method provided in an embodiment of this specification.

As shown in FIG1 , the system architecture may include a terminal 101, a network 102, and a server 103. The network 102 is used to provide a medium for a communication link between the terminal 101 and the server 103. The network 102 may include various types of wired communication links or wireless communication links, for example, the wired communication link includes an optical fiber, a twisted pair, or a coaxial cable, and the wireless communication link includes a Bluetooth communication link, a Wireless-Fidelity (Wi-Fi) communication link, or a microwave communication link.

The terminal 101 can interact with the server 103 through the network 102 to receive messages from the server 103 or send messages to the server 103, or the terminal 101 can interact with the server 103 through the network 102 to receive messages or data sent by other users to the server 103. The terminal 101 can be hardware or software. When the terminal 101 is hardware, it can be various electronic devices, including but not limited to smart watches, smart phones, tablet computers, laptop portable computers and desktop computers. When the terminal 101 is software, it can be installed in the electronic devices listed above, which can be implemented as multiple software or software modules (for example: used to provide distributed services), or it can be implemented as a single software or software module, which is not specifically limited here.

In the embodiment of the present specification, the terminal 101 first determines at least one target professional factor that meets the preset popularity condition from at least one preset professional factor based on the user behavior sequence of at least one user for the historical product strategy. The professional factor is a factor for professional interpretation of the product; then, the terminal 101 needs to determine at least one candidate product strategy corresponding to each target professional factor respectively, and the candidate product strategy includes at least the target professional factor and the strategy element corresponding to the target professional factor; finally, the terminal 101 determines the candidate product strategy that meets the preset delivery conditions from all the candidate product strategies as the target product strategy to be launched.

The server 103 may be a business server that provides various services. It should be noted that the server 103 may be hardware or software. When the server 103 is hardware, it may be implemented as a distributed server cluster consisting of multiple servers, or it may be implemented as a single server. When the server 103 is software, it may be implemented as multiple software or software modules (for example, for providing distributed services), or it may be implemented as a single software or software module, which is not specifically limited here.

Alternatively, the system architecture may not include the server 103. In other words, the server 103 may be an optional device in the embodiments of this specification, that is, the method provided in the embodiments of this specification may be applied to a system structure that only includes the terminal 101, and the embodiments of this specification do not limit this.

It should be understood that the number of terminals, networks, and servers in FIG. 1 is only illustrative, and any number of terminals, networks, and servers may be used according to implementation requirements.

Please refer to Figure 2, which is a flowchart of a product strategy delivery method provided by an embodiment of this specification. The execution subject of the embodiment of this specification can be a terminal that executes product strategy delivery, or a processor in the terminal that executes the product strategy delivery method, or a product strategy delivery service in the terminal that executes the product strategy delivery method. For the convenience of description, the specific execution process of the product strategy delivery method is introduced below by taking the execution subject as a processor in a terminal as an example.

As shown in FIG. 2 , the product strategy delivery method may include at least the following steps.

S202. Based on a user behavior sequence of at least one user for a historical product strategy, determine at least one target professional factor that meets a preset popularity condition from at least one preset professional factor, where the professional factor is a factor for professionally explaining the product.

Alternatively, in the face of users' need for interpretability of product strategies in some professional scenarios, conventional strategy delivery schemes that completely rely on expert experience and black-box strategy delivery models have gradually failed to meet users' deep-seated needs for product strategies. Therefore, in order to help users quickly and effectively select suitable products and complete subsequent purchases and other conversions, the implicit professional factor data in the product strategy can be presented to users in the form of a combination strategy of more trust-enhancing text, charts and action points. Among them, professional factors are factors that professionally explain products, which are data performances in the objective definition of products from the perspective of relevant industry professionals. Different categories of factors explain products from different dimensions. For example, for a financial product, fundamental factors explain fund returns from multiple aspects such as valuation, growth, and profitability. Quantity and price factors are factors constructed around technical indicators such as price and trading volume. Product strategies constructed through these professional factors allow users to quickly and visually obtain the intrinsic benefits and risk information of products through the combined visualized product strategies.

Optionally, in order to present the implicit professional factor data that users are highly concerned about in the product strategy to users in an explainable and explicit manner, so as to increase the user's click, browse and subsequent purchase conversion rate of the product strategy, the target professional factor that the user group is interested in among all the professional factors of the product can be determined first, so as to facilitate the subsequent user-based The highly concerned target professional factors show users the preferred product strategies. Before determining the target professional factors, it is necessary to first clarify the preset professional factors that need to be analyzed for the product in the specific strategy recommendation scenario.

Furthermore, for a product, its professional factors not only have the objective dimension of the product itself, but also need to consider the subjective dimension of the user's preference for the product, that is, the preset professional factors include but are not limited to product professional factors and user professional factors, among which the product professional factors are professional factors of the product dimension defined based on the basic professional knowledge in the product industry field, and the user professional factors are professional factors of the user dimension defined based on user behavior data. For example, for financial scenarios, the product professional factors of financial products can be subdivided into yield, Sharpe value, BRAR (popularity willingness index), CR (energy index), VR (capacity ratio), MAR (potential growth and decline index), Metz line, VCI (Jiaqing variation rate), MAD (William long and short strength line), etc. These product professional factors are professionally defined parameter items in the financial industry, which are used to reflect some objective conditions of the financial product itself; and for users, the user professional factors related to them are also very concerned, such as the amount of user holding income, user holding yield, etc., which are all user-dimensional professional factors of a product defined based on user behavior data.

Specifically, the preset professional factors can be obtained by collecting basic data such as the product's trading data, market data, fundamental data and financial data. Each preset professional factor has its own special calculation formula. By calculating the basic data, the specific situation of each preset professional factor of each product can be determined. For example, through the Sharpe value calculation formula of financial products, the Sharpe value of this financial product can be calculated as a high Sharpe value or a low Sharpe value. Then, when a high Sharpe value is selected as the target professional factor of the user's concern, products with high Sharpe values can be displayed to users in the product strategy.

Optionally, when determining the target professional factors that the user group is interested in among all the professional factors of the product, expert experience selection and algorithm automatic selection can be used, wherein the expert experience selection is that the operation staff manually selects the target professional factors from the preset professional factors, specifically judging the preset professional factors that meet the preset popular conditions as the target professional factors based on the expert's prior experience; the algorithm automatic selection is that the algorithm calculates the user behavior sequence of at least one user for the pushed historical product strategy based on the pre-collected user behavior sequence, and determines at least one target professional factor that meets the preset popular conditions from at least one preset professional factor. The preset popular conditions used in the judgment process are used to measure the degree of attention, that is, the popularity, of various professional factors in the user group. If the preset popular conditions are met, it means that the user group is more interested in the professional factor, and the relevant product strategies of such factors can obtain more user behaviors such as clicks and conversions.

The above user behavior sequence is a sequence composed of user behaviors and professional factors corresponding to user behaviors obtained according to user behavior data. For example, for financial product strategy A, financial product strategy A includes professional factors 1 and professional factors 2. User X's behavior on financial product strategy A is "click", then the user behavior sequence of click behavior is [Product1[professional factor 1, professional factor 2]], and user Y's behavior on financial product strategy A is "purchase conversion", then the user behavior sequence of conversion behavior is [Product2[professional factor 3, professional factor 4]]. This means that the user behavior sequence can reflect the correlation between the user's behavior on historical product strategies and the professional factors in the historical product strategies. Then, combined with the preset popular conditions, at least one target professional factor that meets the preset popular conditions can be determined from the preset professional factors for subsequent product strategy assembly.

S204. Determine at least one candidate product strategy corresponding to each target professional factor, wherein the candidate product strategy includes at least the target professional factor and the strategy element corresponding to the target professional factor.

Optionally, after determining the target professional factors that meet the preset popular conditions, the corresponding candidate product strategies can be further determined according to the target professional factors. The candidate product strategies include at least the target professional factors and the strategy elements corresponding to the target professional factors. That is, at this time, the target professional factors are mainly used as the core to determine the strategy elements that can be assembled into product strategies with the target professional factors, and then assembled into candidate product strategies that can be launched according to some assembly rules. It should be noted that the product strategies in the embodiments of this specification are complete product strategies that can be directly launched, in which each strategy element and each target professional factor has a fixed display position and display attributes.

Furthermore, in general, considering that the meanings of some professional factors may be repeated or contradictory, in order to avoid confusion and redundancy in the content of product strategies, the candidate product strategies may include only one target professional factor. In this case, each target professional factor corresponds to at least one candidate product strategy. When it is necessary to increase the professional factor information in the candidate product strategies, the mutual influence of the professional factors can be considered in advance, and professional factor coexistence rules can be established to stipulate which professional factors can appear at the same time and which professional factors do not need to appear at the same time. In this way, two or even more target professional factors can appear in a candidate product strategy. In this case, the content richness of a product strategy can be increased, and more target professional factor information can be displayed to users through a limited strategy space, thereby improving the efficiency of users in obtaining product intrinsic factor information.

S206: Determine, from among all the candidate product strategies, a candidate product strategy that meets a preset launch condition as the target product strategy to be launched.

Optionally, from all candidate product strategies, a target product strategy to be launched is finally determined. The target product strategy needs to meet preset launch conditions. The preset launch conditions can be specific conditions determined based on user preferences, current market conditions, etc., and can change with scenario requirements. When selecting a target product strategy, it can be manually selected by operations staff based on expert experience, or it can be automatically calculated and decided with the help of algorithms, neural network models, etc., so as to quickly and efficiently determine the preferred target product strategy for launch. In this way, when the popular target professional factors are determined based on the user behavior sequence as the core of the product strategy, the final target product strategy can explicitly display the inherent professional concepts of the product to the user, so that the user can quickly obtain product information based on the product strategy, which can solve the user's urgent need for explainable product professional information.

In an embodiment of the present specification, a product strategy delivery method is provided. Based on the user behavior sequence of at least one user for a historical product strategy, at least one target professional factor that meets a preset popular condition is determined from at least one preset professional factor, and the professional factor is a factor that professionally explains the product; at least one candidate product strategy corresponding to each target professional factor is determined respectively, and the candidate product strategy includes at least the target professional factor and the strategy element corresponding to the target professional factor; from all candidate product strategies, the candidate product strategy that meets the preset delivery condition is determined as the target product strategy to be delivered. Since the professional factor can provide a professional explanation for the product strategy, when the popular target professional factor determined according to the user behavior sequence is used as the core of the product strategy, the corresponding product strategy determined according to the target professional factor can explicitly display the inherent professional concept of the product to the user, so that the user can quickly obtain product information according to the product strategy, which can solve the user's urgent need for explainable product professional information.

Please refer to FIG. 3 , which is a flowchart of a product strategy delivery method provided in an embodiment of this specification.

As shown in FIG3 , the product strategy delivery method may include at least the following steps.

S302: Determine at least one preset The professional factors respectively correspond to at least one target user behavior and the number of user actions for each target user behavior.

Optionally, in order to determine the target professional factors that meet the preset popularity conditions, it is first necessary to accurately measure the popularity of each preset professional factor in the user group. From the introduction of the above description embodiment, it can be understood that the user behavior sequence includes the behavior performed by the user and the preset professional factor corresponding to the behavior, and the more behaviors a user performs on a preset professional factor, the higher the user's attention to the preset professional factor, which means that the preset professional factor is more popular. Then the attention of each preset professional factor can be judged by the number of user actions corresponding to important target user behaviors, that is, according to the user behavior sequence of at least one user for the historical product strategy, at least one target user behavior corresponding to at least one preset professional factor and the number of user actions of each target user behavior are determined. The target user behavior is an important user behavior, such as clicking, browsing, collecting, paying attention, buying, adding positions, etc. The number of user actions of each target user behavior is the total number of actions taken by the user group for the target user behavior in all the collected data.

S304. Determine, according to the number of user actions corresponding to each preset professional factor, a preset professional factor that meets a preset popular condition from at least one preset professional factor as a target professional factor.

Optionally, when the total number of user actions of all target user behaviors corresponding to the preset professional factor is larger, it can be explained to a certain extent that the preset professional factor is more popular. That is, according to the number of user actions corresponding to each preset professional factor, it is possible to determine the preset professional factor that meets the preset popularity condition from at least one preset professional factor as the target professional factor.

Furthermore, considering that different target user behaviors may have different importance levels based on the purpose of recommending product strategies, for example, when showing product strategies to users, the biggest purpose may be to hope that users will click, buy, add positions, and other behaviors that are conducive to substantial conversion, while browsing discussion areas and passive browsing interfaces where strategies are located are behaviors that cannot represent the actual attention of users. In order to more accurately calculate the popularity of the preset professional factors, corresponding popularity weights can be assigned to each target user behavior according to the importance of the target user behavior. For example, important click behaviors and purchase behaviors have higher weights, while unimportant browsing behaviors have lower weights. Then, the popularity scores of each preset professional factor are calculated based on the number of user actions corresponding to each preset professional factor and the popularity weights of each target user behavior. Specifically, "PopularityScore_factorn" is used to represent the popularity score of the preset professional factor n, "clk_uv" represents the user click rate, and its corresponding weight is w ₁ , "trans_uv" represents the user conversion rate, and its corresponding weight is w ₂ , and so on. Each target user behavior has a corresponding popularity weight, so the popularity score calculation expression of the preset professional factor n is:
PopularityScore_factorn= _w1 ×clk_uv+w2×trans_uv+….

After calculating the popularity score of each preset professional factor, the preset professional factor whose popularity score meets the preset popularity condition can be determined as the target professional factor according to the popularity score of each preset professional factor. In actual scenarios, the preset popularity condition can be set as the K preset professional factors with the top K popularity scores as the target professional factors, so as to control the number of target professional factors and improve the efficiency of subsequent product strategy assembly.

S306. Input each target professional factor into a policy element extraction model, and determine at least one policy element corresponding to each target professional factor based on the policy element extraction model.

Optionally, after determining the target professional factor, the relevant strategy elements are determined based on the target professional factor, so that the subsequent product strategy composed of the target professional factor and the strategy elements corresponding to the target professional factor can explicitly display the target professional factor to the user, so as to meet the user's need for explainability of the product strategy. In the background of product strategy delivery, a large number of backup strategy elements will be prepared, including at least text, images, charts, and action points, such as the text "Holding firmly is the easiest way to avoid actual losses..."; the image chart can be a product-related line chart or data chart; the action point is used to reflect the user behavior expected by the strategy. Then, when generating the strategy elements corresponding to the target professional factor, they can be extracted from the pre-prepared element library. The specific extraction process can use the strategy element extraction model. The strategy element extraction model is a trained and converged neural network model. Based on the strategy element extraction model, at least one strategy element corresponding to each target professional factor can be determined respectively.

Optionally, the policy element extraction model f(text, image, chart, action) is trained based on at least one sample policy element and the standard professional factor label corresponding to each sample policy element. That is, when training the policy element extraction model, the sample policy element is used as input, and the standard professional factor label is marked for each sample policy element. The policy element extraction model outputs the predicted professional factor label of each sample policy element based on the sample policy element, and the model loss is calculated based on the standard professional factor label and the predicted professional factor label of each sample policy element. The model loss is used to train the policy element extraction model for parameter adjustment, so that the policy element extraction model learns the corresponding relationship between policy elements and professional factors. After the policy element extraction model converges, it receives the target professional factor as input, and can extract at least one policy element corresponding to each target professional factor based on the knowledge learned by itself.

It should be noted that in actual scenarios, some text information, graphic resources, etc. with higher sensitivity may involve issues such as copyright and usage rights. These policy elements can be subject to a layer of legal and applicable scenario review after being extracted to avoid risks that exist in some actual scenarios. The review rules are usually set based on expert experience, and the specific rule content needs to be set according to the real-time needs of the actual scenario. The embodiments of this specification do not limit this.

S308. Determine at least one candidate product strategy corresponding to each target professional factor based on each target professional factor and the strategy elements corresponding to each target professional factor.

Optionally, after determining the strategy elements corresponding to each target professional factor, it is necessary to assemble each target professional factor and the strategy elements into a product strategy, that is, to determine at least one candidate product strategy corresponding to each target professional factor based on each target professional factor and the strategy elements corresponding to each target professional factor. When assembling strategy elements, it is considered that the color attributes and style attributes of the elements may cause conflicts when some elements appear at the same time. For example, in a product strategy "Bad case" with a poor display effect, the combination of light gray background and white text makes the text content display very unclear, while in a product strategy "Good case" with a better display effect, the combination of light gray background and black text makes the visual display effect of the product strategy better.

Then, in order to assemble a product strategy with good display effect, please refer to FIG4, which is a tree structure diagram of a strategy element assembly rule provided in an embodiment of this specification. As shown in FIG4, the attribute type of each strategy element can be used as a large node according to the attribute of the strategy element, for example, the strategy template attribute is used as a large node A, the text font attribute is used as a large node B, the text color attribute is used as a large node C, and the background color attribute is used as a large node D; under the large node B of the text font attribute, there are 5 fonts, namely _b1 , _b2 , _b3 , _b4 , and _b5 . Under the large node C of the text color attribute, there are three colors, namely c ₁ , c ₂ , and c ₃ ; under the large node D of the background color attribute, there are two background styles, namely d ₁ and d ₂ ; at this time, the connecting lines between each font, text color, and background style represent a parameter attribute combination of an element in a product strategy, so there are a total of 5×3×2=30 possible combinations of product strategies. In the tree structure, "Good case" can be used as the preferred path to solve the problem of assembling strategy elements.

Furthermore, the above tree structure can obtain element combination rules by learning positive sample product strategy cases (sample "Good case") and negative sample product strategy cases (sample "Bad case"), where these sample product strategy cases can be cases manually adjusted in the early strategy assembly process in actual applications. Finally, visual prior experience is learned based on these case tree structures to narrow the search space for each parameter element combination, so that the element combination rules are used when searching for the attribute combination path of the strategy element. Based on the element combination rules, each target professional factor and the strategy elements corresponding to each target professional factor are respectively combined to determine at least one candidate product strategy corresponding to each target professional factor that satisfies the element combination rules. In this way, the candidate product strategy corresponding to each target professional factor is a candidate product strategy that is related to each target professional factor and shows good results.

Optionally, the tree structure can determine the combination of policy elements corresponding to each target professional factor. After that, it is necessary to specifically place these policy elements and each target professional factor in the pre-determined pre-selected policy specifications to complete the assembly of a complete candidate product strategy. The pre-selected policy specifications are product policy templates that need to be launched and are manually determined by the operator in the product policy template library. The selection of pre-selected policy specifications is usually related to user needs, product information, and page functions. For example, please refer to Figure 5, which is a schematic diagram of a display interface of a pre-selected policy specification provided in an embodiment of this specification. As shown in Figure 5, there is a product policy template library open to operation staff in the terminal display interface. The operation staff can select L pre-selected policy specifications as needed as the target template for generating the product strategy. According to these pre-selected policy specifications, each target professional factor and the policy elements corresponding to each target professional factor are respectively assembled by Cartesian product, thereby determining all candidate product strategies that can be used for launch generated by this product strategy.

Optionally, please refer to Figure 6, which is a logical flow chart of a product strategy delivery method provided by an embodiment of this specification. As shown in Figure 6, when it is necessary to deliver a product strategy, enter the starting process of the product strategy delivery method, first determine the preset professional factors based on basic data such as transaction data, market data, fundamental data and financial data in the application scenario, and pre-set the product strategy template library; when generating the product strategy, determine the target professional factors from the preset professional factors, and select the pre-selected strategy specifications from the product strategy template library; generate the strategy elements corresponding to each target professional factor; then combine the pre-selected strategy specifications, and use Cartesian products to assemble all candidate product strategies related to the target professional factors.

S3010. Input each candidate product strategy into a strategy delivery model, and determine, based on the strategy delivery model, a candidate product strategy that meets preset delivery conditions as a target product strategy to be delivered.

Optionally, among all the candidate product strategies, user behavior prediction can be further performed on each candidate product strategy to predict the user behavior that the user will make when facing these candidate product strategies. According to the user behavior prediction results, the candidate product strategy that meets the preset delivery conditions is selected as the target product strategy to be delivered. The preset delivery conditions are specifically set according to the expected user behavior, and the delivery value of each candidate product strategy can be measured.

Specifically, when determining the target product strategy, the strategy delivery model can be used to input each candidate product strategy into the strategy delivery model, and user behavior prediction can be performed based on the strategy delivery model, thereby determining the candidate product strategy that meets the preset delivery conditions as the target product strategy to be delivered. Usually, there are multiple possibilities for user behavior for a product strategy. For example, in the financial fund scenario, by analyzing the user's fund subscription behavior, after the user clicks, there are usually multiple behaviors that are highly correlated with delayed purchase behavior, such as self-selection, fixed investment, and subscription. Delayed purchase refers to the situation in which the purchase conversion occurs within a period of time after the user behavior that often exists in financial scenarios. Then, the PostClick behavior indicator can be divided into decisive behavior (DAction) and other behaviors (OAction) according to whether it is related to delayed purchase behavior. The decisive behavior includes: self-selection behavior, fixed investment behavior, and subscription behavior, and other behaviors include: discussion area browsing behavior, toolbar browsing behavior, and consultation behavior, etc. It can be understood that the decisive behavior and other behaviors are divided according to the probability of association with delayed purchase conversion behavior.

Further, please refer to Figure 7, which is a behavior link modeling in a strategy delivery model provided by an embodiment of this specification. As shown in Figure 7, in the strategy delivery model, user behavior can be established between clicks and delayed purchases, forming a behavior link modeling of "delivery exposure → click → D(O)Action → delayed purchase → delayed GMV (total transaction amount)". On this basis, based on the behavior link modeling, the strategy delivery model can be based on full-space multi-objective modeling, making full use of the behavior link of "delivery exposure → click → D(O)Action → delayed purchase → delayed GMV (total transaction amount)", and decomposing it into multiple target behavior links, specifically "exposure → click", "click → DAction", "DAction → delayed purchase", "OAction → delayed purchase", "delayed purchase → delayed GMV", these 5 target behavior links, each target behavior link corresponds to a prediction subnetwork.

Optionally, please refer to FIG. 8 , which is a model structure diagram of a strategy delivery model provided in an embodiment of this specification. As shown in Figure 8, the initialization input of the model is the original one-hot encoded feature input. In the shared module, the one-hot feature encoding is embedded into a feature representation. The main structure of the strategic delivery model includes the following three modules: (1) Shared Embedding Module: SEM represents the embedding feature representation of sparse features shared by all prediction sub-networks. For example, the embedding feature representation of features such as user ID, habitual preferences, and personality characteristics is shared by all prediction sub-networks, which can alleviate the data sparsity problem faced by a single behavior link to a certain extent. (2) Decomposed Prediction Module: DPM is composed of the above five prediction sub-networks. Each prediction sub-network estimates the user behavior estimates of the five target behavior links: "exposure→click", "click→DAction", "DAction→delayed purchase", "OAction→delayed purchase", and "delayed purchase→delayed GMV". (3) Sequential Composition Module: SCM finally integrates four expected prediction values based on the user behavior estimates of each prediction sub-network. They are:

The click-through rate of product strategy i in the link “exposure→click” This is determined by the "exposure→click" prediction subnetwork in the model, and its specific value is represented by Y ₁ ;

The decisive behavior rate of product strategy i in the "exposure→DAction" link Determinative behavior rate In the model, it is determined by the estimated value Y ₁ of the "exposure→click" prediction subnetwork and the estimated value Y ₂ of the "click→DAction" prediction subnetwork. The calculation formula is expressed as Y ₁ Y ₂ ;

The transaction rate of product strategy i in the link of “exposure→delayed purchase” Transaction rate In the model, it is determined by the estimated value Y ₁ of the “exposure→click” prediction subnetwork, the estimated value Y ₂ of the “click→DAction” prediction subnetwork, the estimated value Y ₃ of the “DAction→delayed purchase” prediction subnetwork, and the estimated value Y ₄ of the “OAction→delayed purchase” prediction subnetwork. The calculation formula is:
Y ₁ [(1-Y ₂ )Y ₄ +Y ₂ Y ₃ ];

And the predicted total transaction amount of product strategy i in "Exposure → Delayed GMV" It needs to be determined by the estimated values Y ₁ , Y ₂ , Y ₃ , Y ₄ , and Y ₅ of the five prediction sub-networks. The calculation formula is:
Y ₁ [(1-Y ₂ )Y ₄ +Y ₂ Y ₃ ]Y ₅ .

Optionally, when training the strategy delivery model, the predicted loss is calculated based on the predicted value output for at least one sample product strategy and the standard value corresponding to each sample product strategy. More specifically, taking the sample product strategy as product strategy i as an example, the click rate based on the sample product strategy i Combined with the corresponding real click rate, we get sub-loss Loss ₁ and decisive behavior rate Combined with the corresponding real decisive behavior rate, we get the sub-loss Loss ₂ and the transaction rate Combined with the corresponding real transaction rate, we get sub-loss Loss ₃ and the predicted value of total transaction amount. Combined with the corresponding actual total transaction amount, the sub-loss Loss ₄ is obtained. Then, according to the loss weights corresponding to the four preset sub-losses, the _total Loss of the final model's prediction is calculated. The model is trained once using the _total Loss. After multiple trainings, the converged strategy delivery model can be used in real scenarios to make product strategy delivery decisions.

Please refer to Figure 6. After the preferred target product strategy is determined, the target product strategy is launched, and the entire product strategy launch process is completed. In addition, after the target product strategy is launched, the target product strategy can also be monitored to collect user behavior feedback data obtained by the target product strategy. The collected user data can be used to update and iterate the algorithms and models involved in the product strategy launch plan, thereby continuously outputting better product strategy plans for users and improving user experience.

In an embodiment of the present specification, a product strategy delivery method is provided. First, different popularity weights are assigned to target user behaviors of different importance, and the popularity score of the preset professional factor is calculated according to the user behavior sequence. The popularity of the preset professional factor in the user group is analyzed from the popularity score, so as to accurately determine the important target professional factor; further, the policy element extraction model is used to quickly and accurately extract the policy elements related to the target professional factor, and the combination of the policy elements is restricted by using a tree structure, so as to finally obtain the candidate product strategy with high correlation to the target professional factor and having a preferred display effect, so as to realize the dynamic optimization selection of the policy element parameters; finally, the policy delivery model is used to select the target product strategy to be delivered. The policy delivery model is based on multiple user behavior link modeling, and can simultaneously consider multiple target behavior links in the target scenario, and then output accurate product strategy delivery decisions after training convergence. Based on the above product strategy delivery scheme, when the popular target professional factor determined according to the user behavior sequence is used as the core of the product strategy, the corresponding product strategy determined according to the target professional factor can explicitly display the inherent professional concept of the product to the user, so that the user can quickly obtain product information according to the product strategy, which can solve the user's urgent need for explainable product professional information.

Please refer to Figure 9, which is a structural block diagram of a product strategy delivery device provided by an embodiment of this specification. As shown in Figure 9, the product strategy delivery device 900 includes: a professional factor selection module 910, which is used to determine at least one target professional factor that meets the preset popular condition from at least one preset professional factor based on the user behavior sequence of at least one user for the historical product strategy, and the professional factor is a factor for professional interpretation of the product; a candidate strategy determination module 920, which is used to respectively determine at least one candidate product strategy corresponding to each target professional factor, and the candidate product strategy at least includes the target professional factor and the strategy element corresponding to the target professional factor; a product strategy delivery module 930, which is used to determine the candidate product strategy that meets the preset delivery condition from all candidate product strategies as the target product strategy to be delivered.

Optionally, the professional factor selection module 910 is also used to determine at least one target user behavior corresponding to at least one preset professional factor and the number of user actions for each target user behavior based on the user behavior sequence of at least one user for the historical product strategy; and determine, from at least one preset professional factor, a preset professional factor that meets a preset popular condition as a target professional factor according to the number of user actions corresponding to each preset professional factor.

Optionally, the professional factor selection module 910 is also used to calculate the popularity score of each preset professional factor according to the number of user actions corresponding to each preset professional factor and the popularity weight of each target user behavior, wherein the popularity weight is set according to the importance of the corresponding target user behavior; and determine the preset professional factor whose popularity score meets the preset popularity condition as the target professional factor.

Optionally, the candidate strategy determination module 920 is also used to input each target professional factor into a strategy element extraction model, and determine at least one strategy element corresponding to each target professional factor based on the strategy element extraction model; and determine at least one candidate product strategy corresponding to each target professional factor based on each target professional factor and the strategy elements corresponding to each target professional factor.

Optionally, the type of policy elements includes at least one of text, image, chart, and action point; the policy element extraction model is trained based on at least one sample policy element and standard professional factor labels corresponding to each sample policy element.

Optionally, the candidate strategy determination module 920 is also used to combine each target professional factor and the strategy elements corresponding to each target professional factor based on the element combination rule, and determine at least one candidate product strategy corresponding to each target professional factor that satisfies the element combination rule. The element combination rule is obtained by learning positive sample product strategy cases and negative sample product strategy cases.

Optionally, the candidate strategy determination module 920 is further used to perform Cartesian product assembly on each target professional factor and the strategy elements corresponding to each target professional factor according to the pre-selected strategy specifications.

Optionally, the candidate strategy determination module 920 is also used to input each candidate product strategy into a strategy delivery model, and determine, based on the strategy delivery model, a candidate product strategy that meets preset delivery conditions as the target product strategy to be delivered; wherein the strategy delivery model includes a prediction subnetwork of at least one target behavior link, the prediction subnetwork is used to predict the probability of a user group achieving a corresponding target behavior link under a given product strategy, and an embedded feature representation of sparse features shared by each prediction subnetwork.

Optionally, the strategy delivery model is trained based on a predicted value outputted for at least one sample product strategy and a predicted loss calculated from a standard value corresponding to each sample product strategy.

Optionally, the preset professional factors include but are not limited to product professional factors and user professional factors; wherein the product professional factors are professional factors of the product dimension defined based on basic professional knowledge in the product industry field, and the user professional factors are professional factors of the user dimension defined based on user behavior data.

In an embodiment of the present specification, a product strategy delivery device is provided, wherein a professional factor selection module is used to determine at least one target professional factor that meets a preset popular condition from at least one preset professional factor based on a user behavior sequence of at least one user for a historical product strategy, wherein the professional factor is a factor for professionally explaining a product; a candidate strategy determination module is used to respectively determine at least one candidate product strategy corresponding to each target professional factor, wherein the candidate product strategy includes at least a target professional factor and a strategy element corresponding to the target professional factor; and a product strategy delivery module is used to determine, from all candidate product strategies, a candidate product strategy that meets the preset delivery condition as the delivered target product strategy. Since professional factors can provide professional explanations for product strategies, when the popular target professional factor determined according to the user behavior sequence is used as the core of the product strategy, the corresponding product strategy determined according to the target professional factor can explicitly display the inherent professional concept of the product to the user, so that the user can quickly obtain product information according to the product strategy, which can solve the user's urgent need for explainable product professional information.

The embodiments of this specification provide a computer program product including instructions. When the computer program product is executed on a computer or a processor, the computer or the processor executes the steps of any one of the methods in the above embodiments.

The embodiments of this specification also provide a computer storage medium, which can store multiple instructions, and the instructions are suitable for being loaded by a processor and executing the steps of any method in the above embodiments.

Please refer to Figure 10, which is a schematic diagram of the structure of a terminal provided in an embodiment of this specification. As shown in Figure 10, the terminal 1000 may include: at least one terminal processor 1001, at least one network interface 1004, a user interface 1003, a memory 1005, and at least one communication bus 1002.

The communication bus 1002 is used to realize the connection and communication between these components.

The user interface 1003 may include a display screen (Display) and a camera (Camera), and the optional user interface 1003 may also include a standard wired interface and a wireless interface.

The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

The terminal processor 1001 may include one or more processing cores. The terminal processor 1001 uses various interfaces and lines to connect various parts of the entire terminal 1000, and executes various functions of the terminal 1000 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 1005, and calling data stored in the memory 1005. Optionally, the terminal processor 1001 can be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), and programmable logic array (PLA). The terminal processor 1001 can integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; and the modem is used to process wireless communication. It is understandable that the above-mentioned modem may not be integrated into the terminal processor 1001, but may be implemented by a separate chip.

The memory 1005 may include a random access memory (RAM), or The memory 1005 may include a read-only memory (ROM). Optionally, the memory 1005 includes a non-transitory computer-readable storage medium. The memory 1005 may be used to store instructions, programs, codes, code sets or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 1005 may also be at least one storage device located away from the aforementioned terminal processor 1001. As shown in Figure 10, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a product strategy delivery program.

In the terminal 1000 shown in FIG10 , the user interface 1003 is mainly used to provide an input interface for the user and obtain the data input by the user; and the terminal processor 1001 can be used to call the product strategy delivery program stored in the memory 1005, and specifically perform the following operations: based on the user behavior sequence of at least one user for the historical product strategy, determine at least one target professional factor that meets the preset popular condition from at least one preset professional factor, where the professional factor is a factor for professionally interpreting the product; respectively determine at least one candidate product strategy corresponding to each target professional factor, where the candidate product strategy includes at least the target professional factor and the strategy element corresponding to the target professional factor; from all candidate product strategies, determine the candidate product strategy that meets the preset delivery condition as the target product strategy to be delivered.

In some embodiments, when the terminal processor 1001 executes a user behavior sequence based on at least one user for a historical product strategy and determines at least one target professional factor that meets a preset popular condition from at least one preset professional factor, the terminal processor 1001 specifically performs the following steps: based on the user behavior sequence of at least one user for a historical product strategy, determine at least one target user behavior corresponding to at least one preset professional factor and the number of user actions for each target user behavior; and determine, from at least one preset professional factor, a preset professional factor that meets the preset popular condition as a target professional factor based on the number of user actions corresponding to each preset professional factor.

In some embodiments, when the terminal processor 1001 determines a preset professional factor that meets a preset popularity condition from at least one preset professional factor as a target professional factor according to the number of user actions corresponding to each preset professional factor, the terminal processor 1001 specifically performs the following steps: calculate the popularity score of each preset professional factor according to the number of user actions corresponding to each preset professional factor and the popularity weight of each target user behavior, wherein the popularity weight is set according to the importance of the corresponding target user behavior; determine the preset professional factor whose popularity score meets the preset popularity condition as the target professional factor.

In some embodiments, when the terminal processor 1001 is executing to determine at least one candidate product strategy corresponding to each target professional factor, it specifically performs the following steps: input each target professional factor into the policy element extraction model, and determine at least one policy element corresponding to each target professional factor based on the policy element extraction model; determine at least one candidate product strategy corresponding to each target professional factor based on each target professional factor and the policy elements corresponding to each target professional factor.

In some embodiments, the type of policy elements includes at least one of text, image, chart, and action point; the policy element extraction model is trained based on at least one sample policy element and the standard professional factor labels corresponding to each sample policy element.

In some embodiments, when the terminal processor 1001 determines at least one candidate product strategy corresponding to each target professional factor based on each target professional factor and the policy elements corresponding to each target professional factor, it specifically performs the following steps: based on the element combination rule, each target professional factor and the policy elements corresponding to each target professional factor are respectively combined to determine at least one candidate product strategy corresponding to each target professional factor that satisfies the element combination rule, and the element combination rule is obtained by learning positive sample product strategy cases and negative sample product strategy cases.

In some embodiments, when the terminal processor 1001 executes the steps of respectively combining each target professional factor and the policy elements corresponding to each target professional factor: Cartesian product assembly is performed on each target professional factor and the policy elements corresponding to each target professional factor according to the preselected policy specifications.

In some embodiments, when the terminal processor 1001 determines, from among all candidate product strategies, a candidate product strategy that meets preset delivery conditions as the target product strategy to be delivered, the following steps are specifically performed: each candidate product strategy is input into a strategy delivery model, and based on the strategy delivery model, a candidate product strategy that meets preset delivery conditions is determined as the target product strategy to be delivered; wherein the strategy delivery model includes a prediction subnetwork of at least one target behavior link, the prediction subnetwork is used to predict the probability of a user group realizing a corresponding target behavior link under a given product strategy, and an embedded feature representation of sparse features shared by each prediction subnetwork.

In some embodiments, the strategy delivery model is trained based on a predicted value outputted for at least one sample product strategy and a predicted loss calculated from a standard value corresponding to each sample product strategy.

In some embodiments, the preset professional factors include but are not limited to product professional factors and user professional factors; wherein the product professional factors are professional factors of the product dimension defined based on basic professional knowledge in the product industry field, and the user professional factors are professional factors of the user dimension defined based on user behavior data.

In the several embodiments provided in this specification, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules is only a logical function division, and there may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The above computer program product includes one or more computer instructions. When the above computer program instructions are loaded and executed on a computer, the above-mentioned processes or functions according to the embodiments of this specification are generated in whole or in part. The above computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The above computer instructions can be stored in a computer-readable storage medium or transmitted via the above computer-readable storage medium. The above computer instructions can be transmitted from a website site, computer, server or data center to another network site by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a digital versatile disc (DVD)), or a semiconductor medium (e.g., a solid state drive (SSD)).

It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

In addition, it should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, stored data, displayed data, etc.) and signals involved in the embodiments of this specification are all authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions. For example, the user behavior sequence and user action quantity involved in this specification are all obtained with full authorization.

In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

The above is a description of a product strategy delivery method, device, storage medium, and terminal provided in the embodiments of this specification. For technicians in this field, according to the ideas of the embodiments of this specification, there may be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the embodiments of this specification.

Claims (14)

A product strategy delivery method, the method comprising: Based on the user behavior sequence of at least one user for the historical product strategy, at least one target professional factor satisfying a preset popularity condition is determined from at least one preset professional factor, wherein the professional factor is a factor for professionally explaining the product; Determine at least one candidate product strategy corresponding to each target professional factor respectively, wherein the candidate product strategy includes at least the target professional factor and the strategy element corresponding to the target professional factor; From all candidate product strategies, determine the candidate product strategy that meets the preset launch conditions as the target product strategy to be launched. According to the method of claim 1, the step of determining at least one target professional factor satisfying a preset popularity condition from at least one preset professional factor based on the user behavior sequence of at least one user for the historical product strategy comprises: Based on a user behavior sequence of at least one user for a historical product strategy, determining at least one target user behavior corresponding to at least one preset professional factor and the number of user actions for each target user behavior; According to the number of user actions corresponding to each preset professional factor, a preset professional factor satisfying a preset popularity condition is determined from at least one preset professional factor as a target professional factor. According to the method of claim 2, the step of determining a preset professional factor that satisfies a preset popularity condition from at least one preset professional factor as a target professional factor according to the number of user actions corresponding to each preset professional factor, comprises: Calculate the popularity score of each preset professional factor according to the number of user actions corresponding to each preset professional factor and the popularity weight of each target user behavior, wherein the popularity weight is set according to the importance of the corresponding target user behavior; The preset professional factor whose popularity score meets the preset popularity condition is determined as the target professional factor. According to the method of claim 1, the step of respectively determining at least one candidate product strategy corresponding to each target professional factor comprises: Inputting each target professional factor into a policy element extraction model, and determining at least one policy element corresponding to each target professional factor based on the policy element extraction model; At least one candidate product strategy corresponding to each target professional factor is determined based on each target professional factor and the strategy elements corresponding to each target professional factor. The method according to claim 4, wherein the type of the policy element comprises at least one of text, image, chart, and action point; The policy element extraction model is trained based on at least one sample policy element and standard professional factor labels corresponding to each sample policy element. According to the method of claim 4, determining at least one candidate product strategy corresponding to each target professional factor based on each target professional factor and the strategy element corresponding to each target professional factor, comprises: Based on the element combination rules, each target professional factor and the strategy elements corresponding to each target professional factor are combined respectively. At least one candidate product strategy corresponding to each target professional factor and satisfying the element combination rule is determined, and the element combination rule is obtained by learning positive sample product strategy cases and negative sample product strategy cases. According to the method of claim 6, the step of respectively combining each target professional factor and the strategy element corresponding to each target professional factor comprises: According to the pre-selected strategy specifications, Cartesian product assembly is performed on each target professional factor and the strategy elements corresponding to each target professional factor. According to the method of claim 1, determining, from all candidate product strategies, a candidate product strategy that meets a preset launch condition as a target product strategy to be launched, comprises: Input each candidate product strategy into the strategy delivery model, and determine the candidate product strategy that meets the preset delivery conditions as the target product strategy to be delivered based on the strategy delivery model; Among them, the strategy delivery model includes at least one prediction subnetwork of a target behavior link, and the prediction subnetwork is used to predict the probability of a user group realizing a corresponding target behavior link under a given product strategy, as well as an embedded feature representation of sparse features shared by each prediction subnetwork. According to the method of claim 8, the strategy delivery model is trained based on the predicted value output for at least one sample product strategy and the predicted loss calculated based on the standard value corresponding to each sample product strategy. The method according to claim 1, wherein the preset professional factors include but are not limited to product professional factors and user professional factors; The product professional factor is a professional factor of the product dimension defined based on basic professional knowledge in the product industry field, and the user professional factor is a professional factor of the user dimension defined based on user behavior data. A product strategy delivery device, comprising: A professional factor selection module, for determining at least one target professional factor satisfying a preset popularity condition from at least one preset professional factor based on a user behavior sequence of at least one user for a historical product strategy, wherein the professional factor is a factor for professionally explaining a product; A candidate strategy determination module, used to determine at least one candidate product strategy corresponding to each target professional factor, wherein the candidate product strategy at least includes the target professional factor and a strategy element corresponding to the target professional factor; The product strategy delivery module is used to determine, from all candidate product strategies, a candidate product strategy that meets preset delivery conditions as the target product strategy to be delivered. A computer program product comprising instructions, when the computer program product is run on a computer or a processor, causes the computer or the processor to perform the steps of the method according to any one of claims 1 to 10. A computer storage medium storing a plurality of instructions, wherein the instructions are suitable for being loaded by a processor and executing the steps of the method according to any one of claims 1 to 10. A terminal comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 10 when executing the program.