A method and device for processing popularization information, a storage medium and an electronic device

Abstract

Embodiments of the present application disclose a promotion information processing method and device, a storage medium and an electronic device, wherein the method comprises: receiving a search request input by a user, obtaining a user search statement and determining a user search semantic vector corresponding to the user search statement, performing promotion information abstract matching processing on the user search semantic vector to obtain a promotion information semantic abstract set, obtaining a candidate promotion auction word set based on the promotion information semantic abstract set, determining a target promotion auction word based on the candidate promotion auction word set, and performing information promotion processing based on the target promotion auction word.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, storage medium and electronic device for promoting information processing. Background Technology

[0002] When users use applications with information search functions, they often receive promotional information pushed by the service provider during the search process. For example, to provide convenience to users, the promotional information may be an advertisement promoting a certain product or service. These promotional messages are significant for service providers, users, and the information promoters. Therefore, how to improve the delivery and recommendation effectiveness of promotional information in information search has always been a key focus of the industry. Summary of the Invention

[0003] This application provides a method, apparatus, storage medium, and electronic device for processing promotional information, which can improve the effectiveness of ad placement and recommendation. The technical solution is as follows:

[0004] In a first aspect, embodiments of this application provide a method for processing promotional information, the method comprising:

[0005] Receive a search request input by a user, obtain the user's search statement, and determine the user's search semantic vector corresponding to the user's search statement;

[0006] The user search semantic vector is subjected to promotion information summary matching processing to obtain a promotion information semantic summary set, and a candidate promotion auction term set is obtained based on the promotion information semantic summary set;

[0007] Based on the set of candidate promotional auction terms, target promotional auction terms are determined, and information promotion processing is performed based on the target promotional auction terms.

[0008] In one feasible implementation, the step of performing promotional information summary matching processing on the user search semantic vector to obtain a promotional information semantic summary set, and obtaining a candidate promotional auction term set based on the promotional information semantic summary set, includes:

[0009] In the promotion information semantic summary table, the user search semantic vector is subjected to promotion information summary matching processing to obtain a preset number of promotion information semantic summary sets;

[0010] Query the set of candidate promotional auction terms corresponding to the set of promotional information semantic summaries from the promotional information semantic summary table.

[0011] In one feasible implementation, the step of performing promotional information summary matching processing on the user search semantic vector in the promotional information semantic summary table to obtain a preset number of promotional information semantic summary sets includes:

[0012] In the promotion information semantic summary table, the vector similarity between the user search semantic vector and the reference promotion information semantic summary is calculated, and a promotion information semantic summary set including a preset number of reference promotion information semantic summaries is selected from the promotion information semantic summary table based on the vector similarity.

[0013] In one feasible implementation, the method further includes:

[0014] Obtain the historical search behavior sequence for the reference user, and determine the historical promotional information text of the long text type and the reference promotional auction terms corresponding to the historical promotional information text from the historical search behavior sequence;

[0015] Semantic summarization is performed on the historical promotional information text to obtain a reference promotional information semantic summary. Based on the reference promotional auction terms and the reference promotional information semantic summary, a promotional information semantic summary table is constructed for each reference promotional auction term.

[0016] In one feasible implementation, the step of extracting a semantic summary of the historical promotional information text to obtain a reference promotional information semantic summary includes:

[0017] Determine the summary task prompt words for the historical promotion information text, input the summary task prompt words and the historical promotion information text into the target large language model for semantic summary extraction, and output the reference promotion information semantic summary.

[0018] In one feasible implementation, the step of constructing a semantic summary table for each reference promotional auction term based on the reference promotional auction term and the semantic summary of the reference promotional information includes:

[0019] The historical promotional information text in each of the historical search behavior sequences is replaced with the semantic summary of the reference promotional information to generate a semantic summary table of promotional information for each of the reference promotional auction terms, wherein the historical search behavior sequences include the reference promotional auction terms corresponding to the historical promotional information text.

[0020] In one feasible implementation, determining the historical promotional information text of the long text type from the historical search behavior sequence includes:

[0021] Identify candidate historical promotional texts of long text type from historical search behavior sequences.

[0022] Based on all historical search behavior sequences, determine the number of times the candidate historical promotional information texts co-occur and the interval time corresponding to the historical promotional information texts;

[0023] The importance score of the candidate historical promotion information text is determined by a target scoring function based on the text co-occurrence frequency and the interval time.

[0024] The target scoring function satisfies the following formula:

[0025] f=*aΔ T

[0026] Wherein, f is the importance score, x is the number of times the text co-occurs, a is the attenuation coefficient, and ΔT is the interval time.

[0027] In one feasible implementation, the step of determining the target promotional auction term based on the candidate promotional auction term set includes:

[0028] Determine the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set, and select a target number of target promotion auction terms from the candidate promotion auction terms based on the promotion effect parameters.

[0029] In one feasible implementation, determining the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set includes:

[0030] Determine the bid coefficient, estimated click-through rate, and recall rate for each candidate promotional auction term in the candidate promotional auction term set;

[0031] The promotion effect parameters are obtained by multiplying the auction keyword bidding coefficient, the estimated click-through rate, and the recall rate.

[0032] Secondly, embodiments of this application provide a promotional information processing device, the device comprising:

[0033] The request processing module is used to receive the search request input by the user, obtain the user's search statement, and determine the user search semantic vector corresponding to the user's search statement.

[0034] The set determination module is used to perform promotion information summary matching processing on the user search semantic vector to obtain a promotion information semantic summary set, and to obtain a candidate promotion auction term set based on the promotion information semantic summary set;

[0035] The information promotion module is used to determine the target promotion auction keyword based on the candidate promotion auction keyword set, and to perform information promotion processing based on the target promotion auction keyword.

[0036] In one feasible implementation, the set determination module is configured to:

[0037] In the promotion information semantic summary table, the user search semantic vector is subjected to promotion information summary matching processing to obtain a preset number of promotion information semantic summary sets;

[0038] Query the set of candidate promotional auction terms corresponding to the set of promotional information semantic summaries from the promotional information semantic summary table.

[0039] In one feasible implementation, the set determination module is configured to:

[0040] In the promotion information semantic summary table, the vector similarity between the user search semantic vector and the reference promotion information semantic summary is calculated, and a promotion information semantic summary set including a preset number of reference promotion information semantic summaries is selected from the promotion information semantic summary table based on the vector similarity.

[0041] In one feasible implementation, the device is further used for:

[0042] Obtain the historical search behavior sequence for the reference user, and determine the historical promotional information text of the long text type and the reference promotional auction terms corresponding to the historical promotional information text from the historical search behavior sequence;

[0043] Semantic summarization is performed on the historical promotional information text to obtain a reference promotional information semantic summary. Based on the reference promotional auction terms and the reference promotional information semantic summary, a promotional information semantic summary table is constructed for each reference promotional auction term.

[0044] In one feasible implementation, the set determination module is configured to:

[0045] Determine the summary task prompt words for the historical promotion information text, input the summary task prompt words and the historical promotion information text into the target large language model for semantic summary extraction, and output the reference promotion information semantic summary.

[0046] In one feasible implementation, the set determination module is configured to:

[0047] The historical promotional information text in each of the historical search behavior sequences is replaced with the semantic summary of the reference promotional information to generate a semantic summary table of promotional information for each of the reference promotional auction terms, wherein the historical search behavior sequences include the reference promotional auction terms corresponding to the historical promotional information text.

[0048] In one feasible implementation, the set determination module is configured to:

[0049] Identify candidate historical promotional texts of long text type from historical search behavior sequences.

[0050] Based on all historical search behavior sequences, determine the number of times the candidate historical promotional information texts co-occur and the interval time corresponding to the historical promotional information texts;

[0051] The importance score of the candidate historical promotion information text is determined by a target scoring function based on the text co-occurrence frequency and the interval time.

[0052] The target scoring function satisfies the following formula:

[0053] f=*aΔ T

[0054] Wherein, f is the importance score, x is the number of times the text co-occurs, a is the attenuation coefficient, and ΔT is the interval time.

[0055] In one feasible implementation, the information promotion module is used for:

[0056] Determine the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set, and select a target number of target promotion auction terms from the candidate promotion auction terms based on the promotion effect parameters.

[0057] In one feasible implementation, the information promotion module is used for:

[0058] Determine the bid coefficient, estimated click-through rate, and recall rate for each candidate promotional auction term in the candidate promotional auction term set;

[0059] The promotion effect parameters are obtained by multiplying the auction keyword bidding coefficient, the estimated click-through rate, and the recall rate.

[0060] Thirdly, embodiments of this application provide a computer storage medium storing a plurality of instructions adapted for loading by a processor and executing the above-described method steps.

[0061] Fourthly, embodiments of this application provide an electronic device that may include: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to execute the above-described method steps.

[0062] The beneficial effects of the technical solutions provided in some embodiments of this application include at least the following:

[0063] In one or more embodiments of this application, by receiving a user's search request, obtaining the user's search statement and determining the user's search semantic vector corresponding to the user's search statement, performing promotional information summary matching processing on the user's search semantic vector to obtain a promotional information semantic summary set, and then obtaining a candidate promotional auction term set based on the promotional information semantic summary set. Matching based on promotional information summaries rather than long text-type promotional information can recall a candidate promotional auction term set with high accuracy. Subsequently, based on the candidate promotional auction term set, target promotional auction terms are determined, and information promotion processing is performed based on the target promotional auction terms. During information promotion processing, the promotional information recalled based on the target promotional auction terms has high relevance. It can retain rich semantic features of the promotional information by leveraging the promotional information summary, thereby improving the delivery and recommendation effect of promotional information in information search scenarios, improving the relevance of promotional information recommendations, and improving the user's promotional information experience. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 This is a schematic diagram of a promotional information processing system provided in an embodiment of this application;

[0066] Figure 2 This is a flowchart illustrating a promotional information processing method provided in an embodiment of this application;

[0067] Figure 3 This is a schematic diagram of a user search scenario provided in an embodiment of this application;

[0068] Figure 4 This is a schematic diagram of a scenario following the recall of promotional information, provided in an embodiment of this application.

[0069] Figure 5 This is a flowchart illustrating a message digest matching method provided in an embodiment of this application;

[0070] Figure 6 This is a schematic diagram illustrating the process of constructing a semantic summary table for promotional information provided in an embodiment of this application;

[0071] Figure 7 This is a schematic diagram of the structure of a promotional information processing device provided in an embodiment of this application;

[0072] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application;

[0073] Figure 9 This is a schematic diagram of the structure of the operating system and user space provided in the embodiments of this application;

[0074] Figure 10 yes Figure 9 Architecture diagram of the Android operating system in China;

[0075] Figure 11 yes Figure 9 Architecture diagram of the iOS operating system. Detailed Implementation

[0076] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0077] In the description of this application, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. In the description of this application, it should be noted that, unless otherwise expressly specified and limited, "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances. Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist; for example, A and / or B can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship.

[0078] In related technologies, promotional information is advertised through a platform with a planned campaign. The platform then uses algorithms to distribute promotional traffic to the target audience. High-quality promotional information, once clicked by users, generates certain promotional performance parameters (such as promotional rewards and revenue) on the platform. In this process, the goal of optimizing the CPC (Cost Per Click) algorithm on the promotional platform is to maximize these performance parameters (such as expected revenue per thousand impressions). The key factors influencing these performance parameters include the promotional information's CTR (Cost Per Transaction), bid-price (bid-per-unit), and PVR (Promotional Viewership).

[0079] Related technologies involve multiple stages, including recall, coarse ranking, and fine ranking. The recall and fine ranking CTR prediction stages rely on textual semantic information such as user search keywords, natural search result summaries and context, promotional bidwords, promotional titles, and promotional descriptions as features. However, features such as summaries, promotional titles, and promotional descriptions are long texts at the sentence level. Directly modeling the original text to extract semantic features is ineffective, resulting in poor relevance of recalled promotional information or even failure to recall promotional information. This leads to the loss of rich semantic features from the promotional information side, indicating poor recommendation performance. Therefore, there is an urgent need to improve the recommendation performance of promotional information in information search scenarios.

[0080] The present application will now be described in detail with reference to specific embodiments.

[0081] Please see Figure 1 This is a schematic diagram of a promotional information processing system provided in this specification. Figure 1 As shown, the promotion information processing system may include at least a client cluster and a service platform 100.

[0082] The client cluster may include at least one client, such as Figure 1 As shown, it specifically includes client 1 corresponding to user 1, client 2 corresponding to user 2, ..., client n corresponding to user n, where n is an integer greater than 0.

[0083] Each client in a client cluster can be an electronic device with communication capabilities, including but not limited to: wearable devices, handheld devices, personal computers, tablets, in-vehicle devices, smartphones, computing devices, or other processing devices connected to a wireless modem. Electronic devices may have different names in different networks, such as: user equipment, access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication equipment, user agent or user device, cellular phone, cordless phone, personal digital assistant (PDA), and electronic devices in 5G networks or future evolved networks.

[0084] The service platform 100 can be a standalone server device, such as a rack-mount, blade, tower, or cabinet-type server device, or a workstation, mainframe, or other hardware device with strong computing power; or it can be a server cluster composed of multiple servers. The servers in the service cluster can be composed in a symmetrical manner, wherein each server is functionally and hierarchically equivalent in the transaction chain, and each server can provide services independently. The independent provision of services can be understood as not requiring the assistance of other servers.

[0085] In one or more embodiments of this specification, the service platform 100 may establish a communication connection with at least one client in the client cluster, and complete the data interaction during the feature processing process, such as online transaction data interaction, based on the communication connection.

[0086] It should be noted that the service platform 100 establishes a communication connection with at least one client in the client cluster via a network for interactive communication. This network can be a wireless network or a wired network. Wireless networks include, but are not limited to, cellular networks, wireless LANs, infrared networks, or Bluetooth networks. Wired networks include, but are not limited to, Ethernet, universal serial bus (USB), or controller area networks. In one or more embodiments of the specification, technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc., are used to represent data exchanged over the network (such as target compressed packets). Furthermore, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), and Internet Protocol Security (IPsec) can be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can be used to replace or supplement the aforementioned data communication technologies.

[0087] The feature processing system embodiments provided in this specification and the promotional information processing methods described in one or more embodiments belong to the same concept. The execution entity corresponding to the promotional information processing method in one or more embodiments of this specification can be the aforementioned service platform 100; the execution entity corresponding to the promotional information processing method in one or more embodiments of this specification can also be the electronic device corresponding to the client, depending on the actual application environment. The implementation process of the feature processing system embodiments can be detailed in the following method embodiments, and will not be repeated here.

[0088] based on Figure 1 The schematic diagram shown below illustrates the feature processing methods provided by one or more embodiments of this specification.

[0089] In one embodiment, such as Figure 2 As shown, a promotional information processing method is proposed. This method can be implemented using a computer program and can run on a promotional information processing device based on the von Neumann architecture. The computer program can be integrated into an application or run as a standalone tool application. The promotional information processing device can be a service platform.

[0090] Specifically, the promotional information processing method includes:

[0091] S102: Receive the search request input by the user, obtain the user's search statement and determine the user search semantic vector corresponding to the user's search statement;

[0092] When a user enters a search query into a search engine, the service platform can receive the search request from the user. The search request carries the search query, and the platform can further extract search query features such as keywords and phrases from the user's search query and convert them into a user search semantic vector.

[0093] For example, the user's search query is input into a semantic processing model pre-trained based on a machine learning model to extract semantic vectors and output the user's search semantic vector. For example, the semantic processing model can be the Query-Title semantic BERT model.

[0094] S104: Perform promotion information summary matching processing on the user search semantic vector to obtain a promotion information semantic summary set, and obtain a candidate promotion auction term set based on the promotion information semantic summary set;

[0095] A corresponding promotional information summary is pre-set for each candidate "promotional information summary". The candidate "promotional information summary" can be a pre-stored short description or keyword list about the promotional information content (such as advertising content). Each candidate promotional information that can be recommended can be pre-set with a corresponding promotional information summary. The user's search semantic vector is matched with these candidate "promotional information summaries" to find a set of promotional information semantic summaries consisting of one or more promotional information semantic summaries that are most relevant to the user's search intent.

[0096] Furthermore, the original promotional information corresponding to each promotional information semantic summary has a pre-existing associated promotional auction term (bidword). The extraction of candidate promotional information semantic summaries and the determination of candidate promotional auction terms (bidword) can be pre-saved. Then, a promotional mapping relationship is established between candidate promotional information, candidate promotional information semantic summaries, and candidate promotional auction terms (bidword). Based on this, in practical applications, after determining the set of promotional information semantic summaries, the candidate promotional auction terms corresponding to each promotional information semantic summary in the promotional information semantic summary set can be queried based on the promotional mapping relationship. The candidate promotional auction terms corresponding to each promotional information semantic summary form a set of candidate promotional auction terms.

[0097] S106: Determine the target promotion auction term based on the candidate promotion auction term set, and perform information promotion processing based on the target promotion auction term.

[0098] Targeted promotional auction keywords are a reference number of recommended promotional auction keywords determined from the set of candidate promotional auction keywords. Using targeted promotional auction keywords to recall and deliver relevant promotional information can improve the delivery and recommendation effect of promotional information in information search scenarios.

[0099] During this query process, the system extracts relevant candidate bidwords for each entry in the promotional information summary set. The set of these candidate bidwords forms the candidate promotional auction term set, namely B = {b1, b2, ..., b...}. k This provides a pool of candidate keywords for subsequent promotional campaigns.

[0100] In one feasible implementation, a coarse-ranking model from related technologies can be used to initially rank the candidate promotion auction keyword set. This typically considers factors such as semantic relevance, historical advertising bids, and recall weights. After coarse-ranking processing, the coarse-ranking model obtains B. * .

[0101] B *The data is input into the click-through rate (CTR) prediction model, which then performs a CTR prediction for each bidword in the candidate set. During the training process of the CTR prediction model, promotional information text (such as ad titles and ad summaries) and semantic feature vector E extracted by the target large language model can also be used to help improve the accuracy of the CTR prediction model.

[0102] Optionally, you can continue to filter target promotion auction keywords from the perspective of historical promotion performance. Select reference indicators for the historical promotion performance dimension, such as click-through rate, conversion rate and cost-effectiveness indicators. You can obtain the reference indicator value corresponding to each candidate promotion auction keyword, and select a reference number of target promotion auction keywords from the candidate promotion auction keyword set based on the magnitude of the reference indicator value.

[0103] For example, determining the target promotional auction term based on the candidate promotional auction term set includes:

[0104] Determine the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set, and select a target number of target promotion auction terms from the candidate promotion auction terms based on the promotion effect parameters.

[0105] For example, using promotional performance parameters such as eCPM (effective cost per thousand impressions) to effectively select candidate promotional auction keywords can help choose the most cost-effective promotional auction keywords from a set of candidate keywords.

[0106] Optionally, you can use eCPM = (Impressions / Total Revenue) × 1000 to calculate the promotional performance parameter. In the formula, total revenue can be the estimated revenue for a specific promotional auction keyword, which can be determined based on historical data or market averages. Impressions are the number of times the ad is shown. Thus, eCPM provides the average revenue generated per thousand impressions.

[0107] Furthermore, through creative work, the calculation of promotion effect parameters has been improved. It is now comprehensively calculated from the auction keyword bid coefficient, the estimated click-through rate, and the recall rate. The method for determining the promotion effect parameters for each candidate auction keyword in the candidate promotion auction keyword set can be as follows:

[0108] The bid coefficient, estimated click-through rate (PCTR), and recall rate (PVR) of each candidate promotional auction term in the candidate promotional auction term set are determined. The promotional effect parameters are obtained based on the product of the bid coefficient, the estimated click-through rate, and the recall rate.

[0109] In other words: Promotion effect parameter = bid * pctr * pvr

[0110] For example, the promotion effect parameter of each candidate promotion auction term in the candidate promotion auction term set is determined, the candidate promotion auction terms are sorted based on the size of the promotion effect parameter, the target number of target promotion auction terms are selected from the candidate promotion auction terms based on the promotion effect parameter, and then the target promotion auction terms are pushed to the promotion information engine to recall the corresponding target promotion information and expose the target promotion information to the user.

[0111] The target quantity is a pre-set empirical value, which can be set based on the actual application situation.

[0112] Indicative: Please see Figure 3 , Figure 3 This is a scenario illustration of a user search. A user enters "children's drawing pictures" into the image search box. Originally, the ad engine lacked suitable promotional information to retrieve and therefore could not display relevant promotional content to the user (e.g., ...). Figure 3 After executing one or more embodiments of the promotional information processing method described in this specification, the user inputs the same search request, obtains the user search statement "a complete collection of children's drawing pictures" and determines the user search semantic vector corresponding to the user search statement, performs promotional information summary matching processing on the user search semantic vector to obtain a promotional information semantic summary set, obtains a candidate promotional auction term set based on the promotional information semantic summary set, determines the target promotional auction term based on the candidate promotional auction term set, and recalls highly relevant promotional information based on the target promotional auction term. Please refer to [link to relevant documentation]. Figure 4 , Figure 4 This is a scenario illustration after a recall of promotional information. Figure 4 In the process, two highly relevant promotional messages were recalled based on the target promotional keywords. Figure 4 As shown in the red box, the effectiveness of promotional information delivery has been significantly improved for the platform.

[0113] In one or more embodiments of this application, by receiving a user's search request, obtaining the user's search statement and determining the user's search semantic vector corresponding to the user's search statement, performing promotional information summary matching processing on the user's search semantic vector to obtain a promotional information semantic summary set, and then obtaining a candidate promotional auction term set based on the promotional information semantic summary set. Matching based on promotional information summaries rather than long text-type promotional information can recall a candidate promotional auction term set with high accuracy. Subsequently, based on the candidate promotional auction term set, target promotional auction terms are determined, and information promotion processing is performed based on the target promotional auction terms. During information promotion processing, the promotional information recalled based on the target promotional auction terms has high relevance. It can retain rich semantic features of the promotional information by leveraging the promotional information summary, thereby improving the delivery and recommendation effect of promotional information in information search scenarios, improving the relevance of promotional information recommendations, and improving the user's promotional information experience.

[0114] Please see Figure 5 , Figure 5 This is a flowchart illustrating an information summary matching method proposed in this application. Specifically, the method involves performing promotional information summary matching on the user search semantic vector to obtain a set of promotional information semantic summaries, and then obtaining a set of candidate promotional auction terms based on this set of promotional information semantic summaries. This can be achieved as follows:

[0115] S202: Perform promotion information summary matching processing on the user search semantic vector in the promotion information semantic summary table to obtain a preset number of promotion information semantic summary sets;

[0116] For example, a promotional information summary is pre-determined from the existing promotional information, thereby constructing a promotional information semantic summary table containing semantic summaries (i.e., promotional information extracts) of various promotional information (such as advertising content). Each promotional information extract is converted into a semantic vector form, which facilitates semantic-level vector matching.

[0117] In some embodiments, the original promotional information corresponding to each promotional information semantic summary has a pre-existing associated promotional auction term bidword. The semantic summary of candidate promotional information and the determination of candidate (or reference) promotional auction term bidword can be pre-saved. Then, a promotional information mapping relationship is established between candidate (or reference) promotional information, candidate (or reference) promotional information semantic summary and candidate (or reference) promotional auction term bidword, forming a promotional information semantic summary table including the promotional information mapping relationship.

[0118] The promotional message summary matching process typically involves calculating the similarity between the user's search semantic vector and each summary vector in the summary table. This can be achieved using methods such as cosine similarity and Euclidean distance. Then, based on the similarity ranking, the most relevant set of promotional message summaries is selected. The number of summaries can be preset, for example, selecting the top 10 most similar promotional message summaries, denoted as {S}. *} k ...

[0119] For example, the step of performing promotional information summary matching processing on the user search semantic vector in the promotional information semantic summary table to obtain a preset number of promotional information semantic summary sets includes:

[0120] A2: Based on the reference promotional information semantic summary in the promotional information semantic summary table, calculate the vector similarity between the user search semantic vector and the reference promotional information semantic summary;

[0121] A4: Based on the vector similarity, select a set of promotional information semantic summaries from the promotional information semantic summary table, which includes a preset number of reference promotional information semantic summaries.

[0122] In a schematic way, the promotion information summary matching process for the user search semantic vector in the promotion information semantic summary table can be to calculate the vector similarity between the user search semantic vector and the promotion information summary in each table (reference). Based on the vector similarity, a preset number of reference promotion information semantic summaries are taken. The preset number of reference promotion information semantic summaries form a set, that is, the promotion information semantic summary set.

[0123] S204: Query the set of candidate promotional auction terms corresponding to the set of promotional information semantic summaries from the promotional information semantic summary table.

[0124] After selecting the set of semantic summaries for promotional messages, the next step is to query the promotional message semantic summaries table for the corresponding candidate promotional auction keywords. Each promotional message summary is associated with a specific candidate promotional auction keyword. These candidate promotional auction keywords are predefined for specific advertising content and can represent the core content or target audience of the promotional message.

[0125] During this query process, the system extracts relevant candidate bidwords for each entry in the promotional information summary set. The set of these candidate bidwords forms the candidate promotional auction term set, namely B = {b1, b2, ..., b...}. k This provides a pool of candidate keywords for subsequent promotional campaigns.

[0126] In one or more embodiments of this application, the device can efficiently connect users' actual search intent with the most relevant promotional information content, while providing scientific, data-driven key promotional auction keywords for promotional information delivery. This not only improves the targeting and effectiveness of promotional information but also optimizes the cost-effectiveness of promotional information delivery, ultimately helping to improve the conversion rate and user satisfaction of promotional information. Furthermore, the automation and intelligence level of the entire process are also improved, enhancing the responsiveness and market adaptability of the promotional information system.

[0127] Please see Figure 6 , Figure 6 This is a flowchart illustrating the construction process of a semantic summary table for promotional information proposed in this application. Specifically:

[0128] S3002: Obtain the historical search behavior sequence for the reference user, and determine the historical promotional information text of the long text type and the reference promotional auction term corresponding to the historical promotional information text from the historical search behavior sequence;

[0129] Based on past information of users on the platform such as user searches, advertisement requests, and click history, determine the historical search behavior sequence for a reference user. The reference user can be a user who uses the search function of the platform. During the process of the reference user using the search function of the platform, the platform can record the behavior data of the reference user. Based on the behavior data, the historical search behavior sequence of the reference user can be determined. The historical search behavior sequence is to extract data of a specified search text type from the behavior data according to one or more specified search text types. After all the data of the specified search text types are arranged and combined, they form the historical search behavior session sequence of the reference user;

[0130] The specified search text types include but are not limited to search statement query types, reference promotion auction word types during the search process, promotion information title types, promotion information description types, retrieval result summary types, etc.;

[0131] The long text type is the long text type set from the specified search text types. For example, setting the promotion information title type and the promotion information description type as long text types, it is found through creative labor that directly extracting semantic features from the historical promotion information text of the long text type has poor subsequent semantic feature effects due to the long sentence granularity, resulting in poor relevance of the recalled promotion information;

[0132] Exemplarily, determine the historical promotion information text of the long text type and the reference promotion auction word corresponding to the historical promotion information text from the historical search behavior sequence;

[0133] For example, assume the historical search behavior sequence is expressed as: <a search query, b advertisement bidword (auction word), c advertisement title, d advertisement description> session sequence, the historical promotion information text is "c advertisement title, d advertisement description", and the reference promotion auction word corresponding to the historical promotion information text is "b advertisement bidword (auction word)";

[0134] In an optional implementation manner, text string aggregation filtering can also be performed on the promotion information text. Exemplarily, the determining of the historical promotion information text of the long text type from the historical search behavior sequence includes:

[0135] A2: Determine the candidate historical promotion information text of the long text type from the historical search behavior sequence;

[0136] The candidate historical promotion information text can be, for example, the promotion information description and the promotion information title;

[0137] A4: Based on all historical search behavior sequences, determine the text co-occurrence times of the candidate historical promotion information text and the interval time corresponding to the historical promotion information text;

[0138] A6: Based on the number of text co-occurrences and the interval time, a target scoring function is used to determine the importance score of the candidate historical promotion information text, and the historical promotion information text is determined from the candidate historical promotion information text based on the importance score;

[0139] The target scoring function satisfies the following formula:

[0140] f=*aΔ T

[0141] Wherein, f is the importance score, x is the number of times the text co-occurs, a is the attenuation coefficient, and ΔT is the interval time.

[0142] For example, 'a' is the attenuation coefficient, and its value ranges from 0.99 to 1.

[0143] Furthermore, a score threshold is set. If the important score is greater than the score threshold, the candidate historical promotion information text is used as the historical promotion information text. If the important score is less than or equal to the score threshold, the candidate historical promotion information text is filtered out.

[0144] Furthermore, all candidate historical promotional information texts are sorted according to their importance scores, and a certain number of the top-ranked historical promotional information texts are selected.

[0145] S3004: Extract semantic summaries from the historical promotional information text to obtain reference promotional information semantic summaries, and construct a promotional information semantic summary table for each reference promotional auction term based on the reference promotional auction terms and the reference promotional information semantic summaries.

[0146] In one feasible implementation, the historical promotional information text is first preprocessed, including noise removal (such as HTML tags, special symbols, etc.), text normalization (such as converting to lowercase, removing stop words, and lemmatization), and word segmentation (such as breaking the text down into words or phrases). Then, a natural language processing model, such as BERT, GPT, or other deep learning models, is used to convert the preprocessed text into semantic vectors. Natural language processing models can capture deep semantic relationships in the text and convert them into numerical representations in a high-dimensional space. Finally, a semantic summary is extracted from the semantic vectors, including key information extraction and summary generation, thus obtaining a semantic summary of the reference promotional information.

[0147] Key information extraction: Using tools to extract key information from semantic vectors, which can include techniques such as topic recognition and keyword extraction. Tools such as TF-IDF, LDA (Latent Dirichlet Allocation), or neural network models can be used to identify key elements in text.

[0148] Summary generation: Generate a semantic summary based on the extracted key information;

[0149] The generated semantic summaries are stored in a database or indexing system. Each reference promotion information semantic summary is associated with its corresponding reference promotion information semantic summary and reference promotion auction terms to generate a promotion information semantic summary table. With the help of the promotion information semantic summary table, users' search requests can be quickly retrieved and matched when needed.

[0150] In one feasible implementation, the step of extracting a semantic summary of the historical promotional information text to obtain a reference promotional information semantic summary can be performed in the following manner:

[0151] Determine the summary task prompt words for the historical promotion information text, input the summary task prompt words and the historical promotion information text into the target large language model for semantic summary extraction, and output the reference promotion information semantic summary.

[0152] Specifically, the summary task prompts are used to instruct the target large language model to extract semantic summaries based on the model input and output semantic summaries of reference generalization information; the summary task prompts can use preset general summary task prompt templates.

[0153] Optional: The summary task prompt could be something like, "You are an advertising copy summary assistant. Based on the given ad title and ad description, generate a short summary sentence. The output character length should be close to the original ad title." A specific example is shown below:

[0154] 1. Model Input: Ad Title: ###Postgraduate Entrance Exam Course, xxx Postgraduate Entrance Exam Training Camp, 23 / 24 Postgraduate Entrance Exam Class Recruitment###, Ad Description: ###Postgraduate Entrance Exam Course, Postgraduate Entrance Exam Tutoring Must-See xxx Postgraduate Entrance Exam Training Camp, 35-person Small Class Face-to-Face Teaching, Scientific Score Improvement, Helping You Achieve a Sprint to 985 / 211 Universities. Model Output: xxx Postgraduate Entrance Exam Training Camp Recruitment, Small Class Face-to-Face Teaching, Scientific Score Improvement, Helping You Achieve a Sprint to 985 / 211 Universities;

[0155] 2. Model Input: Input: Ad Title: ### Fiberglass Transport Tank <xxx>30 years of dedicated focus on FRP transport tank manufacturing, a trusted established manufacturer. Advertisement description: A professional manufacturer of FRP industrial products, our products sell well in over 30 provinces nationwide. We specialize in wound FRP storage tanks, which are sturdy, durable, reasonably priced, acid-resistant, alkali-resistant, corrosion-resistant, and have a long service life. Model output: xxx FRP transport tank, a professional manufacturer, sturdy, durable, long service life, products sold nationwide.

[0156] Furthermore, for the set of historical promotional information texts, by constructing the above-mentioned summary task prompt word "prompt", a batch request is made to the target large language model (such as: gpt-3.5-turbo or 360gpt-pro) to obtain a new semantic summary S* of the reference promotional information.

[0157] For each reference promotional information semantic summary S*, its reference promotional auction term bidword can be determined. At least, a promotional information semantic summary table can be constructed for each reference promotional auction term based on the reference promotional auction term and the reference promotional information semantic summary. Each reference promotional auction term bidword can be used to determine candidate (or reference) promotional information within the system platform. That is, a promotional information semantic summary table including the promotional information mapping relationship can also be formed based on establishing a promotional information mapping relationship between candidate (or reference) promotional information, candidate (or reference) promotional information semantic summaries, and candidate (or reference) promotional auction term bidwords.

[0158] In one or more embodiments of this application, the semantic summary of historical promotional information can not only improve the relevance and effectiveness of promotional information, but also help the maintenance side of promotional information to understand market dynamics and user needs more deeply, thereby making more accurate promotional decisions.

[0159] The following will combine Figure 7 This application provides a detailed description of the promotional information processing device provided in its embodiments. It should be noted that... Figure 7 The promotional information processing device shown is used to execute this application. Figures 1-6 The methods shown in the embodiments are for illustrative purposes only, illustrating the parts relevant to the embodiments of this application. For specific technical details not disclosed, please refer to this application. Figures 1-6 The example shown.

[0160] Please see Figure 7 This diagram illustrates the structure of a promotional information processing device according to an embodiment of this application. The promotional information processing device 1 can be implemented as all or part of a user terminal through software, hardware, or a combination of both. According to some embodiments, the promotional information processing device 1 includes a promotional information processing module 11, a promotional information processing module 12, and a promotional information processing module 13, specifically used for:

[0161] The request processing module 11 is used to receive the search request input by the user, obtain the user's search statement, and determine the user search semantic vector corresponding to the user's search statement;

[0162] The set determination module 12 is used to perform promotion information summary matching processing on the user search semantic vector to obtain a promotion information semantic summary set, and to obtain a candidate promotion auction term set based on the promotion information semantic summary set;

[0163] The information promotion module 13 is used to determine the target promotion auction term based on the candidate promotion auction term set, and to perform information promotion processing based on the target promotion auction term.

[0164] In one feasible implementation, the set determination module 12 is configured to:

[0165] In the promotion information semantic summary table, the user search semantic vector is subjected to promotion information summary matching processing to obtain a preset number of promotion information semantic summary sets;

[0166] Query the set of candidate promotional auction terms corresponding to the set of promotional information semantic summaries from the promotional information semantic summary table.

[0167] In one feasible implementation, the set determination module 12 is configured to:

[0168] In the promotion information semantic summary table, the vector similarity between the user search semantic vector and the reference promotion information semantic summary is calculated, and a promotion information semantic summary set including a preset number of reference promotion information semantic summaries is selected from the promotion information semantic summary table based on the vector similarity.

[0169] In one feasible implementation, the device 1 is further used for:

[0170] Obtain the historical search behavior sequence for the reference user, and determine the historical promotional information text of the long text type and the reference promotional auction terms corresponding to the historical promotional information text from the historical search behavior sequence;

[0171] Semantic summarization is performed on the historical promotional information text to obtain a reference promotional information semantic summary. Based on the reference promotional auction terms and the reference promotional information semantic summary, a promotional information semantic summary table is constructed for each reference promotional auction term.

[0172] In one feasible implementation, the set determination module 12 is configured to:

[0173] Determine the summary task prompt words for the historical promotion information text, input the summary task prompt words and the historical promotion information text into the target large language model for semantic summary extraction, and output the reference promotion information semantic summary.

[0174] In one feasible implementation, the set determination module 12 is configured to:

[0175] The historical promotional information text in each of the historical search behavior sequences is replaced with the semantic summary of the reference promotional information to generate a semantic summary table of promotional information for each of the reference promotional auction terms, wherein the historical search behavior sequences include the reference promotional auction terms corresponding to the historical promotional information text.

[0176] In one feasible implementation, the set determination module 12 is configured to:

[0177] Identify candidate historical promotional texts of long text type from historical search behavior sequences.

[0178] Based on all historical search behavior sequences, determine the number of times the candidate historical promotional information texts co-occur and the interval time corresponding to the historical promotional information texts;

[0179] The importance score of the candidate historical promotion information text is determined by a target scoring function based on the text co-occurrence frequency and the interval time.

[0180] The target scoring function satisfies the following formula:

[0181] f=*aΔ T

[0182] Wherein, f is the importance score, x is the number of times the text co-occurs, a is the attenuation coefficient, and ΔT is the interval time.

[0183] In one feasible implementation, the information promotion module 13 is used for:

[0184] Determine the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set, and select a target number of target promotion auction terms from the candidate promotion auction terms based on the promotion effect parameters.

[0185] In one feasible implementation, the information promotion module 13 is used for:

[0186] Determine the bid coefficient, estimated click-through rate, and recall rate for each candidate promotional auction term in the candidate promotional auction term set;

[0187] The promotion effect parameters are obtained by multiplying the auction keyword bidding coefficient, the estimated click-through rate, and the recall rate.

[0188] It should be noted that the promotional information processing device provided in the above embodiments is only illustrated by the division of the above functional modules when executing the promotional information processing method. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the promotional information processing device and the promotional information processing method embodiments provided in the above embodiments belong to the same concept, and the implementation process is detailed in the method embodiments, which will not be repeated here.

[0189] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0190] In one or more embodiments of this application, by receiving a user's search request, obtaining the user's search statement and determining the user's search semantic vector corresponding to the user's search statement, performing promotional information summary matching processing on the user's search semantic vector to obtain a promotional information semantic summary set, and then obtaining a candidate promotional auction term set based on the promotional information semantic summary set. Matching based on promotional information summaries rather than long text-type promotional information can recall a candidate promotional auction term set with high accuracy. Subsequently, based on the candidate promotional auction term set, target promotional auction terms are determined, and information promotion processing is performed based on the target promotional auction terms. During information promotion processing, the promotional information recalled based on the target promotional auction terms has high relevance. It can retain rich semantic features of the promotional information by leveraging the promotional information summary, thereby improving the delivery and recommendation effect of promotional information in information search scenarios, improving the relevance of promotional information recommendations, and improving the user's promotional information experience.

[0191] This application also provides a computer storage medium that can store multiple instructions, which are adapted to be loaded and executed by a processor as described above. Figures 1-6 The promotional information processing method described in the illustrated embodiment can be found in the following documentation for its specific execution process. Figures 1-6 The specific details of the illustrated embodiments will not be elaborated here.

[0192] This application also provides a computer program product storing at least one instruction, which is loaded and executed by the processor as described above. Figures 1-6 The promotional information processing method described in the illustrated embodiment can be found in the following documentation for its specific execution process. Figures 1-6 The specific details of the illustrated embodiments will not be elaborated here.

[0193] Please refer to Figure 8 This diagram illustrates a structural block diagram of an electronic device provided in an exemplary embodiment of this application. The electronic device in this application may include one or more components such as a processor 110, a memory 120, an input device 130, an output device 140, and a bus 150. The processor 110, memory 120, input device 130, and output device 140 may be connected via the bus 150.

[0194] Processor 110 may include one or more processing cores. Processor 110 connects to various parts of the electronic device via various interfaces and lines, and performs various functions and processes data of electronic device 100 by running or executing instructions, programs, code sets, or instruction sets stored in memory 120, and by calling data stored in memory 120. Optionally, processor 110 may be implemented using at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). Processor 110 may integrate one or more of the following: central processing unit (CPU), graphics processing unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the displayed content; and the modem handles wireless communication. It is understood that the modem may also not be integrated into processor 110 and may be implemented separately using a communication chip.

[0195] The memory 120 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 120 may include a non-transitory computer-readable storage medium. The memory 120 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 120 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as touch functionality, sound playback functionality, image playback functionality, etc.), instructions for implementing the various method embodiments described below, etc. The operating system may be the Android system, including systems deeply developed based on the Android system, the iOS system developed by Apple Inc., including systems deeply developed based on the iOS system, or other systems. The data storage area may also store data created by the electronic device during use, such as phonebook data, audio and video data, chat log data, etc.

[0196] See Figure 9 As shown, the memory 120 can be divided into operating system space and user space. The operating system runs in the operating system space, while native and third-party applications run in the user space. To ensure that different third-party applications can achieve good running performance, the operating system allocates corresponding system resources for each application. However, different application scenarios within the same third-party application have different requirements for system resources. For example, in local resource loading scenarios, third-party applications have high requirements for disk read speed; in animation rendering scenarios, third-party applications have high requirements for GPU performance. Since the operating system and third-party applications are independent of each other, the operating system often cannot promptly perceive the current application scenario of a third-party application, resulting in the operating system's inability to adapt system resources accordingly to the specific application scenario of the third-party application.

[0197] In order for the operating system to distinguish the specific application scenarios of third-party applications, it is necessary to establish data communication between the third-party applications and the operating system. This would allow the operating system to obtain the current scenario information of the third-party applications at any time, and then perform targeted system resource adaptation based on the current scenario.

[0198] Taking the Android operating system as an example, the programs and data stored in memory 120 are as follows: Figure 10 As shown, the memory 120 can store the Linux kernel layer 320, the system runtime library layer 340, the application framework layer 360, and the application layer 380. The Linux kernel layer 320, system runtime library layer 340, and application framework layer 360 belong to the operating system space, while the application layer 380 belongs to the user space. The Linux kernel layer 320 provides low-level drivers for various hardware components of the electronic device, such as display drivers, audio drivers, camera drivers, Bluetooth drivers, Wi-Fi drivers, and power management. The system runtime library layer 340 provides support for key features of the Android system through several C / C++ libraries. For example, the SQLite library provides database support, the OpenGL / ES library provides 3D graphics support, and the Webkit library provides browser kernel support. The system runtime library layer 340 also provides the Android runtime library, which mainly provides core libraries that allow developers to write Android applications using the Java language. The Application Framework Layer 360 provides various APIs that may be used when building applications. Developers can also use these APIs to build their own applications, such as activity management, window management, view management, notification management, content provider, package management, call management, resource management, and location management. At least one application runs in the Application Layer 380. These applications can be native applications that come with the operating system, such as contacts, SMS, clock, and camera apps; or third-party applications developed by third-party developers, such as games, instant messaging, and photo editing apps.

[0199] Taking the operating system as an example (iOS), the programs and data stored in memory 120 are as follows: Figure 11 As shown, the iOS system includes: Core OS layer 420, Core Services layer 440, Media layer 460, and Cocoa Touch layer 480. Core OS layer 420 includes the operating system kernel, drivers, and low-level program frameworks. These low-level program frameworks provide hardware-level functionality for use by the program frameworks located in Core Services layer 440. Core Services layer 440 provides system services and / or program frameworks required by applications, such as Foundation framework, account framework, advertising framework, data storage framework, network connectivity framework, geolocation framework, motion framework, etc. Media layer 460 provides applications with audiovisual interfaces, such as interfaces related to graphics and images, audio technology, video technology, and AirPlay (wireless playback of audio and video transmission technologies). Cocoa Touch layer 480 provides various commonly used interface-related frameworks for application development and is responsible for user touch interaction on electronic devices. Examples include local notification services, remote push services, advertising frameworks, game tool frameworks, message user interface (UI) frameworks, UIKit frameworks, map frameworks, and so on.

[0200] exist Figure 11 The framework shown includes, but is not limited to, the base framework in the core service layer 440 and the UIKit framework in the touchable layer 480. The base framework provides many basic object classes and data types, offering the most basic system services to all applications, and is independent of the UI. The UIKit framework, on the other hand, provides a basic UI class library for creating touch-based user interfaces. iOS applications can use the UIKit framework to provide their UI, thus providing the application's infrastructure for building user interfaces, drawing, handling user interaction events, responding to gestures, and so on.

[0201] The methods and principles for implementing data communication between third-party applications and the operating system in the iOS system can be referenced from the Android system, and will not be elaborated here.

[0202] The input device 130 is used to receive input instructions or data, and includes, but is not limited to, a keyboard, mouse, camera, microphone, or touch device. The output device 140 is used to output instructions or data, and includes, but is not limited to, a display device and a speaker. In one example, the input device 130 and the output device 140 can be combined into a touch screen, which is used to receive touch operations from the user using a finger, stylus, or any suitable object on or near it, and to display the user interface of various applications. The touch screen is usually located on the front panel of the electronic device. The touch screen can be designed as a full-screen, curved screen, or irregularly shaped screen. The touch screen can also be designed as a combination of a full-screen and a curved screen, or a combination of an irregularly shaped screen and a curved screen; this embodiment of the application does not limit this.

[0203] In addition, those skilled in the art will understand that the structure of the electronic device shown in the above figures does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements. For example, the electronic device may also include radio frequency circuits, input units, sensors, audio circuits, wireless fidelity (WiFi) modules, power supplies, Bluetooth modules, etc., which will not be described in detail here.

[0204] In the embodiments of this application, the executing entity for each step can be the electronic device described above. Optionally, the executing entity for each step is the operating system of the electronic device. The operating system can be Android, iOS, or other operating systems; this embodiment of the application does not limit this.

[0205] The electronic device in this embodiment may also be equipped with a display device, which can be various devices capable of display functions, such as: cathode ray tube display (CR), light-emitting diode display (LED), electronic ink screen, liquid crystal display (LCD), plasma display panel (PDP), etc. Users can use the display device on the electronic device 101 to view displayed text, images, videos, and other information. The electronic device may be a smartphone, tablet computer, gaming device, AR (Augmented Reality) device, automobile, data storage device, audio playback device, video playback device, laptop, desktop computing device, wearable device such as electronic watch, electronic glasses, electronic helmet, electronic bracelet, electronic necklace, electronic clothing, etc.

[0206] exist Figure 8 In the illustrated electronic device, the processor 110 can be used to call the application program stored in the memory 120 and specifically perform the following operations:

[0207] Receive a search request input by a user, obtain the user's search statement, and determine the user's search semantic vector corresponding to the user's search statement;

[0208] The user search semantic vector is subjected to promotion information summary matching processing to obtain a promotion information semantic summary set, and a candidate promotion auction term set is obtained based on the promotion information semantic summary set;

[0209] Based on the set of candidate promotional auction terms, target promotional auction terms are determined, and information promotion processing is performed based on the target promotional auction terms.

[0210] In one embodiment, the processor 110 performs the following steps after executing the promotional information semantic summary matching process on the user search semantic vector to obtain a promotional information semantic summary set, and obtains a candidate promotional auction term set based on the promotional information semantic summary set:

[0211] In the promotion information semantic summary table, the user search semantic vector is subjected to promotion information summary matching processing to obtain a preset number of promotion information semantic summary sets;

[0212] Query the set of candidate promotional auction terms corresponding to the set of promotional information semantic summaries from the promotional information semantic summary table.

[0213] In one embodiment, the processor 110 performs promotional information summary matching processing on the user search semantic vector in the promotional information semantic summary table to obtain a preset number of promotional information semantic summary sets, including:

[0214] In the promotion information semantic summary table, the vector similarity between the user search semantic vector and the reference promotion information semantic summary is calculated, and a promotion information semantic summary set including a preset number of reference promotion information semantic summaries is selected from the promotion information semantic summary table based on the vector similarity.

[0215] In one embodiment, when executing the promotional information processing method, the processor 110 further performs the following steps:

[0216] Obtain the historical search behavior sequence for the reference user, and determine the historical promotional information text of the long text type and the reference promotional auction terms corresponding to the historical promotional information text from the historical search behavior sequence;

[0217] Semantic summarization is performed on the historical promotional information text to obtain a reference promotional information semantic summary. Based on the reference promotional auction terms and the reference promotional information semantic summary, a promotional information semantic summary table is constructed for each reference promotional auction term.

[0218] In one embodiment, when the processor 110 performs semantic summary extraction on the historical promotional information text to obtain a reference promotional information semantic summary, it may perform the following steps:

[0219] Determine the summary task prompt words for the historical promotion information text, input the summary task prompt words and the historical promotion information text into the target large language model for semantic summary extraction, and output the reference promotion information semantic summary.

[0220] In one embodiment, when the processor 110 executes the step of constructing a semantic summary table of promotional information for each reference promotional auction term based on the reference promotional auction term and the reference promotional information semantic summary, it may perform the following steps:

[0221] The historical promotional information text in each of the historical search behavior sequences is replaced with the semantic summary of the reference promotional information to generate a semantic summary table of promotional information for each of the reference promotional auction terms, wherein the historical search behavior sequences include the reference promotional auction terms corresponding to the historical promotional information text.

[0222] In one embodiment, the processor 110, when executing the historical promotional information text determined from the historical search behavior sequence as long text type, may perform the following steps:

[0223] Identify candidate historical promotional texts of long text type from historical search behavior sequences.

[0224] Based on all historical search behavior sequences, determine the number of times the candidate historical promotional information texts co-occur and the interval time corresponding to the historical promotional information texts;

[0225] The importance score of the candidate historical promotion information text is determined by a target scoring function based on the text co-occurrence frequency and the interval time.

[0226] The target scoring function satisfies the following formula:

[0227] f=*aΔ T

[0228] Wherein, f is the importance score, x is the number of times the text co-occurs, a is the attenuation coefficient, and ΔT is the interval time.

[0229] In one embodiment, when the processor 110 performs the step of determining the target promotional auction term based on the candidate promotional auction term set, it may execute the following steps:

[0230] Determine the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set, and select a target number of target promotion auction terms from the candidate promotion auction terms based on the promotion effect parameters.

[0231] In one embodiment, the processor 110 may perform the following steps when determining the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set:

[0232] Determine the bid coefficient, estimated click-through rate, and recall rate for each candidate promotional auction term in the candidate promotional auction term set;

[0233] The promotion effect parameters are obtained by multiplying the auction keyword bidding coefficient, the estimated click-through rate, and the recall rate.

[0234] In one or more embodiments of this application, by receiving a user's search request, obtaining the user's search statement and determining the user's search semantic vector corresponding to the user's search statement, performing promotional information summary matching processing on the user's search semantic vector to obtain a promotional information semantic summary set, and then obtaining a candidate promotional auction term set based on the promotional information semantic summary set. Matching based on promotional information summaries rather than long text-type promotional information can recall a candidate promotional auction term set with high accuracy. Subsequently, based on the candidate promotional auction term set, target promotional auction terms are determined, and information promotion processing is performed based on the target promotional auction terms. During information promotion processing, the promotional information recalled based on the target promotional auction terms has high relevance. It can retain rich semantic features of the promotional information by leveraging the promotional information summary, thereby improving the delivery and recommendation effect of promotional information in information search scenarios, improving the relevance of promotional information recommendations, and improving the user's promotional information experience.

[0235] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory, or random access memory, etc.

[0236] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.< / xxx>

Claims

1. A method for processing promotional information, characterized in that, The method includes: Receive a search request input by a user, obtain the user's search statement, and determine the user's search semantic vector corresponding to the user's search statement; In the promotion information semantic summary table, the user search semantic vector is subjected to promotion information summary matching processing to obtain a preset number of promotion information semantic summary sets; Query the set of candidate promotional auction terms corresponding to the set of promotional information semantic summaries from the promotional information semantic summary table; Based on the set of candidate promotion auction terms, target promotion auction terms are determined, and information promotion processing is performed based on the target promotion auction terms. The method further includes: Obtain the historical search behavior sequence for the reference user, and determine the historical promotional information text of the long text type and the reference promotional auction keywords corresponding to the historical promotional information text from the historical search behavior sequence; the long text type includes promotional information title type and promotional information description type; Semantic summarization is performed on the historical promotion information text to obtain a reference promotion information semantic summary. Based on the reference promotion auction terms and the reference promotion information semantic summary, a promotion information semantic summary table is constructed for each reference promotion auction term. The step of determining the long-text type of historical promotional information text from the historical search behavior sequence includes: Identify candidate historical promotional texts of long text type from historical search behavior sequences. Based on all historical search behavior sequences, determine the number of times the candidate historical promotional information texts co-occur and the interval time corresponding to the historical promotional information texts; Based on the number of text co-occurrences and the interval time, a target scoring function is used to determine the importance score of the candidate historical promotion information text, and the historical promotion information text is determined from the candidate historical promotion information text based on the importance score.

2. The method according to claim 1, characterized in that, The step of performing promotional information summary matching processing on the user search semantic vector in the promotional information semantic summary table to obtain a preset number of promotional information semantic summary sets, including: Based on the reference promotional information semantic summary in the promotional information semantic summary table, calculate the vector similarity between the user search semantic vector and the reference promotional information semantic summary, and select a set of promotional information semantic summaries including a preset number of reference promotional information semantic summaries from the promotional information semantic summary table based on the vector similarity.

3. The method according to claim 1, characterized in that, The step of extracting a semantic summary of the historical promotional information text to obtain a reference promotional information semantic summary includes: Determine the summary task prompt words for the historical promotion information text, input the summary task prompt words and the historical promotion information text into the target large language model for semantic summary extraction, and output the reference promotion information semantic summary.

4. The method according to claim 1, characterized in that, The construction of a semantic summary table for each reference promotional auction term based on the reference promotional auction term and the semantic summary of the reference promotional information includes: The historical promotional information text in each of the historical search behavior sequences is replaced with the semantic summary of the reference promotional information to generate a semantic summary table of promotional information for each of the reference promotional auction terms, wherein the historical search behavior sequences include the reference promotional auction terms corresponding to the historical promotional information text.

5. The method according to claim 1, characterized in that, The target scoring function satisfies the following formula: Where f is the importance score, x is the number of times the text co-occurs, and a is the attenuation coefficient. The interval time is mentioned.

6. The method according to claim 1, characterized in that, The process of determining the target promotional auction term based on the candidate promotional auction term set includes: Determine the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set, and select a target number of target promotion auction terms from the candidate promotion auction terms based on the promotion effect parameters.

7. The method according to claim 1, characterized in that, The step of determining the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set includes: Determine the bid coefficient, estimated click-through rate, and recall rate for each candidate promotional auction term in the candidate promotional auction term set; The promotion effect parameters are obtained by multiplying the auction keyword bidding coefficient, the estimated click-through rate, and the recall rate.

8. A promotional information processing device, characterized in that, The device includes: The request processing module is used to perform promotion information summary matching processing on the user search semantic vector in the promotion information semantic summary table to obtain a preset number of promotion information semantic summary sets; and to query the candidate promotion auction term set corresponding to the promotion information semantic summary set from the promotion information semantic summary table. The set determination module is used to perform promotion information summary matching processing on the user search semantic vector to obtain a promotion information semantic summary set, and to obtain a candidate promotion auction term set based on the promotion information semantic summary set; The information promotion module is used to determine the target promotion auction term based on the candidate promotion auction term set, and to perform information promotion processing based on the target promotion auction term; The device is also used for: Obtain the historical search behavior sequence for the reference user, and determine the historical promotional information text of the long text type and the reference promotional auction keywords corresponding to the historical promotional information text from the historical search behavior sequence; the long text type includes promotional information title type and promotional information description type; Semantic summarization is performed on the historical promotion information text to obtain a reference promotion information semantic summary. Based on the reference promotion auction terms and the reference promotion information semantic summary, a promotion information semantic summary table is constructed for each reference promotion auction term. Specifically, when determining historical promotional information text of the long text type from a historical search behavior sequence, the device is used for: Identify candidate historical promotional texts of long text type from historical search behavior sequences. Based on all historical search behavior sequences, determine the number of times the candidate historical promotional information texts co-occur and the interval time corresponding to the historical promotional information texts; Based on the number of text co-occurrences and the interval time, a target scoring function is used to determine the importance score of the candidate historical promotion information text, and the historical promotion information text is determined from the candidate historical promotion information text based on the importance score.

9. The apparatus according to claim 8, characterized in that, The set determination module is used for: In the promotion information semantic summary table, the vector similarity between the user search semantic vector and the reference promotion information semantic summary is calculated, and a promotion information semantic summary set including a preset number of reference promotion information semantic summaries is selected from the promotion information semantic summary table based on the vector similarity.

10. The apparatus according to claim 8, characterized in that, The set determination module is used for: Determine the summary task prompt words for the historical promotion information text, input the summary task prompt words and the historical promotion information text into the target large language model for semantic summary extraction, and output the reference promotion information semantic summary.

11. The apparatus according to claim 8, characterized in that, The set determination module is used for: The historical promotional information text in each of the historical search behavior sequences is replaced with the semantic summary of the reference promotional information to generate a semantic summary table of promotional information for each of the reference promotional auction terms, wherein the historical search behavior sequences include the reference promotional auction terms corresponding to the historical promotional information text.

12. The apparatus according to claim 8, characterized in that, The target scoring function satisfies the following formula: Where f is the importance score, x is the number of times the text co-occurs, and a is the attenuation coefficient. The interval time is mentioned.

13. The apparatus according to claim 8, characterized in that, The information promotion module is used for: Determine the promotion effect parameters for each candidate promotion auction term in the candidate promotion auction term set, and select a target number of target promotion auction terms from the candidate promotion auction terms based on the promotion effect parameters.

14. The apparatus according to claim 13, characterized in that, The information promotion module is used for: Determine the bid coefficient, estimated click-through rate, and recall rate for each candidate promotional auction term in the candidate promotional auction term set; The promotion effect parameters are obtained by multiplying the auction keyword bidding coefficient, the estimated click-through rate, and the recall rate.

15. A computer storage medium, characterized in that, The computer storage medium stores a plurality of instructions, which are adapted to be loaded by a processor and executed as method steps as claimed in any one of claims 1 to 7.

16. An electronic device, characterized in that, include: A processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and executed the method steps as claimed in any one of claims 1 to 7.

Images