Dynamic user portrait advertisement style switching method based on context awareness

Abstract

The invention discloses a dynamic user portrait advertisement style switching method based on context awareness, and relates to the technical field of advertisement putting switching, and the method comprises the steps: recognizing the context information of an advertisement request, determining a current effective festival label time window based on the advertisement request time in the context information, and generating a festival context packet; acquiring real-time behavior data of the user, performing session merging and behavior event classification on the real-time behavior data of the user, generating regular real-time behavior data of the user, associating the festival context packet with the regular real-time behavior data of the user, and generating a portrait context packet; based on the portrait context packet, a target interaction structure and a target festival theme are decided and packaged to form a delivery resource packet; and rendering the target interaction structure in the putting resource packet, and loading a style resource corresponding to the target festival theme to complete advertisement style switching. According to the method, the consistency of advertisement style switching and the current intention of the user is improved by constructing the multi-layer dependency analysis tree.

Description

Technical Field

[0001] This invention relates to the field of advertising delivery switching technology, and in particular to a method for switching advertising styles based on context-aware dynamic user profiles. Background Technology

[0002] In the field of digital advertising, dynamically matching ad styles with users' current interests is a core element in improving interaction efficiency and business conversion. With enhanced user behavior data collection capabilities and advancements in context modeling technology, ad presentation mechanisms are gradually evolving towards a deeper integration of high timeliness, scene awareness, and personalization. Especially in time-sensitive holiday marketing scenarios, the key technological path to achieving precise reach and immersive experiences lies in how to structurally integrate holiday semantic information with users' real-time interactive intentions, and thereby drive adaptive adjustments to ad visuals and interactive elements.

[0003] Existing technologies, when constructing user profiles, often fail to encapsulate the holiday time window with the standardized real-time behavioral data in a joint context. This results in a lack of collaborative awareness of the user's current conversational intent and the holiday context when switching ad styles. Consequently, although the holiday theme is triggered, it is difficult to match the user's actual interaction state, affecting the relevance of the presentation. Existing technologies typically use preset time rules or static tag combinations to approximate this problem, but they struggle to achieve fine-grained and closed-loop driven dynamic adaptation. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a context-aware dynamic user profile ad style switching method to solve the problem of dynamic adaptation that is difficult to achieve fine-grained and closed-loop driven.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for switching dynamic user profile advertising styles based on context awareness, comprising, Identify the context information of the ad request, determine the current valid holiday tag time window based on the ad request time in the context information, and generate a holiday context package; Acquire real-time user behavior data, perform session merging and behavior event categorization on the real-time user behavior data, generate standardized real-time user behavior data, associate the holiday context package with the standardized real-time user behavior data, and generate a profile context package. Based on the profile context package, the target interaction structure and target holiday theme are determined and encapsulated into a delivery resource package; Render the target interactive structure in the resource package, load the style resources corresponding to the target holiday theme, complete the ad style switching, collect the display feedback data generated after the ad style switching, and use the display feedback data as the real-time user behavior data input for the next switching to drive the next switching decision.

[0007] As a preferred embodiment of the context-aware dynamic user profile ad style switching method of the present invention, wherein: the generation of the holiday context package specifically includes, Extract the ad request time from the context information; Match the ad request time with the holiday calendar database to determine the current valid holiday tag time window; The holiday tag time window is merged with the ad placement specifications and media page type in the context information to generate a holiday context package.

[0008] As a preferred embodiment of the context-aware dynamic user profile advertising style switching method of the present invention, wherein: determining the currently valid holiday tag time window specifically involves, Convert ad request times to a unified time base; Retrieve festival records with a unified time base from the festival calendar database; Filter the holiday event records that are active, and use them as the current valid holiday tag time window.

[0009] As a preferred embodiment of the context-aware dynamic user profile advertising style switching method of the present invention, wherein: the generation of regularized real-time user behavior data specifically includes, Acquire real-time user behavior data, perform a merge operation on the real-time user behavior data according to the session identifier, and output the session-level behavior sequence; Perform semantic classification on behavioral events in a session-level behavioral sequence to generate a set of classified behavioral events; The categorized behavioral event sets are reorganized in chronological order to generate well-organized real-time user behavior data.

[0010] As a preferred embodiment of the context-aware dynamic user profile advertising style switching method of the present invention, wherein: the generation of the profile context package specifically includes, Extract the holiday tag time window and interaction structure constraint information from the holiday context package; Based on the holiday tag time window, the regularized real-time user behavior data is aligned with the time window to obtain the behavior data within the window; The behavioral data within the window is fused with the interaction structure constraint information to generate a profile context package.

[0011] As a preferred embodiment of the context-aware dynamic user profile advertising style switching method of the present invention, wherein: the generation of the categorized behavioral event set specifically includes, Construct a multi-layer dependency parsing tree for behavior events, perform cross-layer semantic propagation on behavior events on the multi-layer dependency parsing tree, and generate enhanced semantic tags; The enhanced semantic tags are mapped to predefined behavior category rules for classification, generating a set of classified behavior events.

[0012] As a preferred embodiment of the context-aware dynamic user profile advertising style switching method of the present invention, the session identifier is a unique tracking identifier obtained from the context information of the advertising request, which is generated and transmitted by the media terminal at the beginning of the user session.

[0013] As a preferred embodiment of the context-aware dynamic user profile ad style switching method of the present invention, the interaction structure constraint information is derived from the ad slot specifications and media page type, and limits the set of rules that restrict the currently available ad interaction forms.

[0014] As a preferred embodiment of the context-aware dynamic user profile ad style switching method of the present invention, wherein: the formation of the delivery resource package specifically refers to, Extract in-window behavior data and interaction structure constraint information from the portrait context package; By analyzing user behavior patterns within the holiday time window using in-window behavioral data, the holiday phase status can be obtained. Based on the festival stage status and interaction structure constraints, the target interaction structure is matched from a predefined set of interaction structures; Use in-window behavioral data to identify user interests and preferences, and match target holiday theme templates from a holiday theme template library; Combine and encapsulate the target interactive structure with the target holiday theme template to generate a deployment resource package.

[0015] As a preferred embodiment of the context-aware dynamic user profile ad style switching method of the present invention, wherein: the collection of display feedback data generated after ad style switching specifically includes, Separate the rendering instruction set corresponding to the target interactive structure and the style resource descriptor list corresponding to the target holiday theme template from the resource package; Create an ad rendering sandbox in the ad rendering client, execute the rendering instruction set in the ad rendering sandbox, instantiate the target interaction structure, and output the interaction structure component; The instantiated target interaction structure refers to the creation of specific, interactive advertising interface component instances in the advertising rendering sandbox based on the rendering instruction set. Activate style resources through the list of style resource descriptors, inject style resources into the ad rendering sandbox of interactive structural components, and complete the ad style switching; An event listening pipeline is established in the ad rendering sandbox of the interactive structural components to capture user interaction behavior and generate the original interactive event stream; The original interactive event stream is formatted and associated with the target interactive structure and the target identifier of the target holiday theme template, and the output display feedback data package is generated.

[0016] The beneficial effects of this invention are as follows: By constructing a multi-layer dependency parse tree, it achieves structured enhancement of behavioral events and semantic classification in holiday scenarios. Original behaviors such as page browsing, searching, clicking, and adding to cart are semantically fused within contextual dependencies to generate a regularized behavioral sequence with holiday intent recognition capabilities, which is then aligned with the holiday context package to form a profile context package. The multi-layer dependency parse tree allows for precise characterization of user interests and preferences based on the behavioral evolution path within the holiday time window, avoiding misjudgments of interest caused by traditional behavioral statistics methods that ignore behavioral temporal dependencies and semantic associations. This provides a high-fidelity, scenario-sensitive decision-making basis for matching target interaction structures with holiday themes, significantly improving the consistency between ad style switching and the user's current intent. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a context-aware, dynamic user profile ad style switching method.

[0019] Figure 2 A flowchart for generating a portrait context package.

[0020] Figure 3 A flowchart for creating resource delivery packages.

[0021] Figure 4 A flowchart for ad style switching and feedback. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for switching dynamic user profile advertising styles based on context awareness, including the following steps: S1. Identify the context information of the ad request, determine the current valid holiday tag time window based on the ad request time in the context information, and generate a holiday context package.

[0026] Identify the contextual information of the ad request and extract the ad request time from the contextual information.

[0027] Specifically, the process involves identifying ad requests using predefined fields in the context information uploaded by the media terminal. These predefined fields include ad placement identifier, device type, media page type, user identifier, ad request time, and location information. When an ad request arrives, the media terminal parses the context information uploaded along with it. This context information includes the ad placement identifier, device type, media page type, user identifier, ad request time, and location information. Content is read from each field of the context information. When the predefined ad request time field is encountered, its value is extracted completely as the ad request time, represented as a timestamp. The extracted ad request time is then directly used for subsequent matching with the holiday calendar database.

[0028] Match the ad request time with the holiday calendar database to determine the current valid holiday tag time window; Specifically, this involves converting the ad request time to a unified time base; the unified time base is the Harmonized Universal Time (UTC) timestamp base. Retrieve festival records with a unified time base from the festival calendar database.

[0029] Specifically, the holiday calendar database stores holiday records using Coordinated Universal Time (UTC) timestamps as index keys. Each holiday record contains a holiday start timestamp field, a holiday end timestamp field, and a holiday tag field. During retrieval, the extracted ad request time is used as the UTC timestamp. All holiday records in the holiday calendar database are traversed directly. For each holiday record, the ad request time is compared to see if it falls between the holiday start timestamp field and the holiday end timestamp field. If the ad request time falls between these two fields, the corresponding holiday record is added to the candidate holiday record set. After traversal, the candidate holiday record set is output as the retrieved holiday record. If the ad request time does not fall between the holiday start timestamp field and the holiday end timestamp field of any holiday record, the corresponding holiday record is not added to the candidate holiday record set. After traversing all holiday records, if the candidate holiday record set is empty, an empty set is output as the retrieved holiday record, indicating that there are currently no valid holidays.

[0030] The holiday calendar database is constructed by pre-collecting global Gregorian calendar holidays, lunar calendar holidays, religious holidays, e-commerce promotional events, and brand-customized holiday events. Each holiday record includes a holiday tag, a holiday start timestamp field, a holiday end timestamp field, and a corresponding holiday description field. All timestamps are uniformly converted to Coordinated Universal Time (UTC) timestamps before storage. This is used to quickly retrieve and determine whether the current time is within the active time window of a certain holiday event when an ad request arrives, thereby outputting the currently valid holiday tag time window and supporting the subsequent generation of holiday context packets.

[0031] Filter the holiday event records that are active, and use them as the current valid holiday tag time window.

[0032] Specifically, the process involves: traversing each retrieved holiday record, where each record includes a holiday status field with a value indicating whether it is active or inactive; checking the value of the holiday status field for each record; adding the record to the valid holiday record set if it is active, and not adding it if it is inactive; using the valid holiday record set as the current valid holiday tag time window if the set contains multiple records, and ensuring that the window covers the holiday tag field, holiday start timestamp field, and holiday end timestamp field for all active holiday records; and using the empty holiday record set as the empty window, indicating no active holidays. This window is then directly used for subsequent merging with the ad slot specifications and media page type in the context information.

[0033] The holiday tag time window is merged with the ad slot specifications and media page types in the context information to generate a holiday context package. Specifically, the holiday tag field, holiday start timestamp field, and holiday end timestamp field in the holiday tag time window are directly combined with the ad slot specifications and media page types in the context information into a structured data package as the holiday context package.

[0034] S2. Obtain real-time user behavior data, perform session merging and behavior event categorization on the real-time user behavior data, generate standardized real-time user behavior data, associate the holiday context package with the standardized real-time user behavior data, and generate a profile context package.

[0035] Acquire real-time user behavior data and perform a merging operation on the real-time user behavior data according to the session identifier, and output the session-level behavior sequence; wherein, the session identifier is a unique tracking identifier obtained from the context information of the advertising request, which is generated and transmitted by the media terminal at the beginning of the user session.

[0036] Specifically, real-time user behavior data is collected through event tracking logs reported in real time by media terminals when users perform actions such as browsing pages, clicking on ad slots, viewing products, adding items to carts, searching keywords, and staying on pages. Each piece of real-time user behavior data includes a user identifier field, a behavior type field, a behavior timestamp field, a behavior target field, and a session identifier field. When acquiring real-time user behavior data, all real-time user behavior data matching the user identifier field of the current ad request are read from the event tracking logs reported by the media terminals to form an initial set of real-time user behavior data. When performing a merge operation, all real-time user behavior data in the initial set of real-time user behavior data are grouped according to the session identifier field, with each group corresponding to the same session identifier field value. The real-time user behavior data within the same group are arranged in ascending order of the behavior timestamp field to form an ordered subsequence of real-time user behavior data. After all grouping is completed, each ordered subsequence of real-time user behavior data is connected in the order of most recent session time to form a complete session-level behavior sequence.

[0037] Perform semantic classification on the behavioral events in the session-level behavioral sequence to generate a set of classified behavioral events.

[0038] Specifically, this involves constructing a multi-layered dependency parsing tree for behavior events, performing cross-layer semantic propagation on the behavior events within the multi-layered dependency parsing tree, and generating enhanced semantic tags.

[0039] Furthermore, the behavioral events in the session-level behavioral sequence are added as nodes in chronological order to the multi-level dependency parse tree. The first level of the multi-level dependency parse tree contains all behavioral event nodes, and each node corresponds to a behavioral event.

[0040] Starting with the first behavior event in the session-level behavior sequence, a root node is created as the first behavior event node. Then, subsequent behavior events are processed in sequence. For the currently processed behavior event, a new node is created as the current behavior event node, and the current behavior event node is connected to the previous behavior event node as a parent-child relationship, thus forming a chain-like basic structure.

[0041] Based on the chain-like infrastructure, for each behavior event node, if the value of the behavior type field indicates page browsing or keyword search, a child node layer is added below the current behavior event node to record the value of the behavior target field as a page or keyword node; if the value of the behavior type field indicates clicking an ad slot, viewing a product, or adding it to the shopping cart, the current behavior event node is connected to the most recent previous page or keyword node as a dependency relationship, thus forming a multi-level dependency parse tree. The upper-level nodes of the multi-level dependency parse tree represent the context behavior, and the lower-level nodes represent the specific target behavior.

[0042] During cross-layer semantic propagation, the process traverses upwards from the leaf nodes of the multi-layer dependency parse tree to the root node. The semantic labels of the leaf nodes are initially set to the combined values ​​of the behavior type field and the behavior target field. This propagation continues upwards layer by layer. Upon reaching a parent node, the semantic labels of the child nodes are added to the semantic label set of the parent node. Simultaneously, the process traverses downwards from the root node, passing the semantic label set of the parent node down to all child nodes. This ensures that each behavior event node receives extended semantic labels from both the upper and lower layers.

[0043] After propagation is complete, the extended semantic tag attached to each behavioral event node serves as an enhanced semantic tag.

[0044] It should be noted that the multi-layer dependency parsing tree adopts a tree-like hierarchy formed by the expansion of the chain-based basic structure. The tree depth is determined by the number of times the page browsing or search keyword behavior occurs in the behavior event sequence. Each page browsing or search keyword behavior corresponds to the addition of a child node layer. The node types include behavior event nodes and page or keyword nodes. Behavior event nodes correspond to each piece of real-time user behavior data in the session-level behavior sequence, and page or keyword nodes correspond to the value of the behavior target field.

[0045] The input feature dimensions are fixed as five fields included in the user's real-time behavior data: user identifier, behavior type, behavior timestamp, behavior target, and session identifier.

[0046] The optimization goal is to improve the accuracy of enhanced semantic tags in expressing user interests in holiday scenarios, and to ensure that the extended semantic tags attached to each behavioral event node after cross-layer semantic propagation can fully reflect the behavioral intent related to the upper and lower layer contexts.

[0047] The enhanced semantic tags are mapped to predefined behavior category rules for classification, generating a set of classified behavior events. The predefined behavior category rules are set based on common user behavior patterns in advertising business and interest classification requirements in holiday scenarios. They contain multiple rule entries, each consisting of a rule condition field and a behavior category field. The rule condition field is defined as an exact match combination of one or more enhanced semantic tags.

[0048] Specifically, it involves traversing all behavior event nodes in the multi-level dependency parse tree and extracting the accompanying enhanced semantic tags for each behavior event node.

[0049] For the enhanced semantic label of the current behavior event node, it is compared with the rule condition field of each rule entry in turn. If the enhanced semantic label completely matches the rule condition field of a certain rule entry, the value of the corresponding behavior category field of the current behavior event node is assigned as the classification result.

[0050] If the enhanced semantic label matches multiple rule entries, the rule entry that appears first in the list of rule entries will be selected as the classification result.

[0051] If the enhanced semantic tag does not match any rule entry, the value of the behavior category field is assigned to the default category.

[0052] After all behavior event nodes have been assigned, all behavior event nodes with a behavior category field will be collected into a categorized set of behavior events.

[0053] The categorized behavioral event sets are reorganized in chronological order to generate well-organized real-time user behavior data.

[0054] Extract the holiday tag time window and interaction structure constraint information from the holiday context package; the interaction structure constraint information is derived from the ad placement specifications and media page type, and is a set of rules that limit the currently available ad interaction forms.

[0055] The interaction structure constraint information is obtained by directly looking up the corresponding set of ad interaction rules based on the ad slot specifications and media page types contained in the holiday context package through a preset ad slot interaction mapping table. The preset ad slot interaction mapping table is set based on the compatibility of different ad slot specifications and media page types in the advertising business and the user interaction experience requirements. The preset ad slot interaction mapping table defines the set of ad interaction rules allowed under different combinations of ad slot specifications and media page types. For example, the large image ad slot on the information flow page supports carousel and video interaction, while the pop-up ad slot supports close and click-to-jump interaction.

[0056] Based on the holiday tag time window, the regularized real-time user behavior data is aligned with the time window to obtain the behavior data within the window.

[0057] Specifically, the process is as follows: The start and end timestamp fields of the holiday are retrieved from the holiday tag time window as alignment boundaries. All behavior items in the user's real-time behavior data are traversed and normalized. The value of the behavior timestamp field for each behavior item is checked to see if it falls between the start and end timestamp fields. If it does, the corresponding behavior item is added to the window's behavior data set. If it does not, the corresponding behavior item is not added to the window's behavior data set. After traversal, the window's behavior data set is merged into the window's behavior data. If the holiday tag time window contains multiple holiday records, the same alignment operation is performed on the start and end timestamp fields of each holiday record, merging all eligible behavior items into the same window's behavior data set to avoid duplication.

[0058] The behavior data within the window is fused with the interaction structure constraint information to generate a profile context package. Specifically, the set of advertising interaction rules in the interaction structure constraint information is directly combined as additional fields into the behavior data set within the window to form a complete structured data package as the profile context package.

[0059] S3. Based on the profile context package, determine the target interaction structure and target holiday theme, and encapsulate them into a delivery resource package.

[0060] Extract in-window behavior data and interaction structure constraint information from the portrait context package; By analyzing user behavior patterns within the holiday time window, the holiday phase status can be obtained.

[0061] Specifically, the process involves: arranging all behavior items in the window's behavior data set in ascending order of the behavior timestamp field to form an ordered behavior sequence within the window; traversing the ordered behavior sequence within the window, counting the frequency of occurrence of different behavior category fields, and recording the order of occurrence of the behavior category fields as holiday search interest, holiday browsing products, holiday adding to cart, and holiday purchase completion; and matching predefined holiday stage rules based on the statistical results and the order of occurrence. If the behavior category fields in the ordered window's behavior sequence are only holiday search interests and holiday browsed products, and the occurrence counts of the behavior category fields for "added to cart" and "purchase completed" are zero, then the holiday phase is in the pre-heating phase. If the occurrence counts of the behavior category field for "added to cart" in the ordered window's behavior sequence are greater than zero, and the occurrence counts of the behavior category field for "purchase completed" are zero, then the holiday phase is in the peak phase. If the occurrence counts of the behavior category field for "purchase completed" in the ordered window's behavior sequence are greater than the sum of the occurrence counts of the behavior category fields for "added to cart," "search interests," and "browsed products," then the holiday phase is in the climax phase. If all behavior timestamp fields in the ordered window's behavior sequence fall within the interval of one hour before to one hour after the holiday end timestamp, and the occurrence of the behavior category field is "viewing orders" or "reviewing," then the holiday phase is in the residual heat phase. If the behavior data set in the window is empty, or the occurrence counts of the behavior category fields for "search interests," "browsed products," "added to cart," "purchase completed," and "viewing orders" and "reviewing" in the ordered window's behavior sequence are all zero, then the holiday phase is in the no-holiday phase.

[0062] The holiday stage status obtained after matching is directly used for subsequent operations to match the target interaction structure based on the holiday stage status and interaction structure constraint information. The predefined holiday stage rules are set based on experience summarizing user behavior patterns during the holiday event. For example, the preheating stage is mainly for searching and browsing, the peak stage is mainly for adding to cart, the climax stage is mainly for purchasing, and the after-sales stage is mainly for after-sales service.

[0063] Based on the festival stage status and interaction structure constraints, the target interaction structure is matched from a predefined set of interaction structures.

[0064] Specifically, it iterates through all interaction structure entries in the predefined interaction structure set, checks whether the value of the applicable holiday stage state field matches the current holiday stage state for each interaction structure entry, and checks whether the value of the applicable advertising interaction form rule field is compatible with the advertising interaction form rule set in the interaction structure constraint information.

[0065] If the value of the applicable holiday stage state field of the interaction structure entry is completely consistent with the current holiday stage state, and the value of the applicable advertising interaction form rule field is included in the advertising interaction form rule set in the interaction structure constraint information, then the corresponding interaction structure entry will be marked as a candidate interaction structure.

[0066] After traversal, the interaction structure with the highest target priority is selected from all candidate interaction structures as the target interaction structure. The target priority is determined according to the order of the items in the predefined interaction structure set, with the interaction structure item at the beginning corresponding to the highest target priority.

[0067] If no candidate interaction structure is available, the default interaction structure entry in the predefined interaction structure set will be selected as the target interaction structure. The predefined interaction structure set is set based on the compatibility between the different holiday stages and the advertising interaction form rule set in the advertising business. For example, the carousel interaction structure is prioritized during the preheating stage, the video interaction structure is prioritized during the peak stage, and the strong reminder pop-up interaction structure is prioritized during the climax stage.

[0068] Use in-window behavior data to identify user interests and preferences, and match target holiday theme templates from a holiday theme template library.

[0069] Specifically, the process involves: traversing all behavior items in the behavior data set within the window, statistically analyzing the occurrence of different behavior category fields, and extracting product categories, brands, or keywords appearing in the behavior target field as interest items; selecting the behavior category field with the highest frequency in the statistical results as the primary interest category, and selecting the product category, brand, or keyword corresponding to the repeated occurrence or most recent behavior timestamp field in the behavior target field as specific user interest preferences; user interest preferences are formed by combining the primary interest category and specific interest preferences, for example, the primary interest category is holiday purchase intention, and the specific interest preference is electronic product category or specific holiday gift keywords; the holiday theme template library contains multiple holiday theme template entries, each of which consists of a template identifier field, an applicable holiday tag field, an applicable interest category field, and an applicable interest keyword field.

[0070] Iterate through all holiday theme template entries in the holiday theme template library. For each entry, check if the value of the applicable holiday tag field matches the holiday tag field in the holiday tag time window, and if the value of the applicable interest category field matches the primary interest category, and if the value of the applicable interest keyword field is compatible with the product category, brand, or keyword in the specific interest preference. If the applicable holiday tag field, applicable interest category field, and applicable interest keyword field of a holiday theme template entry are completely consistent with or contain the current holiday tag field, primary interest category, and specific interest preference, then mark the corresponding holiday theme template entry as a candidate holiday theme template. After iteration, select the holiday theme template entry with the highest priority from all candidate holiday theme templates as the target holiday theme template. The priority of the entries is determined according to the order of the entries in the holiday theme template library, with the holiday theme template entry at the beginning of the order having the highest priority. If there are no candidate holiday theme templates, select the default holiday theme template entry in the holiday theme template library as the target holiday theme template.

[0071] Among them, the holiday theme template library is a custom template library, which is set based on the advertising visual presentation needs of different holiday tags and user interests and preferences. It is used to match the most suitable target holiday theme template based on the user interest preferences identified by the holiday tag time window and the behavioral data within the window. For example, the Spring Festival tag combined with the interest in electronic products corresponds to a red festive electronic device theme template, and the Double 11 tag combined with the interest in beauty corresponds to a promotional discount beauty theme template.

[0072] Combine and encapsulate the target interactive structure with the target holiday theme template to generate a deployment resource package.

[0073] S4. Render the target interactive structure in the resource package, load the style resources corresponding to the target holiday theme, complete the ad style switch, collect the display feedback data generated after the ad style switch, use the display feedback data as user behavior data for the next switch, and add it to the user real-time behavior data to drive the next switch decision.

[0074] The rendering instruction set corresponding to the target interactive structure and the style resource descriptor list corresponding to the target holiday theme template are separated from the resource package. Specifically, the rendering instruction set corresponding to the target interactive structure field and the style resource descriptor list corresponding to the target holiday theme template field are directly read from the resource package as the separation result.

[0075] An ad rendering sandbox is created in the ad rendering client. The rendering instruction set is executed in the ad rendering sandbox, and the target interaction structure is instantiated and the interaction structure component is output. Instantiating the target interaction structure means creating a specific, interactive ad interface component instance in the ad rendering sandbox according to the rendering instruction set.

[0076] Specifically, the ad rendering client creates an independent ad rendering sandbox by calling the iframe element or Shadow DOM (Document Object Tree) technology provided by the browser. Specifically, it first generates an independent DOM container node, sets the DOM container node to isolation mode, restricts the DOM container node's access permissions to the main document of the media page, and allocates an independent event queue and style namespace to the DOM container node to ensure that the ad content runs within the DOM container node without affecting other content on the media page.

[0077] After creating an independent ad rendering sandbox, the separated rendering instruction set is passed into the independent ad rendering sandbox. After receiving the rendering instruction set, the independent ad rendering sandbox parses each rendering instruction in the rendering instruction set in turn. Each rendering instruction specifies the type, position, size and initial attributes of the ad interface component. When executing the rendering instruction set, the independent ad rendering sandbox creates the corresponding ad interface component instance according to the ad interface component type of each rendering instruction, such as creating an image component instance, a text component instance, a button component instance or a video component instance.

[0078] When instantiating the target interactive structure, the independent ad rendering sandbox assembles and connects all created ad interface component instances according to the hierarchical and positional relationships specified by the rendering instruction set to form a complete interactive ad interface component instance, which has the ability to respond to user clicks, swipes, or hovering interactions; after assembly, the complete interactive ad interface component instance is output as an interactive structure component.

[0079] Activate style resources through the list of style resource descriptors, inject style resources into the ad rendering sandbox of the interactive structural component, and complete the ad style switching.

[0080] Specifically, when activating a style resource through the style resource descriptor list, the ad rendering client iterates through each descriptor in the style resource descriptor list. Each descriptor contains a style type field, a style value field, and a target component identifier field.

[0081] Load the corresponding style resource based on the value of the style type field. For example, if the style type field is color, load the color value; if the style type field is font, load the font file link; if the style type field is background image, load the image URL; and if the style type field is animation, load the animation sequence description.

[0082] Once loaded, all activated style resources are integrated to form a style resource set. When the style resource set is injected into the ad rendering sandbox of the interactive structure component, the ad rendering sandbox receives the style resource set, iterates through all ad interface component instances in the interactive structure component, matches the corresponding style resource for each ad interface component instance based on the target component identifier field, and applies the matched style resource to the properties of the ad interface component instance. For example, color values ​​are applied to the background property, fonts are applied to the text property, background images are applied to the image property, and animation sequences are applied to the dynamic effect property.

[0083] Once all style resources are applied, the appearance and dynamic effects of the interactive structural components change accordingly, thus completing the advertisement style switch.

[0084] Once all style resources are applied, the appearance and dynamic effects of the interactive structural components change accordingly, thus completing the advertisement style switch.

[0085] An event listening pipeline is established in the ad rendering sandbox of the interactive structural components to capture user interaction behavior and generate the original interaction event stream.

[0086] When establishing an event listener pipeline in the ad rendering sandbox of the interactive structural component, the ad rendering sandbox registers an event listener for each ad interface component instance in the interactive structural component. The registered event types include click events, swipe events, hover events, and close events.

[0087] During registration, the ad rendering sandbox binds each event listener to the corresponding ad interface component instance and specifies that the listener records the event type, trigger timestamp, and trigger location when the event is triggered. When the user interacts, such as clicking a button component instance or sliding a carousel component instance, the ad rendering sandbox immediately captures the corresponding user interaction event through the bound event listener. The captured content includes the event type field, the event timestamp field, the event target component identifier field, and the event details field.

[0088] Each captured user interaction event is added sequentially to the original interaction event stream, which is a sequence of user interaction events arranged in order of the event timestamp field.

[0089] The original interactive event stream is formatted and associated with the target interactive structure and the target identifier of the target holiday theme template, and a display feedback data package is output. Specifically, each user interaction event in the original interactive event stream is converted into a standard format feedback record. Each feedback record includes an event type field, an event timestamp field, an event target component identifier field, an event detail field, a target interactive structure identifier field, and a target holiday theme template identifier field. All feedback records are then integrated into a display feedback data package.

[0090] The displayed feedback data packets are treated as user behavior data and appended to the user's real-time behavior data to drive the next ad style switching decision.

[0091] It should be noted that this step utilizes an independent ad rendering sandbox to render the target interactive structure and inject style resources, ensuring dynamic switching of ad styles within milliseconds. This significantly enhances the ad's festive atmosphere and visual appeal. Simultaneously, an event listening pipeline is established within the ad rendering sandbox to register click events, swipe events, hover events, etc., in real time. This captures user interaction behavior and generates raw interaction event streams. These streams are then formatted to form display feedback data packets, which are appended to the user's real-time behavior data, creating a closed-loop feedback loop. This strengthens the responsiveness and personalized experience of human-computer interaction, improves the efficiency of user interaction with ads, reduces distractions, promotes clicks and conversions, and drives the next ad style switching decision, achieving continuously optimized dynamic user profile ad delivery.

[0092] This embodiment also provides a computer device applicable to the context-aware dynamic user profile advertising style switching method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the context-aware dynamic user profile advertising style switching method proposed in the above embodiment.

[0093] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0094] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the context-aware dynamic user profile advertising style switching method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0095] In summary, this invention achieves structured enhancement of behavioral events and semantic classification in holiday scenarios by constructing a multi-layered dependency parse tree. It semantically fuses original behaviors such as page browsing, searching, clicking, and adding to cart within contextual dependencies, generating a regularized behavioral sequence with holiday intent recognition capabilities, and aligning it with the holiday context package to form a profile context package. The multi-layered dependency parse tree allows for precise characterization of user interests and preferences based on the behavioral evolution path within the holiday time window, avoiding misjudgments of interests caused by traditional behavioral statistics methods that ignore behavioral temporal dependencies and semantic associations. This provides a high-fidelity, scenario-sensitive decision-making basis for matching target interaction structures with holiday themes, significantly improving the consistency between ad style switching and the user's current intent.

[0096] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for switching dynamic user profile ad styles based on context awareness, characterized in that: include, Identify the context information of the ad request, determine the current valid holiday tag time window based on the ad request time in the context information, and generate a holiday context package; Acquire real-time user behavior data, perform session merging and behavior event categorization on the real-time user behavior data, generate standardized real-time user behavior data, associate the holiday context package with the standardized real-time user behavior data, and generate a profile context package. Based on the profile context package, the target interaction structure and target holiday theme are determined and encapsulated into a delivery resource package; Render the target interactive structure in the resource package, load the style resources corresponding to the target holiday theme, complete the ad style switching, collect the display feedback data generated after the ad style switching, and use the display feedback data as the real-time user behavior data input for the next switching to drive the next switching decision.

2. The context-aware dynamic user profile ad style switching method as described in claim 1, characterized in that: The generation of the holiday context package specifically involves, Extract the ad request time from the context information; Match the ad request time with the holiday calendar database to determine the current valid holiday tag time window; The holiday tag time window is merged with the ad placement specifications and media page type in the context information to generate a holiday context package.

3. The context-aware dynamic user profile ad style switching method as described in claim 2, characterized in that: The determination of the currently valid holiday tag time window is specifically as follows: Convert ad request times to a unified time base; Retrieve festival records with a unified time base from the festival calendar database; Filter the holiday event records that are active, and use them as the current valid holiday tag time window.

4. The context-aware dynamic user profile ad style switching method as described in claim 1, characterized in that: The generation of standardized real-time user behavior data specifically includes: Acquire real-time user behavior data, perform a merge operation on the real-time user behavior data according to the session identifier, and output the session-level behavior sequence; Perform semantic classification on behavioral events in a session-level behavioral sequence to generate a set of classified behavioral events; The categorized behavioral event sets are reorganized in chronological order to generate well-organized real-time user behavior data.

5. The context-aware dynamic user profile ad style switching method as described in claim 1, characterized in that: The generated image context package is specifically... Extract the holiday tag time window and interaction structure constraint information from the holiday context package; Based on the holiday tag time window, the regularized real-time user behavior data is aligned with the time window to obtain the behavior data within the window; The behavioral data within the window is fused with the interaction structure constraint information to generate a profile context package.

6. The context-aware dynamic user profile ad style switching method as described in claim 4, characterized in that: The generated set of categorized behavioral events specifically refers to... Construct a multi-layer dependency parsing tree for behavior events, perform cross-layer semantic propagation on behavior events on the multi-layer dependency parsing tree, and generate enhanced semantic tags; The enhanced semantic tags are mapped to predefined behavior category rules for classification, generating a set of classified behavior events.

7. The context-aware dynamic user profile ad style switching method as described in claim 4, characterized in that: The session identifier is a unique tracking identifier obtained from the context information of the advertising request, and is generated and transmitted by the media terminal at the start of the user session.

8. The context-aware dynamic user profile ad style switching method as described in claim 5, characterized in that: The interaction structure constraint information is derived from the ad slot specifications and media page type, and is a set of rules that limit the currently available ad interaction forms.

9. The context-aware dynamic user profile ad style switching method as described in claim 1, characterized in that: The formation of the resource delivery package specifically refers to... Extract in-window behavior data and interaction structure constraint information from the portrait context package; By analyzing user behavior patterns within the holiday time window using in-window behavioral data, the holiday phase status can be obtained. Based on the festival stage status and interaction structure constraints, the target interaction structure is matched from a predefined set of interaction structures; Use in-window behavioral data to identify user interests and preferences, and match target holiday theme templates from a holiday theme template library; Combine and encapsulate the target interactive structure with the target holiday theme template to generate a deployment resource package.

10. The context-aware dynamic user profile ad style switching method as described in claim 1, characterized in that: The display feedback data generated after the recycling ad style is switched is specifically as follows: Separate the rendering instruction set corresponding to the target interactive structure and the style resource descriptor list corresponding to the target holiday theme template from the resource package; Create an ad rendering sandbox in the ad rendering client, execute the rendering instruction set in the ad rendering sandbox, instantiate the target interaction structure, and output the interaction structure component; The instantiated target interaction structure refers to the creation of specific, interactive advertising interface component instances in the advertising rendering sandbox based on the rendering instruction set. Activate style resources through the list of style resource descriptors, inject style resources into the ad rendering sandbox of interactive structural components, and complete the ad style switching; An event listening pipeline is established in the ad rendering sandbox of the interactive structural components to capture user interaction behavior and generate the original interactive event stream; The original interactive event stream is formatted and associated with the target interactive structure and the target identifier of the target holiday theme template, and the output display feedback data package is generated.

Images