Webpage element point burying method and device and electronic equipment
By receiving user click operation signals and using the buried point recommendation model to bury the web element, the problems of large workload and low performance caused by manual code addition in the existing technology are solved, and more accurate buried point recommendation and efficient buried point Click to process.
Patent Information
- Application Number
- CN202410991999.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-23
- Publication Date
- 2024-11-12
AI Technical Summary
In the existing web element burying technology, manual code addition is required to cause developers to work hard and may experience invalid burying, affecting website performance.
By receiving user click operation signals, creating web page element tags, and using pre-trained buried point recommendation model to perform buried point prediction processing on target web page elements, determine whether to perform buried point processing, and inject buried point code using document object model.
It reduces the workload of developers, improves the efficiency of burying points and website performance, and reduces the occurrence of invalid burying points.
Smart Images

Figure CN118939548A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to a web page element embedding method, device and electronic device. Background Art
[0002] In the modern Internet era, web page optimization has become a key factor in website operation and user experience. As a common data collection and analysis method, web page element tracking aims to determine which web page elements are most needed to be clicked or observed, as well as the location and frequency of these elements on the page. The ultimate goal of web page element tracking is to optimize the user experience and increase the conversion rate of the website.
[0003] Existing web page element tracking technology solutions mainly include code tracking, full tracking and visual tracking. Taking code tracking as an example, code tracking requires developers to manually analyze the interaction between web page elements and users, and then insert specific tracking codes into the code corresponding to the web page elements. Whenever a user interacts with a web page element, the tracking code will collect and report relevant behavior data.
[0004] However, in the existing method, each web page element that needs to be tracked needs to be manually added with code, which not only increases the workload of developers, but also may cause a large number of invalid tracking points (i.e., waste points) in the code. These waste points not only occupy resources, but also may cause the performance of the website to deteriorate. Summary of the invention
[0005] In view of this, the present application provides a web page element tracking method, device and electronic device, the main purpose of which is to solve the current technical problem that in code tracking, each web page element that needs to be tracked needs to be manually added with code, which not only increases the workload of developers, but also may cause a large number of invalid tracking points (i.e., waste points) in the code. These waste points not only take up resources, but also may cause reduced website performance.
[0006] According to a first aspect of the present disclosure, a method for embedding elements in a web page is provided, the method comprising:
[0007] Receiving a click operation signal from a user, and creating a web page element tag in response to the click operation signal, wherein the web page element tag is used to identify a target web page element related to the user click;
[0008] Using the web page element tag to cover the target web page element, and using the pre-trained embedding recommendation model to perform embedding prediction processing on the covered target web page element to obtain element recommendation information, wherein the element recommendation information includes element type and embedding attribute;
[0009] Determining whether to perform tracking processing on the target web page element according to the element type and the tracking attribute;
[0010] If the target web page element is to be subjected to embedding processing, embedding code is injected into the target web page element using the document object model, and the target web page element is marked to complete the embedding processing of the target web page element.
[0011] According to a second aspect of the present disclosure, a web page element embedding device is provided, the device comprising:
[0012] A creation module, used for receiving a click operation signal from a user, and creating a web page element tag in response to the click operation signal, wherein the web page element tag is used for identifying a target web page element related to the user click;
[0013] A prediction module is used to cover the target web page element with the web page element tag, and use the pre-trained embedding recommendation model to perform embedding prediction processing on the covered target web page element to obtain element recommendation information, wherein the element recommendation information includes element type and embedding attribute;
[0014] A determination module, used to determine whether to perform tracking processing on the target web page element according to the element type and the tracking attribute;
[0015] The processing module is used for injecting the embedding code into the target web page element by using the document object model and marking the target web page element to complete the embedding processing of the target web page element.
[0016] According to the third aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method of the aforementioned first aspect.
[0017] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable a computer to execute the method of the aforementioned first aspect.
[0018] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program implements the method of the aforementioned first aspect when executed by a processor.
[0019] The web page element tracking method, device and electronic device provided by the present invention, compared with the prior art, receive a user's click operation signal, and create a web page element label in response to the click operation signal, the web page element label is used to identify the target web page element related to the user's click; use the web page element label to cover the target web page element, and use a pre-trained tracking recommendation model to perform tracking prediction processing on the covered target web page element to obtain element recommendation information, the element recommendation information includes element type and tracking attributes; determine whether to perform tracking processing on the target web page element according to the element type and tracking attributes; if the target web page element is to be tracked, use the document object model to inject tracking code into the target web page element, and mark the target web page element to complete the tracking processing of the target web page element. For the disclosed scheme, a tracking point recommendation model is used to perform tracking point prediction processing on the covered target web page elements to obtain element recommendation information, and determine whether to perform tracking point processing on the target web page elements based on the element recommendation information; if the target web page elements are to be processed, the document object model is used to inject tracking point code into the target web page elements. This method replaces the traditional tracking point method that requires R&D personnel to manually write code on the web page, and provides more accurate tracking point recommendation information, thereby greatly reducing the difficulty of operation, reducing the workload of developers, and improving website performance and tracking point efficiency.
[0020] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.
[0022] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0023] Figure 1 A schematic diagram of a flow chart of a web page element tracking method provided by an embodiment of the present disclosure;
[0024] Figure 2 A schematic diagram of a flow chart of a web page element tracking method provided by an embodiment of the present disclosure;
[0025] Figure 3A schematic diagram of a flow chart of a web page element tracking method provided by an embodiment of the present disclosure;
[0026] Figure 4 A schematic diagram of the structure of a web page element embedding device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION
[0027] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description. It should be noted that, in the absence of conflict, the embodiments of the present disclosure and the features in the embodiments may be combined with each other.
[0028] The following describes the web page element embedding method, device and electronic device of the embodiments of the present disclosure with reference to the accompanying drawings.
[0029] In order to effectively solve the problem that in the current code tracking, each web page element that needs to be tracked needs to be manually added with code, which increases the workload of developers and may also cause the performance of the website to deteriorate. This embodiment provides a web page element tracking method, such as Figure 1 As shown, the method includes:
[0030] Step 101: Receive a click operation signal from a user, and create a web page element tag in response to the click operation signal, where the web page element tag is used to identify a target web page element related to the user click.
[0031] Among them, web page elements can be the various parts and components that make up a web page. In web page design and development, each visible or interactive component can be called a web page element. Web page elements can be defined by HTML (Hypertext Markup Language), styled by CSS (Cascading Style Sheets), and interactive functions implemented by scripting languages such as JavaScript.
[0032] Web page elements may include, but are not limited to, text, images, links, forms, buttons, media, navigation bars, layout containers, etc.
[0033] The target web page element is the web page element that can be directly related to the user's click action.
[0034] For the disclosed embodiment, a new button may be added to the product webpage, and the new button may allow the user to perform a new operation on the product webpage. The new button may be fixed to a corner of the screen by default, such as the upper corner or the lower corner of the screen, and no specific limitation is made to this. The new button is not affected by page sliding, so that the user can access it at any time when browsing the page content.
[0035] The visual style of the new button can be displayed using the "" icon. In actual applications, the visual style of the new button can be adjusted as needed, such as changing the color, size, shape, etc. It can also be integrated with other function buttons to form a button group to provide users with multiple operation options and ensure that the interactive information between buttons is clear to avoid confusion.
[0036] When the user clicks the New button, a click operation signal is sent to the product webpage, and the product webpage can pop up a New Tag window in response to the click operation signal. The New Tag window allows the user to enter or select relevant information to create a new tag (i.e., a webpage element tag). The webpage element tag can be used to identify the user click location or the target webpage element related to the user click.
[0037] Step 102: Use web page element tags to cover target web page elements, and use a pre-trained embedding recommendation model to perform embedding prediction processing on the covered target web page elements to obtain element recommendation information, which includes element type and embedding attributes.
[0038] For the disclosed embodiment, after creating a new web page element tag, the user can drag the web page element tag to the relevant element area in the web page by dragging the mouse, and the currently covered element attribute needs to be highlighted with a border to prompt the user (this is only a recommended solution and can be changed according to actual conditions in actual applications). After completing the drag and releasing the mouse button, the element attribute of the target web page element (such as a button, picture, text box, etc.) at the mouse position can be automatically identified, and according to the identified element attribute, a window or pop-up box is displayed to display the element embedding description of the target web page element.
[0039] In actual operation, the display content of the display window or pop-up box can be adjusted according to the needs of the developer, so that the display window or pop-up box can display text, charts, links, buttons or any other relevant information and controls, so that users can better understand the purpose and operation method of the tracking point.
[0040] For the embodiments of the present disclosure, in order to display the element tracking description of the target web page elements, a pre-trained tracking recommendation model can be used to perform tracking prediction processing on the covered target web page elements to obtain element recommendation information, which may include element type and tracking attributes.
[0041] Among them, the element type can represent the function or content of the element on the page; the embedding attribute can be specific information related to the embedding, such as user click events, scrolling events, dwell time, etc.; the embedding recommendation model can be used to identify and recommend the most valuable web page elements for embedding.
[0042] In actual application scenarios, as a possible implementation method, the embedding recommendation model can receive the positioning information of the target web page element on the current web page. The positioning information may include the position, size, content, etc. of the target web page element.
[0043] The embedding recommendation model can use a pre-trained training set, combined with the element features of the target web page elements, to predict how users may interact with the target web page elements, as well as possible user behavior data of the interaction methods, and output element recommendation information, where the element recommendation information can be a list of the most valuable web page elements predicted by the embedding recommendation model. The web page element list can contain multiple web page elements, and each recommended web page element can include an element type and embedding attributes.
[0044] Step 103: determine whether to perform embedding processing on the target web page element according to the element type and embedding attribute. If the target web page element is to be embedded, inject embedding code into the target web page element using the document object model and mark the target web page element to complete the embedding processing on the target web page element.
[0045] After the user views the window of element tracking instructions, that is, based on the element type and tracking attributes, the user decides whether to perform tracking on the target web page element. If the user decides to perform tracking on the target web page element, the user clicks the confirmation button, where the confirmation button can be used to submit the user's decision and trigger the next step.
[0046] When the user clicks the confirmation button, the document object model (DOM) can be used to modify the source code of the web page and inject the necessary tracking code to implement the tracking function, so that every time the user interacts with the target web page elements, the system will automatically collect relevant user behavior data.
[0047] After injecting the tracking code, the system can close the window displaying the element tracking instructions and visually mark the target web page elements that have been tracked on the web page so that users can identify which elements have been set with tracking.
[0048] After the user views the window of element tracking instructions, if he does not decide to track the target web page element, he can click the Cancel button. The system can abandon the operation of injecting the tracking code, close the window, and restore the user interface to the position where the user originally created the new web page element label, so that the user can re-select other elements for tracking.
[0049] For the embodiment of the present disclosure, the specific solution process is as follows Figure 2 shown.
[0050] In summary, compared with the prior art, the web page element tracking method disclosed in the present invention receives a user's click operation signal, creates a web page element tag in response to the click operation signal, and the web page element tag is used to identify the target web page element related to the user's click; the target web page element is covered with the web page element tag, and the covered target web page element is predicted and processed by a pre-trained tracking recommendation model to obtain element recommendation information, and the element recommendation information includes element type and tracking attributes; according to the element type and tracking attributes, it is determined whether to perform tracking processing on the target web page element; if the target web page element is to be tracked, the target web page element is injected with tracking code by using the document object model, and the target web page element is marked to complete the tracking processing of the target web page element. For the disclosed scheme, a tracking point recommendation model is used to perform tracking point prediction processing on the covered target web page elements to obtain element recommendation information, and determine whether to perform tracking point processing on the target web page elements based on the element recommendation information; if the target web page elements are to be processed, the document object model is used to inject tracking point code into the target web page elements. This method replaces the traditional tracking point method that requires R&D personnel to manually write code on the web page, and provides more accurate tracking point recommendation information, thereby greatly reducing the difficulty of operation, reducing the workload of developers, and improving website performance and tracking point efficiency.
[0051] Further, as a refinement and extension of the above embodiment, in order to fully illustrate the specific implementation process of the method of this embodiment, this embodiment provides the following Figure 3 The specific method shown comprises the following steps:
[0052] Step 201: Obtain historical full-point data of multiple websites through a full-point method, and use the historical full-point data to build a feature model; use the feature model to extract features from the historical full-point data to obtain a historical feature vector; perform vector operation processing on the historical feature vector to obtain a prediction result, so as to generate historical element recommendation information based on the prediction result.
[0053] Among them, the historical full-point data is the detailed data information of the user's interaction with the website elements within the historical time period, for example, the user's various behaviors on the web page, such as clicks, scrolling, and dwell time.
[0054] For the embodiments of the present disclosure, as a possible implementation method, when using deep learning to analyze and process tracking information, it is first necessary to collect a large amount of tracking configuration information. The tracking configuration information can obtain tracking data and user feedback data of multiple websites in a full tracking manner as initial data for training. Among them, the tracking data can be automatically captured and recorded on the entire website. All user behavior data may include user clicks, scrolling, dwell time, etc.
[0055] The collected data can be cleaned and preprocessed, including removing duplicate data, processing missing values, standardizing data, etc., and the processed data can be quantified and weighted to establish a preliminary feature model.
[0056] Among them, data quantization can be to convert continuous data into discrete values for model processing; weight configuration can be to assign different weight values according to the importance of different data features, so that the model can pay more attention to the data features that have a greater impact on the prediction results. The weight value can be a value between 0 and 1, or other forms of measurement, such as standardized scores, which are not specifically limited. Feature models can be used to extract data features from buried point data and conduct further analysis and processing.
[0057] Specifically, the historical full-point data mainly involves three feature models: pageview (element browsing model), weboperate (element operation model) and webstay (viewing area stay model).
[0058] Among them, the element browsing model can focus on the user's browsing behavior on the web page. The characteristic values of the element browsing model may include: is_first_time (i.e., whether the user is visiting the page for the first time), title (i.e., the title of the web page), url_path (i.e., the URL path of the web page), referrer_host (i.e., the domain name of the reference link), referrer (i.e., the full URL of the reference link), and utm_campaign (i.e., parameters used to track the source of marketing activities).
[0059] The element operation model may focus on the interactive operations between users and web page elements. The characteristic values of the element operation model may include: element_id (i.e., element ID), element_content (i.e., element content), element_class_name (i.e., element class name), element_type (i.e., element type, such as button, text box, etc.), element_selector (i.e., CSS selector for selecting elements).
[0060] The viewport dwell model may focus on the user's dwell time in a specific viewport. The characteristic values of the viewport dwell model may include: viewport_width (i.e., the width of the viewport), viewport_position (i.e., the position of the viewport), viewport_height (i.e., the height of the viewport), and event_duration (i.e., the duration of an event, such as the dwell time of a user on a certain element).
[0061] As a possible implementation method, the historical full-point data can be standardized. First, a historical full-point data vector M can be defined. The historical full-point data vector consists of three sub-models, namely the element browsing model (pageview, pv), the element operation model (weboperate, wp) and the view area stay model (webstay, ws)
[0062] That is, M = P{pv,wp,ws}
[0063] Next, the user operation data obtained from the full embedding method can be used to convert each feature attribute into a numerical value, and finally a embedding feature vector value M is formed according to the element attributes in the website. Multiple embedding feature vector values M are combined to form a standardized data set D. The data set D contains multiple embedding feature vectors {M1, M2, ..., Mi}, where each vector Mi represents the user behavior characteristics of different web page elements, that is, D = {M1, M2, ..., M i}
[0064] In the convolution kernel calculation, since different features contribute differently to the model, it is necessary to assign a weight coefficient to each feature. The weight coefficient can be used as a measure of the importance of the feature and is used to adjust the contribution of the feature in the calculation.
[0065] The weight coefficients of the three feature models (i.e., element browsing model pv, element operation model wp, and viewport stay model ws) can be counted according to the user's operation data on the DOM elements, and finally the weight value ω={ω pv ,ω wp ,ω ws}.
[0066] When building a tracking point recommendation model, you can use multiple convolutional layers for feature extraction and structured modeling to achieve element type prediction and automatic generation of tracking point attributes. The specific steps are as follows:
[0067] When building a buried point recommendation model, the convolutional layer can be used to obtain a large amount of buried point data D = {M1, M2, ..., M i}, extract eigenvalues from them, and generate corresponding feature maps.
[0068] The shape of each sub-feature in Mi can be defined based on the feature model obtained previously. For example, for the element browsing model (PV), the shape is set to (C pv ,H pv W pv ), where C pv Indicates the number of channels (can be 6), H pv represents the length of the time series, W pv Indicates the number of fields (can be 6).
[0069] Similarly, for the element operation model (WP), set the shape to (C wp ,,H wp W wp ), where C wp Indicates the number of channels (can be 5), H wp Indicates the length of the element operation, W wp Indicates the number of fields (can be 5).
[0070] Similarly, for the viewport stay model (WS), set the shape to (C wa ,H ws W ws ), where C ws Indicates the number of channels (can be 4), H ws Indicates the length of the window height sequence, W wp Indicates the number of fields (can be 4).
[0071] Define the size and number of filters. For example, you can set the filter size to a 3*3 convolution kernel. You can set the convolution kernel for each feature based on the weight coefficients obtained previously.
[0072] A convolution operation is performed on each feature. For example, for the element browsing model (PV), m 6-channel filters with shape (m, 6, 6) can be performed, and the shape of the output feature map is (m, Hpv, Wpv).
[0073] For each sub-feature map, the corresponding output feature map is concatenated by channel. The final output feature map can combine the information of all sub-features and the shape can be:
[0074] (m*3,min(Hpv,Hws,Hwp),min(Wpv,Wws,Wwp))
[0075] Then, the pooling layer can be used to further process the previously acquired feature map, reduce the size of the feature map, extract the most significant features, and improve the calculation rate of the model.
[0076]
[0077] y i,j =max(x (2i,2j) ,x (2i+1,2j) ,x (2i,2j+1) ,x (2i+1,2j+1) )
[0078] In the formula, y i,j It can represent the new matrix obtained by extracting the maximum value of the two-dimensional matrix in the (f*g)(t) eigenfunction.
[0079] Then, the fully connected layer can be used to flatten the feature map output by the previous pooling layer into a one-dimensional vector, and pass it through a series of fully connected layers to perform linear classification, regression and other tasks. The fully connected layer can learn higher-level feature combinations, thereby improving the expressiveness of the model.
[0080] In the fully connected layer, each neuron has a weight matrix WX = [w1…wn] and a bias matrix b = [b1…bn]. The weight matrix determines the relationship between the input features and the output features, while the bias matrix represents the offset of each neuron, as shown in the following formula, where x represents the input pooling layer and f represents the activation function (usually a nonlinear activation function such as ReLU).
[0081] y=f(W x +b)
[0082] Through the above steps, the convolutional neural network can extract useful features from the embedded data corresponding to the input element attributes, and use these features for classification and identification. As the model deepens, each step increases its expressiveness and performance, and eventually forms a embedded recommendation model that can recommend the most valuable DOM element set and its characteristic attributes.
[0083] Accordingly, the feature model is constructed using the historical full-point data, which may include:
[0084] The buried point features of the historical full buried point data are extracted and converted into numerical values. The buried point features after numerical conversion are weighted and configured to construct a feature model according to the buried point features after weight configuration. The feature model includes an element browsing model, an element operation model and a view area dwell model.
[0085] Accordingly, the feature model is used to extract features from the historical full-buried point data to obtain historical feature vectors, which may specifically include:
[0086] The feature model is used to extract features from the historical full-point data to obtain element browsing features, element operation features, and view area dwell features, and generate feature maps of the corresponding features; convolution operations are performed on each feature map to obtain a historical feature map; and pooling is performed on the historical feature map to obtain a historical feature vector.
[0087] Accordingly, a convolution operation is performed on each feature map to obtain a historical feature map, which may specifically include:
[0088] According to the feature map shape and the number of convolution kernels corresponding to each feature map, a convolution operation is performed on each feature map to obtain an output feature map corresponding to each feature map; each output feature map is spliced according to the channel to obtain a historical feature map.
[0089] Accordingly, vector operations are performed on the historical feature vectors to obtain prediction results, which may specifically include:
[0090] Perform linear transformation operations on historical feature vectors according to the weight matrix and the bias value matrix; perform activation function processing on the linear transformation results to obtain the prediction results.
[0091] Step 202: Use the historical feature vector as the input feature of the embedding recommendation model, use the historical element recommendation information as the training label of the embedding recommendation model, iteratively train the embedding recommendation model until the loss function value of the embedding recommendation model is less than the preset loss function threshold, and determine that the embedding recommendation model training is completed.
[0092] For the disclosed embodiment, a historical feature vector may be extracted from the acquired historical full-point data, and the historical feature vector may be used to reflect detailed information on the user's interaction with the web page elements.
[0093] Historical element recommendation information can be generated based on historical feature vectors. The historical feature vectors can be used as input features of the embedding recommendation model, and the historical element recommendation information can be used as training labels to iteratively train the embedding recommendation model, and continuously adjust the model parameters of the embedding recommendation model until the loss function value of the embedding recommendation model is less than the preset loss function threshold, and the embedding recommendation model training is determined to be completed.
[0094] The trained tracking recommendation model can be evaluated using a data set that was not involved in the model training to ensure that the tracking recommendation model can accurately predict or classify new data. The trained tracking recommendation model can be applied to new tracking data to predict or classify the results based on the input features, so that the document object model can implement tracking based on the model prediction results.
[0095] In summary, compared with the prior art, the web page element tracking method disclosed in the present invention receives a user's click operation signal, creates a web page element tag in response to the click operation signal, and the web page element tag is used to identify the target web page element related to the user's click; the target web page element is covered with the web page element tag, and the covered target web page element is predicted and processed by a pre-trained tracking recommendation model to obtain element recommendation information, and the element recommendation information includes element type and tracking attributes; according to the element type and tracking attributes, it is determined whether to perform tracking processing on the target web page element; if the target web page element is to be tracked, the target web page element is injected with tracking code by using the document object model, and the target web page element is marked to complete the tracking processing of the target web page element. For the disclosed scheme, a tracking point recommendation model is used to perform tracking point prediction processing on the covered target web page elements to obtain element recommendation information, and determine whether to perform tracking point processing on the target web page elements based on the element recommendation information; if the target web page elements are to be processed, the document object model is used to inject tracking point code into the target web page elements. This method replaces the traditional tracking point method that requires R&D personnel to manually write code on the web page, and provides more accurate tracking point recommendation information, thereby greatly reducing the difficulty of operation, reducing the workload of developers, and improving website performance and tracking point efficiency.
[0096] Based on the above Figure 1 and Figure 3 The present disclosure provides a web page element embedding device, such as Figure 4 As shown, the device includes: a creation module 31, a prediction module 32, a determination module 33 and a processing module 34;
[0097] A creation module 31, configured to receive a click operation signal from a user, and create a web page element tag in response to the click operation signal, wherein the web page element tag is used to identify a target web page element related to the user click;
[0098] A prediction module 32 is used to cover the target web page element with the web page element tag, and use the pre-trained embedding recommendation model to perform embedding prediction processing on the covered target web page element to obtain element recommendation information, wherein the element recommendation information includes element type and embedding attribute;
[0099] A determination module 33, used to determine whether to perform a point embedding process on the target web page element according to the element type and the point embedding attribute;
[0100] The processing module 34 is used for injecting the embedding code into the target web page element by using the document object model and marking the target web page element to complete the embedding process for the target web page element if embedding process is performed on the target web page element.
[0101] In a specific application scenario, the prediction module 32 can be used to obtain historical full-point data of multiple websites in a full-point manner, and use the historical full-point data to build a feature model, where the historical full-point data is detailed data information on the interaction between users and website elements in a historical time period;
[0102] Using the feature model to extract features from the historical all-buried-point data to obtain a historical feature vector;
[0103] Performing vector operation processing on the historical feature vector to obtain a prediction result, so as to generate historical element recommendation information according to the prediction result;
[0104] The historical feature vector is used as the input feature of the embedding point recommendation model, and the historical element recommendation information is used as the training label of the embedding point recommendation model. The embedding point recommendation model is iteratively trained until the loss function value of the embedding point recommendation model is less than the preset loss function threshold, and it is determined that the training of the embedding point recommendation model is completed.
[0105] In a specific application scenario, the prediction module 32 can be used to extract features from the historical full-buried point data using the feature model, obtain element browsing features, element operation features, and visual area stay features, and generate feature maps of corresponding features;
[0106] Performing a convolution operation on each of the feature maps to obtain a historical feature map;
[0107] Pooling is performed on the historical feature map to obtain a historical feature vector.
[0108] In a specific application scenario, the prediction module 32 may be used to perform a convolution operation on each of the feature maps according to the feature map shape and the number of convolution kernels corresponding to each of the feature maps, to obtain an output feature map corresponding to each of the feature maps;
[0109] Each of the output feature maps is spliced according to the channel to obtain a historical feature map.
[0110] In a specific application scenario, the prediction module 32 may be used to perform a linear transformation operation on the historical feature vector according to a weight matrix and a deviation value matrix;
[0111] The linear transformation result is processed by activation function to obtain the prediction result.
[0112] It should be noted that for other corresponding descriptions of the functional units involved in the web page element embedding device provided by the present disclosure, please refer to Figure 1 and Figure 3 The corresponding description in will not be repeated here.
[0113] Based on the above Figure 1 and Figure 3 The method shown in the present disclosure, accordingly, also provides a computer-readable storage medium having a computer program stored thereon, which implements the above-mentioned Figure 1 and Figure 3 The method shown.
[0114] Based on this understanding, the technical solution of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of various implementation scenarios of the present disclosure.
[0115] Based on the above Figure 1 and Figure 3 The method shown, and Figure 4 In order to achieve the above-mentioned purpose, the embodiment of the present disclosure also provides an electronic device that can be configured on the vehicle (such as an electric vehicle) side, and the device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 and Figure 3 The method shown.
[0116] Optionally, the above-mentioned physical device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.
[0117] Those skilled in the art will appreciate that the above-mentioned physical device structure provided by the present disclosure does not constitute a limitation on the physical device, and may include more or fewer components, or a combination of certain components, or different arrangements of components.
[0118] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the above-mentioned physical device, and supports the operation of the information processing program and other software and / or programs. The network communication module is used to realize the communication between the components inside the storage medium, and the communication with other hardware and software in the information processing physical device.
[0119] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present disclosure can be implemented by means of software plus necessary general hardware platforms, or by hardware. According to the web page element embedding method, device and electronic device provided by the present disclosure, compared with the prior art, the present disclosure receives a user's click operation signal and creates a web page element tag in response to the click operation signal. The web page element tag is used to identify the target web page element related to the user's click; the target web page element is covered with the web page element tag, and the embedded point recommendation model trained in advance is used to perform embedding prediction processing on the covered target web page element to obtain element recommendation information, and the element recommendation information includes element type and embedded point attributes; according to the element type and embedded point attributes, it is determined whether to perform embedded point processing on the target web page element; if the target web page element is to be embedded, the document object model is used to inject the embedded point code into the target web page element, and the target web page element is marked to complete the embedded point processing on the target web page element. For the disclosed scheme, a tracking point recommendation model is used to perform tracking point prediction processing on the covered target web page elements to obtain element recommendation information, and determine whether to perform tracking point processing on the target web page elements based on the element recommendation information; if the target web page elements are to be processed, the document object model is used to inject tracking point code into the target web page elements. This method replaces the traditional tracking point method that requires R&D personnel to manually write code on the web page, and provides more accurate tracking point recommendation information, thereby greatly reducing the difficulty of operation, reducing the workload of developers, and improving website performance and tracking point efficiency.
[0120] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.
[0121] The above are only specific embodiments of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to these embodiments herein, but will conform to the widest scope consistent with the principles and novel features applied herein.
Claims
1. A method for embedding elements in a web page, characterized in that: The method comprises: Receiving a click operation signal from a user, and creating a web page element tag in response to the click operation signal, wherein the web page element tag is used to identify a target web page element related to the user click; Using the web page element tag to cover the target web page element, and using the pre-trained embedding recommendation model to perform embedding prediction processing on the covered target web page element to obtain element recommendation information, wherein the element recommendation information includes element type and embedding attribute; Determining whether to perform tracking processing on the target web page element according to the element type and the tracking attribute; If the target web page element is to be subjected to embedding processing, embedding code is injected into the target web page element using the document object model, and the target web page element is marked to complete the embedding processing of the target web page element.
2. The method according to claim 1, characterized in that: The training process of the embedding recommendation model includes: Obtain historical full-point data of multiple websites through full-point method, and use the historical full-point data to build a feature model, where the historical full-point data is detailed data information on the interaction between users and website elements in a historical time period; Using the feature model to extract features from the historical all-buried-point data to obtain a historical feature vector; Performing vector operation processing on the historical feature vector to obtain a prediction result, so as to generate historical element recommendation information according to the prediction result; The historical feature vector is used as the input feature of the embedding point recommendation model, and the historical element recommendation information is used as the training label of the embedding point recommendation model. The embedding point recommendation model is iteratively trained until the loss function value of the embedding point recommendation model is less than the preset loss function threshold, and it is determined that the training of the embedding point recommendation model is completed.
3. The method according to claim 2, characterized in that The method of constructing a feature model using the historical all-embedded point data includes: Extracting the buried point features of the historical full buried point data, and converting the buried point features into numerical values; The buried point features after numerical conversion are weighted and configured to construct a feature model according to the buried point features after weight configuration, wherein the feature model includes an element browsing model, an element operation model and a view area stay model.
4. The method according to claim 2, characterized in that: The method of extracting features from the historical all-buried point data using the feature model to obtain a historical feature vector includes: The feature model is used to extract features from the historical full-buried point data to obtain element browsing features, element operation features, and visual area stay features, and a feature map of corresponding features is generated; Performing a convolution operation on each of the feature maps to obtain a historical feature map; Pooling is performed on the historical feature map to obtain a historical feature vector.
5. The method according to claim 4, characterized in that The performing a convolution operation on each of the feature maps to obtain a historical feature map includes: According to the feature map shape and the number of convolution kernels corresponding to each feature map, a convolution operation is performed on each feature map to obtain an output feature map corresponding to each feature map; Each of the output feature maps is spliced according to the channel to obtain a historical feature map.
6. The method according to claim 2, characterized in that The performing vector operation processing on the historical feature vector to obtain a prediction result includes: Performing a linear transformation operation on the historical feature vector according to a weight matrix and a deviation value matrix; The linear transformation result is processed by activation function to obtain the prediction result.
7. A web page element embedding device, the device comprising: A creation module, used for receiving a click operation signal from a user, and creating a web page element tag in response to the click operation signal, wherein the web page element tag is used for identifying a target web page element related to the user click; A prediction module is used to cover the target web page element with the web page element tag, and use the pre-trained embedding recommendation model to perform embedding prediction processing on the covered target web page element to obtain element recommendation information, wherein the element recommendation information includes element type and embedding attribute; A determination module, used to determine whether to perform tracking processing on the target web page element according to the element type and the tracking attribute; The processing module is used for injecting the embedding code into the target web page element by using the document object model and marking the target web page element to complete the embedding processing of the target web page element.
8. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.
9. A non-transitory computer-readable storage medium having computer instructions stored thereon, characterized in that: The computer instructions are used to make a computer execute the method as claimed in claims 1-6.
10. A computer program product, characterized in that The invention comprises a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 6.
Citation Information
Cited By
Recommendation method and system based on user product use behavior
CN120144751A
Artificial intelligence-based front-end burying point recommendation method and system
CN122153169A
