Page recommendation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a page recommendation method and device, electronic equipment and a storage medium. The method comprises the steps that keyword information input by a user is acquired, and the information type of the keyword information is determined; wherein the keyword information represents the demand of the user on the page to be pushed, and the information type represents the expression form of the keyword information; determining a page set according to the keyword information and the information type of the keyword information; wherein the page set comprises a plurality of pages related to the keyword information; acquiring burying point information of each page in the page set; wherein the buried point information represents the situation that the page is consumed, wherein the situation is collected through a buried point preset in the page; and according to the burying point information of each page in the page set, determining a target page from each page, and pushing the target page to the user. According to the method, page recommendation is performed according to user requirements, and the page creation efficiency and precision are improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, electronic device, and storage medium for recommending pages. Background Technology

[0002] Content creation and publishing platforms typically provide users with page editors. Users can add elements or information to pages according to their actual business needs, thereby creating and publishing pages. When consumers browse the pages, they can participate in related activities based on the information displayed.

[0003] However, the current page creation process requires a lot of manual intervention, resulting in inconsistent page quality and low page creation efficiency, which affects the promotion of the campaign. Summary of the Invention

[0004] This application provides a method, apparatus, electronic device, and storage medium for recommending pages, which enables automatic page recommendations based on user needs, facilitating users to create new pages and improving user experience.

[0005] In a first aspect, embodiments of this application provide a method for recommending pages, including:

[0006] Obtain keyword information input by the user and determine the information type of the keyword information; wherein, the keyword information represents the user's demand for the page to be pushed, and the information type represents the presentation form of the keyword information;

[0007] Based on the keyword information and the information type of the keyword information, a page set is determined; wherein, the page set includes multiple pages related to the keyword information;

[0008] Obtain the event tracking information for each page in the page set; wherein, the event tracking information represents the consumption of the page by means of the event tracking information preset in the page.

[0009] Based on the tracking information of each page in the page set, a target page is determined from the pages and pushed to the user; wherein, the target page is used to assist the user in creating a new page.

[0010] Secondly, embodiments of this application provide a page recommendation device, comprising:

[0011] An information acquisition unit is used to acquire keyword information input by the user and determine the information type of the keyword information; wherein, the keyword information represents the user's demand for the page to be pushed, and the information type represents the presentation form of the keyword information;

[0012] A collection determining unit is configured to determine a page collection according to the keyword information and the information type of the keyword information, wherein the page collection includes a plurality of pages related to the keyword information;

[0013] A point acquisition unit is configured to acquire point information of each page in the page collection, wherein the point information represents a situation of the page being consumed by collecting the point pre-set in the page;

[0014] A page recommending unit is configured to determine a target page from the pages according to the point information of each page in the page collection, and push the target page to the user, wherein the target page is used to assist the user in creating a new page.

[0015] In a third aspect, an electronic device is provided, including a memory and a processor.

[0016] The memory stores computer execution instructions.

[0017] The processor executes the computer execution instructions stored in the memory, so that the processor executes the implementation manner of the first aspect.

[0018] In a fourth aspect, a computer readable storage medium is provided, and the computer readable storage medium stores computer execution instructions. When the processor executes the computer execution instructions, the computer execution instructions are used to implement the implementation manner of the first aspect.

[0019] In a fifth aspect, a computer program product is provided, and the computer program product includes a computer program. When the processor executes the computer program, the computer program implements the implementation manner of the first aspect.

[0020] The page recommendation method, device, electronic device and storage medium provided by the embodiments of the present application can determine the demand of the user for the page to be pushed, that is, determine what kind of page the user wants to create, by obtaining the keyword information input by the user. According to the keyword information and the information type of the keyword information, some pages can be determined from the plurality of pages that have been created, as a page set. That is, a plurality of pages related to the keyword information can be determined. For each page in the page set, the page hit information of the pages is obtained according to the preset page hit. The page hit information represents the situation of the page being consumed by collecting the page hit preset in the page, for example, the number of times of browsing the page, the number of people browsing the page, etc. According to the page hit information of each page, the target page is determined from the pages, and the target page is pushed to the user, so that the user can create a new page by referring to the target page. By determining the page set, the page closely related to the demand of the user can be pushed to the user, and by obtaining the page hit information, the determination accuracy of the target page is further improved. It is beneficial for the user to quickly create a new page, improve the quality of page creation, meet the actual demand of the user, improve the user experience, and improve the effect of marketing promotion. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings, which are incorporated herein and form a part of the specification, illustrate embodiments consistent with the present application and, together with the description, further serve to explain the principles of the present application.

[0022] Figure 1 A flowchart of a page recommendation method provided by the embodiments of the present application is shown in the figure.

[0023] Figure 2 A flowchart of a page recommendation method provided by the embodiments of the present application is shown in the figure.

[0024] Figure 3 A schematic diagram of a topology graph provided by the embodiments of the present application is shown in the figure.

[0025] Figure 4 A flowchart of a page recommendation method provided by the embodiments of the present application is shown in the figure.

[0026] Figure 5 A structural diagram of a page recommendation device provided by the embodiments of the present application is shown in the figure.

[0027] Figure 6 A structural diagram of an electronic device provided by the embodiments of the present application is shown in the figure.

[0028] The above-described figures have shown the specific embodiments of the present application, and the following will have a more detailed description. These figures and textual descriptions are not intended to limit the scope of the concept of the present application by any means, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0029] The exemplary embodiments will be described in detail herein with reference to the attached drawings. In the following description, like reference numerals refer to like elements, unless the context clearly dictates otherwise. The following exemplary embodiments are not representative of all embodiments consistent with the present application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the present application.

[0030] First, the terms involved in the present application are explained:

[0031] LMM: Large Multimodal Model, a multi-modal model, is a general-purpose model for fusing text, image, audio and other multi-modal data.

[0032] LLM: Large Language Model, a large language model, is a deep learning model that focuses on understanding and generating natural language text.

[0033] In a content creation and publishing platform, a landing page editor is usually provided, and an operator can upload materials, then edit and integrate pictures, texts and other materials through drag-and-drop and other methods, and configure marketing activity information, and finally generate a landing page. When a potential consumer clicks on the page, the page usually displays marketing activities, discounts and other information, and guides the potential consumer to make a purchase, thereby further improving the conversion rate of the consumer.

[0034] Currently, it is generally necessary to pre-configure a page template of the landing page, and the operator manually searches for a template according to a marketing category, and makes a new page based on the pre-configured template. The operator usually manually creates marketing materials, which has high manual involvement, low page creation efficiency and high cost. In addition, the pages created by the operator have uneven quality, low conversion rate, and low resource utilization rate of page materials.

[0035] It should be noted that the data in the present embodiment is not for a specific user, and cannot reflect the personal information of a specific user. It should be noted that the data in the present embodiment comes from a public data set. In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in the technical solution of the present disclosure comply with the relevant legal regulations and do not violate public order and good customs.

[0036] The technical solutions of the present application and how the technical solutions of the present application solve the above technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described again in some embodiments. The embodiments of the present application will be described below with reference to the drawings.

[0037] Figure 1 A flowchart of a page recommendation method provided in an embodiment of the present application is shown in FIG. 1. The method comprises the following steps. Figure 1

[0038] In S101, keyword information input by a user is acquired, and the information type of the keyword information is determined. The keyword information represents the user's demand for a page to be pushed, and the information type represents the form of the keyword information.

[0039] For example, the user can be an operator of a marketing activity, and can create a page according to actual business needs. Different modal materials can be deployed in the page, such as text, pictures, audio and video materials. The materials in the page can show information about activities related to the business, such as marketing information about an activity promotion. The user uploads the created page to a network platform for consumers to browse. Consumers can participate in the activity by browsing the information in the page, thereby improving the conversion rate of the page. The conversion rate refers to the probability of participating in the activity by browsing the page.

[0040] When the user wants to create a new page, the user can input keyword information according to actual needs, so as to push an existing page to the user according to the keyword information. The keyword information represents what kind of page the user wants to create, i.e., represents the user's demand for a page to be pushed. For example, the user wants to create a page for a promotion activity of a dining consumption coupon, and the keyword information can be "dining consumption coupon". The keyword information input by the user can be acquired in real time, and the keyword information can include one or more.

[0041] In this embodiment, the input method of the keyword information is not specifically limited. For example, the input method can be keyboard input or voice input. The format of the keyword information is also not limited, for example, the keyword information can be in Chinese, English, numbers, etc.

[0042] The user can input multiple keyword information, and each keyword information can correspond to its own information type, which represents the form of the keyword information. The information types of different keyword information can be the same or different. For example, the information type can include text, numbers, strings, etc. After acquiring the keyword information input by the user, the type of the keyword information can be identified based on a preset recognition algorithm to obtain the information type of each keyword information. In this embodiment, the preset recognition algorithm is not specifically limited.

[0043] ​The presentation form is the format of the keyword information, that is, the user can input keyword information in various formats, for example, the format of the keyword information can be a text format, a string format, a picture format, and the like. For example, the keyword information input by the user is "dining consumption coupon" and a coupon identifier (Identifier, ID), wherein "dining consumption coupon" is in a text format, and the coupon ID is "0000001", that is, in a string format.

[0044] In S102, a page set is determined according to the keyword information and the information type of the keyword information, wherein the page set includes a plurality of pages related to the keyword information.

[0045] For example, all pages created in a historical time period can be obtained, and one or more pages are determined from the pages as the page set according to the keyword information and the information type of the keyword information. For example, different page selection strategies can be preset based on the information type of the keyword information. According to the preset page selection strategy, the page related to the keyword information is searched according to the keyword information and added to the page set. For further example, if the information type of the keyword information is text, it can be determined whether the keyword information is contained in the page, if yes, the page is added to the page set; if not, the page does not need to be added to the page set. In this embodiment, the preset page selection strategy is not limited. By setting the page selection strategy, the pages can be screened according to the information type of the keyword information, the determination accuracy of the page set is improved, and the accuracy of the page recommendation is improved.

[0046] A label library can also be preset, and a large number of labels are stored in the label library. The labels in the label library are labels in the created pages. For example, the pages created in a historical time period can be obtained, and the labels are obtained from the pages, and the obtained labels are stored in association with the pages in the label library. Different labels can correspond to one page, and one label can exist in different pages.

[0047] The label can represent the promotion of the information related to the activity in the page, for example, the promotion of the marketing information. For the pages created in the historical time period, the marketing information in the pages can be obtained, and the label of the page is determined according to the marketing information in the page. The marketing information in the page can be identified, and the corresponding label is obtained according to the identification result, for example, the type of the marketing product, the marketing stage, the region of the marketing activity represented in the marketing information are determined by image recognition or semantic recognition, that is, the label can represent the type of the marketing product, the marketing stage, the region of the marketing activity.

[0048] After obtaining keyword information, it can be compared with various tags in the tag library to identify one or more tags, forming a candidate tag set. For example, the semantic similarity between keyword information and various tags can be calculated. If the semantic similarity is greater than a preset threshold, the tag corresponding to that semantic similarity is included in the candidate tag set. For pages that have already been created, if any one or more tags from the candidate tag set exist on the page, the page can be added to the page collection.

[0049] Different pages can contain the same tag; that is, one tag can correspond to multiple pages. After determining the candidate tag set, for each tag in the candidate tag set, we can determine all the pages corresponding to that tag, thus obtaining the page set.

[0050] Different information types can correspond to different methods for determining candidate tag sets. Based on the information type of the keyword information, a corresponding method can be used to determine the candidate tag set from a pre-defined tag library. For example, for text format, semantic recognition can be used to find tags with similar semantics from the tag library; for string format, tags that are completely identical to the string of the keyword information can be found from the tag library.

[0051] S103. Obtain the event tracking information for each page in the page set; wherein, the event tracking information represents the consumption of the page by means of the event tracking information preset in the page.

[0052] For example, the "My Page" in the page collection refers to pages that have already been created. For each page in the page collection, its event tracking information can be directly obtained. This event tracking information can be updated in real time. It represents the consumption of the page data collected through pre-set event tracking points. Event tracking points refer to specific code snippets pre-embedded in a webpage to collect data such as consumer behavior, interactions, and the page's overall performance. For example, event tracking information may include device information, channel, page dwell time, PV (Page Views), UV (Unique Visitors), and button clicks. Channels can include in-app and out-of-app components. In-app components refer to webpages displayed on platforms developed by the company itself, while out-of-app components refer to webpages displayed on third-party platforms.

[0053] S104. Based on the tracking information of each page in the page collection, determine the target page from each page and push the target page to the user; the target page is used to assist the user in creating a new page.

[0054] Exemplarily, for each page in the page set, one or more pages are determined from the pages as target pages according to the page information of the pages. The target pages are pushed to the user, and the user can create a page required by the user according to the layout of the target pages and information such as the material in the target pages. For example, the user can extract the material from the target pages, and add the material to a new page, to realize secondary use of the material and improve the efficiency of page creation.

[0055] The pages can be sorted according to the page information of the pages, and the pages ranked in the first few positions are determined as the target pages. For example, the pages can be sorted according to the PV amount, and the first three pages with the largest PV amount are determined as the target pages.

[0056] The number of labels in the candidate label set contained in each page can also be determined, the pages are sorted according to the number, and the first few pages with a larger number of labels in the candidate label set are retained. That is, the pages most closely related to the user demand are determined. Then, the pages are sorted according to the page information of the pages, to obtain the final target pages.

[0057] In the embodiment of the application, the keyword information input by the user is obtained, the demand of the user for the page to be pushed is determined, that is, what kind of page the user wants to create is determined. According to the keyword information and the information type of the keyword information, some pages are determined from the plurality of pages that have been created as the page set. That is, a plurality of pages related to the keyword information are determined. For each page in the page set, the page information of the page is obtained according to the preset page information. The page information represents the consumption of the page collected through the preset page information, for example, the number of times of browsing the page, the number of people browsing the page, and the like. According to the page information of each page, the target pages are determined from the pages, and the target pages are pushed to the user, so that the user can create a new page by referring to the target pages. By determining the page set, the pages closely related to the demand of the user can be pushed to the user, and by obtaining the page information, the accuracy of the determination of the target pages is further improved. This is conducive to the user to quickly create a new page, improves the quality of page creation, meets the actual demand of the user, improves the user experience, and improves the effect of marketing promotion.

[0058] Figure 2 A flowchart of a page recommendation method provided in the embodiment of the application is shown in Figure 1 The page recommendation method is described in detail based on the embodiment. As shown in Figure 2 The method comprises the following steps.

[0059] S201, acquire keyword information input by a user, and determine an information type of the keyword information; wherein the keyword information represents a demand of the user for a page to be pushed, and the information type represents a form of the keyword information.

[0060] By way of example, this step can refer to step S101 described above, and will not be described again.

[0061] S202, if the information type of the keyword information is a page identifier, find a node in which the keyword information is located in a topology graph, determine the node as an initial node, and determine a node connected to the initial node by an edge as a neighboring node.

[0062] By way of example, the information type of the keyword information can include a page identifier, each page corresponds to a page identifier, and the page identifier can be in the form of a string, for example, the page identifier can be a coupon Id. A topology graph is pre-set, the topology graph includes a plurality of nodes, and there is an edge connection between different nodes, each node represents a page, that is, each node corresponds to a page identifier. The edge connection represents a jump relationship between different pages.

[0063] For the case that the information type of the keyword information is a page identifier, it can be found from the topology graph which node corresponds to the keyword information. The node corresponding to the keyword information is determined as the initial node. It can also be found from the topology graph that the node connected to the initial node by the edge connection, and the node connected to the initial node by the edge connection is determined as the neighboring node.

[0064] S203, determine a page set according to a page corresponding to the initial node and a page corresponding to the neighboring node.

[0065] By way of example, each node corresponds to a page, and the page corresponding to the initial node and the page corresponding to the neighboring node are determined. According to the page corresponding to the initial node and the page corresponding to the neighboring node, the page set is determined, for example, the page corresponding to the initial node and the page corresponding to the neighboring node can be determined as the pages in the page set.

[0066] In this embodiment, the page set is determined according to the page corresponding to the initial node and the page corresponding to the neighboring node, including: acquiring a label in the page corresponding to the initial node from a pre-set label library as an initial label, and acquiring a label in the page corresponding to the neighboring node as a neighboring label; wherein each page corresponds to at least one label, and the label represents an introduction to information in the page; obtaining a first label set according to the initial label and the neighboring label; wherein the first label set includes at least one label, and each label corresponds to at least one page; and determining the page corresponding to the label in the first label set to obtain the page set.

[0067] Specifically, the keyword information can be in the format of a page identifier, for example, the keyword information can be a coupon ID, an applet ID, etc.

[0068] The topology graph is preset according to all the pages created in a preset historical time period. The topology graph includes multiple nodes, and edges can exist between different nodes. Each node represents a page, and the edge represents a jump relationship between different pages. For example, the A node points to the B node, which means that the page of the A node can jump to the page of the B node. Each node can correspond to a page identifier, and the page identifier can be an identifier of marketing information in the page. For example, the identifier of the marketing information can include an ID of a coupon in the marketing information, a link of a marketing activity, an ID of an applet where the marketing activity is located, etc.

[0069] For the case where the keyword information is a page identifier, the node where the page identifier is located in the preset topology graph is found as an initial node. According to the topology relationship between the nodes, the nodes connected by the edge with the initial node are determined as adjacent nodes.

[0070] The preset tag library stores multiple tags, and each tag can come from a different page, that is, the preset tag library can store the association relationship between the tags and the pages. The tag can represent the introduction of the information in the page, for example, it can represent the promotion of the marketing information. Each node corresponds to a page, and all the tags associated with the page corresponding to the initial node are obtained from the tag library as initial tags, and all the tags associated with the page corresponding to the adjacent node are obtained from the tag library as adjacent tags. Each page can correspond to multiple tags, that is, multiple initial tags and multiple adjacent tags can be obtained. The initial tags and the adjacent tags are combined into a first tag set, that is, the first tag set includes the initial tags and the adjacent tags.

[0071] For each tag in the first tag set, determine which pages the tag corresponds to, and determine these pages as the pages in the page set. That is, the pages corresponding to one or more tags in the first tag set can be added to the page set.

[0072] Different pages can correspond to the same tag, that is, one tag can correspond to multiple pages. After determining the first tag set, for each tag in the first tag set, all the pages corresponding to the tag can be determined, thereby obtaining the page set.

[0073] The beneficial effect of such a setting is that the pages related to the keyword information are determined from the topology graph, and the tags in these pages are determined as the first tag set, which facilitates the user to find the pages that may meet the demand and improves the accuracy of page recommendation.

[0074] In the embodiment, the method further includes: extracting a page link from the page; determining a jump page of the page according to the page link; and establishing an edge connection between a node corresponding to the page and a node corresponding to the jump page of the page, to obtain the topology graph.

[0075] Specifically, all pages created in a preset historical time period are obtained, and the pages contain marketing information, for example, the pages can include pictures or texts representing the marketing information. Each marketing information can correspond to its own string, for example, CouponId (coupon identifier), applet Id, native page, link, etc. can be extracted from the marketing information, which can all be used as a page identifier related to the marketing information. In the embodiment, the way of extracting the string is not limited. For example, a page field can be obtained from an HTML (HyperText Markup Language) page for parsing, and the page field is the main part to distinguish different pages. CouponId, applet Id, native page, link, etc. can be extracted from the page field.

[0076] For each page, a page link can also be extracted from the page, and each page corresponds to its own page link. According to the page link, a jump page of the page is determined. The jump page refers to the next page jumped from the page through the page link, for example, the user clicks the marketing information on the page to jump to the next page, and the next page is the jump page. The jump page is also a node in the topology graph, and an edge connection is established between the node corresponding to the page and the node corresponding to the jump page of the page, to obtain the topology graph.

[0077] Figure 3 The topology graph is shown in the following schematic diagram. Figure 3 The nodes include node one, node two, node three, and node four. The page identifiers in the pages of node one and node four are in the form of H5 link, the page identifier in the page of node two is in the form of applet Id, and the page identifier in the page of node three is in the form of CouponId. Node one points to node two, which represents that the page of H5 link can jump to the page of applet through a control on the page, that is, the page link of node one can jump to the page link of node two. Similarly, node one points to node three, which represents that the page of H5 link can jump to a ticket through a control, node two points to node three, which represents that the page of applet can jump to a ticket through a control, and node one points to node four, which represents that the page of H5 link can jump to the page of another H5 link through a control on the page.

[0078] The beneficial effect of such an arrangement is that the next page jumped to by a page can be determined according to the page link, a topology graph is created according to the jump relationship between pages, a topology network of marketing information is obtained, and thus the associated pages can be found according to the topology relationship, and the accuracy and comprehensiveness of page recommendation are improved. In this embodiment, the keyword information information type includes text; the page set is determined according to the keyword information and the information type of the keyword information, including: if the information type of the keyword information is text, generating prompt word information according to the keyword information and a preset prompt word template; wherein the prompt word information includes the keyword information, and the prompt word information represents the processing requirement of the preset large model; inputting the prompt word information into the preset large model, and obtaining a second label set based on a preset label library; wherein the preset large model is used to determine a label from the preset label library that meets the processing requirement of the prompt word information; determining the page corresponding to the label in the second label set to obtain the page set.

[0079] Specifically, the keyword information can be in the format of text. For example, the keyword information can be the name of a marketing activity input by a user, the name of a coupon, an activity area, and the like.

[0080] For the case where the keyword information is text, all labels in the preset label library are obtained. Each label can be represented by a unique identifier, and each label identifier corresponds to its own semantics. For example, the label 001 represents the semantic content of “dining”; the label 002 represents the semantic content of “retail”; and the label 003 represents the semantic content of “travel”.

[0081] A large model is preset, which is an LLM. The model can store semantic information corresponding to each label. For each label in the label library, the keyword information can be input into the large model. A prompt word template can also be preset. The prompt word template is in the form of text data and can assist the model in understanding the task. According to the keyword information and the preset prompt word template, complete prompt word information can be generated. The prompt word information includes the keyword information, and the prompt word information can represent the processing requirement of the preset large model. The prompt word information is input into the large model, and the large model uses the preset label library to filter a plurality of labels from the label library as a second label set. That is, the large model can be used to determine a label from the preset label library that meets the processing requirement of the prompt word information.

[0082] For example, for each label in the label library, the semantic similarity between the label and the keyword information can be calculated by the large model. The semantic similarity represents the similarity of the semantics between the semantic information corresponding to the label and the keyword information. In this embodiment, the structure of the large model and the calculation process of the semantic similarity are not specifically limited. Each label in the label library can correspond to a semantic similarity. According to the semantic similarities corresponding to the labels in the label library, a plurality of labels are determined from the label library as the second label set. For example, the semantic similarities can be sorted from large to small, and the labels ranked in the top K positions are taken as the labels in the second label set. For example, the keyword information is “dining consumption coupon”, and the large model can filter out the labels related to dining and consumption coupons from the label library as the second label set through the inverted index technology. From the created pages, the pages corresponding to one or more labels in the second label set are determined as the page set. For example, as long as a page contains at least one label in the second label set, the page is added to the page set.

[0083] The same label can correspond to different pages, that is, one label can correspond to a plurality of pages. After determining the second label set, for each label in the second label set, all the pages corresponding to the label can be determined, thereby obtaining the page set.

[0084] The beneficial effect of such setting is that the large model determines the second label set according to semantic understanding, improves the determination accuracy of the second label set, and further improves the accuracy of page recommendation.

[0085] The keyword information can include both the page identifier and the text. For the case of containing both the page identifier and the text, the page identifier in the keyword information can be used to find the node where the keyword information is located in the topology graph, as the initial node, and the node connected to the initial node through the edge is determined as the adjacent node; the label in the page corresponding to the initial node in the preset label library is obtained as the initial label, and the label in the page corresponding to the adjacent node is obtained as the adjacent label; according to the initial label and the adjacent label, a label set is obtained as the first label set. For the text in the keyword information, the semantic similarity between the labels in the preset label library and the keyword information can be determined based on the preset large model; according to the semantic similarities corresponding to the labels in the preset label library, a label set is determined as the second label set. The union of the first label set and the second label set is taken as the final candidate label set. Then, according to the candidate label set, the page set is obtained. By performing two layers of label screening, label omission can be avoided, and the accuracy of page recommendation is improved.

[0086] In the embodiment, the method further includes: extracting material information from the page, the material information including at least one of text, image and video; performing fusion processing on the material information in the page based on a preset multi-modal fusion model to obtain description information of the page, the multi-modal fusion model being configured to fuse material information of different modalities into description information, the description information representing content expressed by all the material information in the page; obtaining a label corresponding to the page according to the description information of the page; and obtaining a preset label library according to labels corresponding to all the pages in the preset historical time period.

[0087] Specifically, the label library is constructed in advance according to the pages created in the preset historical time period. All the labels in the pages can be stored in the label library, and an association relationship between the labels and the pages is also stored.

[0088] All the pages created in the preset historical time period are obtained, and for each page, material information contained in the page can be extracted. The material information can include at least one of text, image and video, that is, multi-modal material information can be extracted.

[0089] The multi-modal fusion model is set in advance, and the multi-modal fusion model is an open-source neural network model, that is, LMM, for example, the Qwen2-VL model can be used. In the embodiment, the model architecture of the multi-modal fusion model is not limited. For each page, all the material information in the page is input into the preset multi-modal fusion model, and the multi-modal fusion model performs fusion processing on the material information in the page to obtain description information of the page. That is, the input data of the multi-modal fusion model is the material information, and the output data is the description information, and the output format can be JSON format. The multi-modal fusion model is configured to fuse material information of different modalities into unified description information, and the description information can represent content expressed by all the material information in the page.

[0090] The description information can be in the form of text, that is, the content in the page can be expressed through the description information, and the description information and the content in the page can allow deviation. For example, the page contains material information of two modalities of text and image, and the content expressed is about travel, food, airlines, and various preferential activities, and the description information output by the multi-modal fusion model is “This content summarizes various preferential information, covering travel, car maintenance, refueling, shopping, tourism and food. It includes random discounts for XX rental cars, XX car maintenance, full and reduced discounts for XX travel, and full and reduced discounts for XXX brand. In addition, there are hotel stay red packet, XX travel full and reduced discount, and XX airline full 500 yuan discount. Overall, it provides rich consumer discounts to help consumers enjoy more benefits in different scenarios.”

[0091] The description information of the page is processed by word segmentation or semantic recognition to obtain the label corresponding to the page. For example, a prompt template of a large model for a marketing industry scenario can be customized. The large model and the large model for calculating semantic similarity can be different. The prompt template and the description information are input into the large model for the scene to be labeled, that is, all labels corresponding to the page are obtained.

[0092] All labels corresponding to all pages in a historical time period are obtained, and these labels are associated with the corresponding pages to obtain a final label library. One page can correspond to multiple labels, and one label can correspond to multiple pages.

[0093] The beneficial effect of such a setting is that by performing multi-modal fusion, all labels in the existing page are determined in advance, a label library is created, and subsequent quick finding of labels and pages related to user needs is facilitated, improving the efficiency and accuracy of page recommendation.

[0094] In this embodiment, based on the preset multi-modal fusion model, the material information in the page is fused to obtain the description information of the page, including: inputting the material information in the page into the preset multi-modal fusion model to obtain the text explanation information corresponding to the material information; wherein the text explanation information represents the content expressed by the material information; and the text explanation information corresponding to each material information in the page is fused to obtain the description information of the page.

[0095] Specifically, for each page, all material information in the page is input into the preset multi-modal fusion model. The multi-modal fusion model first performs semantic recognition on each material information to obtain the text explanation information corresponding to the material information. The text explanation information represents the content expressed by the material information. That is, a text description is obtained for each material information.

[0096] After obtaining the text explanation information corresponding to all material information in the page, the text explanation information corresponding to each material information in the page is fused to obtain the unified description information of the page. For example, the text explanation information corresponding to each material information can be spliced to obtain the description information; or after splicing, the semantic adjustment and addition / deletion processing can be performed to obtain the final description information.

[0097] The beneficial effect of such a setting is that one material information can correspond to one text description, and multiple material information can be unified into one text description, avoiding omission of the content represented by the material information, improving the determination accuracy of the description information, and further improving the recommendation accuracy of the page.

[0098] In this embodiment, after receiving each material information, the multi-modal fusion model can perform data preprocessing on the material information, for example, multi-modal encoding and alignment of text, pictures and videos in the page. Specifically, different resolution images and videos can be dynamically processed through a naive dynamic resolution mechanism to generate a unified visual representation, thereby improving the accuracy and efficiency of subsequent description information determination.

[0099] In this embodiment, according to the description information of the page, the label corresponding to the page is obtained, including: obtaining the configuration information of the page; wherein the configuration information represents the information required in the creation process of the page; and obtaining the label corresponding to the page according to the description information and the configuration information of the page.

[0100] Specifically, for each page created in a preset historical time period, the configuration information of the page is obtained. The configuration information represents the information required in the creation process of the page, for example, the configuration information can include title (title), page description, page type, marketing campaign validity period, promotion channel, approval log and the like. Each page stores its own configuration information, and the configuration information of the page can be obtained at any time.

[0101] Combined with the description information and the configuration information of the page, all labels corresponding to the page can be obtained. For example, the description information and the configuration information can be spliced, and the spliced information can be processed by word segmentation or semantic recognition to obtain the label of the page.

[0102] The beneficial effects of such setting are that the labels in the page are determined in combination with the description information and the configuration information, avoiding label omission and improving the comprehensiveness of the label library.

[0103] In this embodiment, according to the description information and the configuration information of the page, the label corresponding to the page is obtained, including: performing semantic recognition processing on the description information of the page, extracting initial labels from the description information of the page; and supplementing the initial labels according to the configuration information of the page to obtain the label corresponding to the page.

[0104] Specifically, after obtaining the description information and the configuration information of the page, the description information and the configuration information can be processed respectively to extract the corresponding labels. For example, the description information of the page can be processed by semantic recognition to extract the labels from the description information of the page as first labels, and the configuration information of the page can be processed by semantic recognition to extract the labels from the configuration information of the page as second labels. The first labels and the second labels are determined as the label corresponding to the page.

[0105] Alternatively, you can first perform word segmentation or semantic recognition on the description information of the page based on a large model of the marketing industry scenario to obtain the initial tags corresponding to the page. Then, based on the configuration information of the page, you can perform secondary processing on the obtained initial tags to obtain the final tags of the page.

[0106] The secondary processing can involve supplementing the initial tags extracted from the description information with the configuration information, i.e., increasing the number of tags corresponding to the page. The supplemented tags, along with the initial tags, together constitute the tags corresponding to the page. For example, multiple tags can be extracted from the configuration information to determine if these tags have already been determined in the description information. If not, they are added as page tags; otherwise, they do not need to be added. One tag can be extracted from each piece of configuration information. For example, if the configuration information includes a page type, and the page type is "long page," then the tag "long page" can be added. Simple characters can be used to represent tags; for example, "T4" can be used to represent a long page. Table 1 is a schematic table of tag types.

[0107] Table 1. Illustration of Label Types

[0108]

[0109] As shown in Table 1, the types of tags can include marketing product type tags, marketing promotion type tags, marketing stage tags, page type tags, channel tags, language tags, region tags, coupon tags, etc. Different types of tags can include multiple representations of content. For example, if all the text on a page is in Chinese, then the language tag for that page can be "Chinese".

[0110] The advantage of this setup is that it tags pages based on description information and then supplements the tags based on configuration information, thereby improving the accuracy of the tag library.

[0111] S204. Obtain the event tracking information for each page in the page set; wherein, the event tracking information represents the consumption of the page by means of the event tracking information preset in the page.

[0112] For example, this step can refer to step S103 above, and will not be repeated here.

[0113] S205. Based on the tracking information of each page in the page collection, determine the target page from each page and push the target page to the user; wherein, the target page is used to assist the user in creating a new page.

[0114] For example, this step can refer to step S104 above, and will not be repeated here.

[0115] In the embodiment of the present application, by acquiring the keyword information input by the user, the demand of the user for the page to be pushed can be determined, that is, what kind of page the user wants to create is determined. According to the keyword information and the information type of the keyword information, part of the pages can be determined from the multiple pages that have been created as the page set. That is, multiple pages related to the keyword information can be determined. For each page in the page set, according to the preset burying point, the burying point information of these pages is acquired. The burying point information represents the situation of the page being consumed collected through the preset burying point in the page, for example, the browsing times, the number of browses and the like of the page can be acquired. According to the burying point information of each page, the target page is determined from these pages, and the target page is pushed to the user, so that the user can create a new page by referring to the target page. By determining the page set, the page closely related to the demand of the user can be pushed to the user, and by acquiring the burying point information, the determination accuracy of the target page is further improved. It is beneficial for the user to quickly create a new page, improve the quality of page creation, meet the actual demand of the user, improve the user experience, and improve the effect of marketing promotion.

[0116] Figure 4 The flowchart of the page recommendation method provided in the embodiment of the present application is shown in Figure 1 The page recommendation method is described in detail based on the embodiment. As shown in Figure 4 The method comprises the following steps:

[0117] S401, acquiring keyword information input by a user, and determining an information type of the keyword information; wherein the keyword information represents a demand of the user for a page to be pushed, and the information type represents a form of the keyword information.

[0118] By way of example, this step can refer to step S101 described above, and will not be described again.

[0119] S402, determining a page set according to the keyword information and the information type of the keyword information; wherein the page set comprises multiple pages related to the keyword information.

[0120] By way of example, this step can refer to step S102 described above, and will not be described again.

[0121] S403, acquiring burying point information of each page in the page set; wherein the burying point information represents a situation of the page being consumed collected through a preset burying point in the page.

[0122] By way of example, this step can refer to step S103 described above, and will not be described again.

[0123] S404. For each page in the page set, determine the page's rating information based on the page's tracking information; whereby the rating information represents the user's level of interest in the page.

[0124] For example, the page collection includes multiple pages. For each page in the page collection, the tracking information for that page is obtained. Based on the tracking information, the page is evaluated to obtain a rating. The rating represents the user's level of interest in the page; the higher the rating, the more likely the page is to be pushed to the user.

[0125] Tracking data can include metrics such as device information, button click data, page dwell time, page views (PV), unique visitors (UV), conversions, and shares. These metrics can be comprehensively considered to score a page. For example, a higher PV value or a longer page dwell time generally results in a higher score. Different weights can be preset for different metrics within the tracking data, and the page's score can be determined through weighted summation and other formulas.

[0126] In this embodiment, the rating information of the page is determined based on the tracking information of the page, including: obtaining the creation time of the page, and determining the weight information of the page based on the creation time; wherein, the weight information represents the importance of the page; and determining the rating information of the page based on the tracking information and weight information of the page.

[0127] Specifically, the page creation time can be obtained, and a dynamic weight can be assigned to the page based on this time, serving as the page's weight information. This weight information represents the importance of the page; for example, the earlier the creation time, the lower the weight. As the current time changes, the weight information can be dynamically adjusted in real-time or periodically.

[0128] By combining the page's tracking data and weight information, a page rating is obtained. First, the page can be rated based on its tracking data to obtain an initial rating. Then, the final page rating is obtained by combining the initial rating with the weight information. For example, one could first comprehensively consider various metrics from the tracking data to calculate an initial rating, and then multiply the initial rating by the weight information to obtain the final rating.

[0129] The advantage of this setup is that it dynamically adjusts the page's weight information based on its creation time, avoiding recommending overly old historical pages to users, thus improving recommendation accuracy and enhancing user experience.

[0130] S405. Based on the rating information of each page in the page collection, sort the pages, determine the target page from the pages according to the sorting results, and push the target page to the user.

[0131] For example, after obtaining the rating information for each page in the page set, this rating information is sorted, for example, in descending order. Based on the sorting result, the target page is determined from these pages; for example, the top N pages can be identified as the target pages. Alternatively, a preset rating threshold can be used to identify pages with ratings greater than the preset threshold as target pages.

[0132] Once the target page is identified, it is pushed to the user, who can then extract creative materials from it for creating new pages. Alternatively, a page recommendation list can be generated based on the sorting results and pushed to the user. The pages in the recommendation list are arranged in the same order as the sorting results. Users can view links to pages and the creative materials they contain using the recommendation list, improving page creation efficiency and quality, thereby enhancing the effectiveness of marketing campaigns.

[0133] In this embodiment, by obtaining keyword information input by the user, the user's needs for the pages to be pushed can be determined, that is, what kind of page the user wants to create. Based on the keyword information and its type, a subset of pages can be identified from multiple previously created pages, forming a page set. That is, multiple pages related to the keyword information can be identified. For each page in the page set, tracking information is obtained based on preset tracking points. Tracking information represents the consumption status of the page collected through preset tracking points; for example, the number of page views and the number of viewers can be obtained. Based on the tracking information of each page, a target page is determined from these pages and pushed to the user, allowing the user to create new pages based on the target page. By determining the page set, pages closely related to the user's needs can be pushed to the user. Obtaining tracking information further improves the accuracy of target page determination. This facilitates users in quickly creating new pages, improves the quality of page creation, meets the user's actual needs, enhances user experience, and improves the effectiveness of marketing promotion.

[0134] Figure 5 This is a schematic diagram of the structure of a page recommendation device provided in an embodiment of this application, as shown below. Figure 5 As shown, the page recommendation device 50 provided in this embodiment includes:

[0135] The information acquisition unit 501 is used to acquire keyword information input by the user and determine the information type of the keyword information; wherein, the keyword information represents the user's demand for the page to be pushed, and the information type represents the presentation form of the keyword information;

[0136] The collection determining unit 502 is configured to determine a page collection according to the keyword information and the information type of the keyword information, wherein the page collection includes a plurality of pages related to the keyword information.

[0137] The burying point acquiring unit 503 is configured to acquire burying point information of each page in the page collection, wherein the burying point information represents a situation of the page being consumed by the burying point preset in the page.

[0138] The page recommending unit 504 is configured to determine a target page from the pages in the page collection according to the burying point information of the pages, and push the target page to the user, wherein the target page is used to assist the user in creating a new page.

[0139] In a possible implementation, the information type of the keyword information includes a page identifier; a topology graph is preset, the topology graph includes a plurality of nodes, and there is an edge connection between different nodes, the node represents a page, and the edge connection represents a jump relationship between different pages; a page is in one-to-one correspondence with a page identifier; the collection determining unit 502 is specifically configured to:

[0140] If the information type of the keyword information is a page identifier, the node where the keyword information is located is searched from the topology graph as an initial node, and the node connected to the initial node by the edge connection is determined as a neighboring node;

[0141] According to the page corresponding to the initial node and the page corresponding to the neighboring node, the page collection is determined.

[0142] In a possible implementation, the collection determining unit 502 is specifically configured to:

[0143] The label in the page corresponding to the initial node is acquired from a preset label library as an initial label, and the label in the page corresponding to the neighboring node is acquired as a neighboring label; each page corresponds to at least one label, and the label represents an introduction situation of information in the page;

[0144] According to the initial label and the neighboring label, a first label collection is obtained; the first label collection includes at least one label, and each label corresponds to at least one page;

[0145] The page corresponding to the label in the first label collection is determined to obtain the page collection.

[0146] In a possible implementation, the method further includes:

[0147] The link extracting unit is configured to extract a page link from a page created in a preset historical time period.

[0148] a jump page determination unit, configured to determine a jump page of the page according to the page link, wherein the jump page represents a next page jumped to from the page through the page link;

[0149] a topology graph determination unit, configured to establish an edge connection between a node corresponding to the page and a node corresponding to the jump page of the page, to obtain the topology graph.

[0150] In a possible implementation, the information type of the keyword information includes text; the set determination unit 502 is specifically configured to:

[0151] if the information type of the keyword information is text, generate prompt word information according to the keyword information and a preset prompt word template; wherein the prompt word information includes keyword information, and the prompt word information represents a processing requirement for a preset large model;

[0152] input the prompt word information into the preset large model, and obtain a second label set based on a preset label library; wherein the preset large model is used to determine a label meeting the processing requirement of the prompt word information from the preset label library;

[0153] determine a page corresponding to a label in the second label set, to obtain the page set.

[0154] In a possible implementation, the method further includes:

[0155] a material extraction unit, configured to extract material information from a page created in a preset historical time period; wherein the material information includes at least one modality of text, image, and video;

[0156] a material fusion unit, configured to perform fusion processing on the material information in the page based on a preset multi-modal fusion model, to obtain description information of the page; wherein the preset multi-modal fusion model is used to fuse material information of different modalities into description information, and the description information represents content expressed by all material information in the page;

[0157] a label determination unit, configured to obtain a label corresponding to the page according to the description information of the page;

[0158] a label library determination unit, configured to obtain the preset label library according to labels corresponding to all pages in the preset historical time period.

[0159] In a possible implementation, the material fusion unit is specifically configured to:

[0160] The material information on the page is input into a preset multimodal fusion model to obtain the text explanation information corresponding to the material information; wherein, the text explanation information represents the content expressed by the material information;

[0161] The text explanation information corresponding to each material information on the page is fused to obtain the description information of the page.

[0162] In one possible implementation, the label determination unit is specifically used for:

[0163] Obtain the configuration information of the page; wherein the configuration information represents the information required by the page during the creation process;

[0164] Based on the description and configuration information of the page, the tags corresponding to the page are obtained.

[0165] In one possible implementation, the label determination unit is specifically used for:

[0166] Semantic recognition processing is performed on the description information of the page to extract initial tags from the description information of the page;

[0167] Based on the configuration information of the page, the initial tags are supplemented to obtain the tags corresponding to the page.

[0168] In one possible implementation, the page recommendation unit 504 is specifically used for:

[0169] For each page in the page set, the page rating information is determined based on the page's tracking information; wherein, the rating information represents the user's level of interest in the page;

[0170] Based on the rating information of each page in the page set, the pages are sorted, and based on the sorting results, the target page is determined from the pages.

[0171] In one possible implementation, the page recommendation unit 504 is specifically used for:

[0172] Obtain the creation time of the page, and determine the page's weight information based on the creation time; wherein the weight information represents the importance of the page;

[0173] Based on the tracking information and weight information of the page, determine the page's rating information.

[0174] This embodiment provides a page recommendation device that can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0175] Figure 6A structural schematic diagram of an electronic device is provided in an embodiment of the present application. As shown in the figure Figure 6 The electronic device 60 provided in the embodiment includes at least one processor 601 and a memory 602. Optionally, the device 60 further includes a communication component 603. The processor 601, the memory 602 and the communication component 603 are connected through a bus 604.

[0176] In the implementation process, the at least one processor 601 executes the computer execution instructions stored in the memory 602, so that the at least one processor 601 executes the above-mentioned method.

[0177] The specific implementation process of the processor 601 can refer to the method embodiments described above, which have similar implementation principles and technical effects, and will not be described here again in the embodiment.

[0178] In the above-mentioned embodiments, it should be understood that the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC) and the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor and the like. The steps of the method disclosed in the application can be directly embodied as execution completed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0179] The memory can include a random access memory (RAM), and can also include a non-volatile memory (NVM), for example, at least one disk memory.

[0180] The bus can be an industry standard architecture (ISA) bus, a peripheral component (PCI) bus or an extended industry standard architecture (EISA) bus and the like. The bus can be divided into an address bus, a data bus, a control bus and the like. For the convenience of representation, the bus in the drawings of the present application does not limit to only one bus or one type of bus.

[0181] The present application also provides a computer program product, including a computer program, which is executed by a processor to realize the above-mentioned method.

[0182] The application further provides a computer readable storage medium, wherein computer execution instructions are stored in the computer readable storage medium, and when a processor executes the computer execution instructions, the method described above is realized.

[0183] The readable storage medium described above can be realized 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 read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0184] An exemplary readable storage medium is coupled to the processor, so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in the device.

[0185] The division of units is only a logical function division, and in actual implementation, there can be another division mode, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0186] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or they can be distributed on a plurality of network units. According to actual needs, some or all of the units can be selected to achieve the purpose of the embodiment.

[0187] In addition, the functional units in each embodiment of the application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0188] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the embodiments of the method of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0189] It can be understood by those skilled in the art that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware. The aforementioned program can be stored in a computer readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, and various media that can store program codes.

[0190] Finally, it should be noted that: those skilled in the art will easily think of other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application that follow the general principles of the present application and include common knowledge or conventional technical means in the art that are not disclosed in the present application, and is not limited to the precise structure described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present application is only limited by the appended claims.

Claims

1. A method of recommending pages, characterized by, The method comprises the following steps: obtaining keyword information input by a user, and determining an information type of the keyword information; wherein the keyword information represents a demand of the user for a page to be pushed, and the information type represents a form of the keyword information; determining a page set according to the keyword information and the information type of the keyword information; wherein the page set comprises a plurality of pages related to the keyword information; obtaining page tracking information of each page in the page set; wherein the page tracking information represents a situation of the page being consumed through a tracking point preset in the page; determining a target page from the pages according to the page tracking information of the pages in the page set, and pushing the target page to the user; wherein the target page is used to assist the user in creating a new page.

2. The method of claim 1, wherein, The information type of the keyword information comprises a page identifier; a topology graph is preset, the topology graph comprises a plurality of nodes, there is an edge connection between different nodes, the nodes represent pages, and the edge connection represents a jump relationship between different pages; a page is in one-to-one correspondence with a page identifier; the page set is determined according to the keyword information and the information type of the keyword information, comprising: if the information type of the keyword information is a page identifier, a node where the keyword information is located in the topology graph is found as an initial node, and a node connected to the initial node through an edge connection is determined as a neighboring node; the page set is determined according to the page corresponding to the initial node and the page corresponding to the neighboring node.

3. The method of claim 2, wherein, The page set is determined according to the page corresponding to the initial node and the page corresponding to the neighboring node, comprising: an initial label in the page corresponding to the initial node is obtained from a preset label library as an initial label, and a label in the page corresponding to the neighboring node is obtained as a neighboring label; wherein each page corresponds to at least one label, and the label represents an introduction to information in the page; a first label set is obtained according to the initial label and the neighboring label; wherein the first label set comprises at least one label, and each label corresponds to at least one page; pages corresponding to the labels in the first label set are determined to obtain the page set.

4. The method of claim 2, wherein, Further comprising: for a page created in a preset historical time period, a page link is extracted from the page; a jump page of the page is determined according to the page link; wherein the jump page represents a next page jumped from the page through the page link; an edge connection is established between a node corresponding to the page and a node corresponding to the jump page of the page to obtain the topology graph.

5. The method of claim 1, wherein, The information type of the keyword information comprises text; the page set is determined according to the keyword information and the information type of the keyword information, comprising: if the information type of the keyword information is text, prompt word information is generated according to the keyword information and a preset prompt word template; wherein the prompt word information comprises keyword information, and the prompt word information represents a processing requirement for a preset large model. inputting the prompt word information into a preset large model, and obtaining a second label set based on a preset label library; the preset large model is used to determine a label meeting a processing requirement of the prompt word information from the preset label library; determining a page corresponding to a label in the second label set to obtain the page set.

6. The method according to claim 3 or 5, characterized in that, Further comprising: extracting material information from the page, wherein the material information includes at least one modality of text, image, and video; based on a preset multi-modal fusion model, the material information in the page is fused to obtain the description information of the page; wherein the preset multi-modal fusion model is used to fuse material information of different modalities into description information, and the description information represents the content expressed by all material information in the page; obtaining the label corresponding to the page according to the description information of the page; obtaining the preset label library according to the labels corresponding to all pages in the preset historical time period.

7. The method of claim 6, wherein, based on a preset multi-modal fusion model, the material information in the page is fused to obtain the description information of the page, including: inputting the material information in the page into a preset multi-modal fusion model to obtain text explanation information corresponding to the material information; wherein the text explanation information represents the content expressed by the material information; fusing the text explanation information corresponding to each material information in the page to obtain the description information of the page.

8. The method of claim 6, wherein, obtaining the label corresponding to the page according to the description information of the page, including: obtaining the configuration information of the page; wherein the configuration information represents the information required in the creation process of the page; obtaining the label corresponding to the page according to the description information and the configuration information of the page.

9. The method of claim 8, wherein, obtaining the label corresponding to the page according to the description information and the configuration information of the page, including: performing semantic recognition processing on the description information of the page to extract an initial label from the description information of the page; performing supplementary processing on the initial label according to the configuration information of the page to obtain the label corresponding to the page.

10. The method of claim 1, wherein, determining a target page from the pages according to the page set, including: for each page in the page set, determining the scoring information of the page according to the page's page information; wherein the scoring information represents the degree of interest of the user to the page; sorting the pages according to the scoring information of each page in the page set, and determining the target page from the pages according to the sorting result.

11. The method of claim 10, wherein, determining the scoring information of the page according to the page's page information, including: obtaining the creation time of the page, and determining the weight information of the page according to the creation time; wherein the weight information represents the importance of the page; determining the scoring information of the page according to the page information and the weight information of the page.

12. A page recommendation apparatus characterized by comprising: including: An information acquisition unit is configured to acquire keyword information input by a user and determine an information type of the keyword information, wherein the keyword information represents a demand of the user for a page to be pushed, and the information type represents a form of the keyword information. A set determination unit is configured to determine a page set according to the keyword information and the information type of the keyword information, wherein the page set includes a plurality of pages related to the keyword information. A point acquisition unit is configured to acquire point information of each page in the page set, wherein the point information represents a situation of the page being consumed by collecting the page through a preset point in the page. A page recommendation unit is configured to determine a target page from the pages according to the point information of the pages in the page set, and push the target page to the user, wherein the target page is used to assist the user in creating a new page.

13. An electronic device, comprising: An apparatus includes: a memory and a processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory, so that the processor executes the method in any one of claims 1-11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are executed by the processor to implement the method in any one of claims 1-11.

15. A computer program product, comprising a computer program, wherein the computer program is executed by a processor to implement the method in any one of claims 1-11.

Images