Link data processing method and device, equipment, storage medium and program product

Abstract

The invention discloses a link data processing method and device, equipment, a storage medium and a program product, and relates to the technical field of data processing technologies and large language models.The method comprises the steps that to-be-processed target information is displayed, and the target information comprises at least one embedded link; identifying an embedding position of the embedded link in the target information; generating a prediction semantic tag corresponding to the embedding position based on the context content corresponding to the embedding position; on the basis of semantic analysis of link content corresponding to the embedded link by the predicted semantic tag, generating a summarized semantic tag and structured abstract content corresponding to the embedded link; and based on an index relationship between a pre-constructed predicted semantic tag and a summarized semantic tag, integrating the structured abstract content to an embedding position of the embedded link in the target information. By implementing the method and the device, the semantic visibility and the context readability of the embedded link are realized, and the bearing information of the embedded link can be efficiently obtained without repeatedly clicking and skipping.

Description

Technical Field

[0001] This disclosure relates to the fields of data processing and large language model technology, specifically to methods, apparatus, devices, storage media, and program products for processing linked data. Background Technology

[0002] During data processing on the data platform, business requirement documents, data request forms, or task descriptions often contain embedded links (such as Hive table addresses, dataset jump links, SQL query image links, visualization dashboard URLs, external document references, etc.), and these embedded links carry important information. However, due to the lack of contextual readability and semantic visibility, users need to repeatedly click, jump, and read to record key information, which is inefficient and detrimental to the knowledge accumulation of the data platform. Summary of the Invention

[0003] In view of this, the present disclosure provides a method, apparatus, device, storage medium, and program product for processing linked data to solve the problem of poor performance of linked content in embedded links.

[0004] In a first aspect, this disclosure provides a method for processing link data, comprising: displaying target information to be processed, the target information including at least one embedded link; in response to a triggering operation for any embedded link, identifying the embedding position of the embedded link in the target information; generating a predicted semantic tag corresponding to the embedding position based on the context content corresponding to the embedding position; generating a summary semantic tag and structured summary content corresponding to the embedded link based on the semantic parsing of the link content corresponding to the embedded link based on the predicted semantic tag; and in response to a content insertion operation triggered for the structured summary content, integrating the structured summary content into the embedding position of the embedded link in the target information based on the index relationship between the pre-built predicted semantic tag and the summary semantic tag.

[0005] Secondly, this disclosure provides a processing apparatus for link data, comprising: an information display module for displaying target information to be processed, the target information including at least one embedded link; an identification module for identifying the embedding position of the embedded link in the target information in response to a triggering operation for any embedded link; a semantic prediction module for generating a predicted semantic tag corresponding to the embedding position based on the context content corresponding to the embedding position; a semantic summary module for generating a summary semantic tag and structured summary content corresponding to the embedded link based on the semantic parsing of the link content corresponding to the embedded link using the predicted semantic tags; and a content integration module for integrating the structured summary content into the embedding position of the embedded link in the target information based on a pre-built index relationship between the predicted semantic tags and the summary semantic tags in response to a content insertion operation triggered for the structured summary content.

[0006] Thirdly, this disclosure provides an electronic device, including: a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the link data processing method of the first aspect or any corresponding embodiment described above.

[0007] Fourthly, this disclosure provides a computer-readable storage medium storing computer instructions for causing a computer to perform the link data processing method described in the first aspect or any corresponding embodiment.

[0008] Fifthly, this disclosure provides a computer program product, including computer instructions for causing a computer to execute the link data processing method described in the first aspect or any corresponding embodiment thereof.

[0009] The link data processing method, apparatus, device, storage medium, and program product provided in this disclosure automatically generate predicted semantic tags for embedded links by identifying the context content of the embedded position in the target information, and summarizes the link content based on the predicted semantic tags, generating corresponding summary semantic tags and structured summary content. This achieves semantic visibility for embedded links, eliminating the need for users to repeatedly click and jump, and efficiently obtaining the information carried by the embedded links. By constructing an index relationship between predicted semantic tags and summary semantic tags, the structured summary content is integrated into the embedded position in response to content insertion operations, ensuring the contextual readability of the embedded link and its embedded position, improving the information transparency of the embedded link, and facilitating knowledge accumulation of the link content. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the specific embodiments or related technologies of this disclosure, the accompanying drawings used in the description of the specific embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram illustrating an application scenario according to an embodiment of this disclosure; Figure 2 This is a flowchart illustrating a method for processing linked data according to an embodiment of the present disclosure; Figure 3 This is a schematic diagram illustrating the operation of structured digest content according to an embodiment of this disclosure; Figure 4 This is a flowchart illustrating another method for processing linked data according to an embodiment of the present disclosure; Figure 5 This is a flowchart illustrating another method for processing linked data according to an embodiment of the present disclosure; Figure 6 This is a schematic diagram of the embedded linking process of a business requirements document according to an embodiment of this disclosure; Figure 7 This is a structural block diagram of a data processing apparatus according to an embodiment of the present disclosure; Figure 8 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present disclosure. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0013] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0014] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0015] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0016] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0017] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0018] Currently, in data processing environments such as big data platforms, data middleware, business intelligence (BI) tools, and intelligent agent systems, the following problems mainly exist: (1) Embedded links such as Hive table addresses, dataset jump links, SQL query image links, visualization dashboard URLs, and external document references carry important business information, but due to the lack of contextual readability and semantic visibility, users need to click, jump, and read repeatedly, which is extremely inefficient, easy to miss, and has poor reusability; and it is not conducive to knowledge accumulation. (2) Only the original embedded link strings are retained, lacking context awareness and semantic interpretation capabilities, making it difficult to establish a connection with business intent, and difficult to provide semantic enhancement services such as field completion, link planning, and question-and-answer interaction, resulting in a gap in understanding and disconnection between upstream and downstream. Moreover, the extraction of image content only stays at the literal text level, without structured extraction and semantic mapping, and cannot support complex interactions. (3) The lack of structured summarization mechanisms and indexing methods makes it impossible to efficiently store, retrieve, and reuse link content, hindering the development of key capabilities such as large model-assisted analysis, content recommendation, and task planning.

[0019] Based on this, the disclosed technical solution constructs a complete processing chain of "understanding links from context, generating semantic summaries, and establishing a dual-key structure index." By automatically identifying the link context and generating semantic summaries, it efficiently obtains the information carried by the links without requiring users to repeatedly click and jump, significantly improving usage efficiency and information transparency. A structured summary generation mechanism driven by predictive semantic tags is introduced to combine contextual reasoning with link intent for intelligent question answering and knowledge enhancement. A dual-key structure is constructed for semantic retrieval, question answering, and content evolution.

[0020] As one optional application scenario of this disclosure embodiment, such as Figure 1 As shown, the optional application scenario includes a data processing tool 101, a data processing model 102, and an electronic device 103. The data processing tool 101 is deployed in the electronic device 103, and the data processing tool 101 communicates with the data processing model 102 to enable the invocation of the data processing model 102.

[0021] Data processing tool 101 provides an interactive interface to display inputs for target information and semantic understanding results for embedded links.

[0022] Data processing model 102 is a model for achieving semantic understanding capabilities, such as a large language model, machine learning model, multimodal model, etc.

[0023] Electronic device 103 can be a device with computing capabilities. For example, electronic device 103 may be equipped with a processor and memory, and may also be equipped with a dedicated accelerator (such as a graphics processing unit (GPU)). In addition, electronic device 103 can store and maintain data.

[0024] Examples of electronic device 103 may include supercomputers, personal computers, laptop computers, in-vehicle computing devices, mobile devices (such as smartphones, tablets, etc.), or combinations thereof. It should be understood that the electronic devices described herein are merely exemplary and not limiting; other different types of electronic devices may also be employed.

[0025] According to an embodiment of this disclosure, a method for processing linked data is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0026] This embodiment provides a method for processing linked data, which can be used in the aforementioned electronic devices, such as computers and tablets. Figure 2 This is a flowchart of a method for processing linked data according to an embodiment of this disclosure, such as... Figure 2 As shown, the process includes the following steps: Step S201: Display the target information to be processed, which includes at least one embedded link.

[0027] The target information is the information that the data processing tool needs to process. This target information can be business requirement documents, data request forms, task descriptions, etc. The data processing tool can be a big data platform, a data middle platform, a business intelligence (BI) tool, a data warehouse, etc.

[0028] Embedded links are links embedded in the target information. These embedded links can be Hive table addresses, dataset jump links, SQL query image links, visualization dashboard URLs, external document references, etc.

[0029] Specifically, the data processing tool provides an interactive page that can respond to user-triggered upload operations for target information. Accordingly, the data processing tool can obtain the target information uploaded by the user and display it in the interactive interface, and the target information includes one or more embedded links.

[0030] Step S202: In response to a trigger operation for any embedded link, identify the embedding position of the embedded link in the target information.

[0031] The embedding position refers to the location of the embedded link within the target information; the triggering operation refers to the click operation, selection operation, etc., triggered for the embedded link, and the embedded link currently selected by the user is determined through the triggering operation.

[0032] Specifically, the target information includes one or more embedded links, and the display method of the information corresponding to the embedded links is different from that of the information corresponding to the non-links. The data processing tool can respond to the user's trigger operation for any embedded link, call the pre-trained data processing model to perform semantic understanding of the target information, and combine regular expressions to identify the various types of embedded links from the content context of the target information, as well as the embedding position of each type of embedded link in the target information.

[0033] The data processing model can be trained based on a large language model or other models; no specific limitation is made here.

[0034] Step S203: Generate predicted semantic labels corresponding to the embedding positions based on the context content corresponding to the embedding positions.

[0035] Predicted semantic tags are used to characterize the semantics of the content to be embedded at the embedding location; that is, the embedded content and the context content must have semantic coherence and consistency. A context window is constructed centered on the embedding location, and the content within this context window constitutes the context content. Specifically, based on the data processing model's token budget (e.g., 4000 tokens), syntactic structure, and semantic density, the context window corresponding to the embedding location is dynamically determined, centered on the embedding location, to ensure the completeness of the context content input to the data processing model. If the context window exceeds a preset range, it is compressed.

[0036] Furthermore, the data processing model predicts the embedded content that should be present at the embedding location by understanding the complete context, and obtains the corresponding predicted semantic label predicted_key.

[0037] Step S204: Based on the semantic parsing of the content corresponding to the embedded link using the predicted semantic tags, generate the summary semantic tags and structured summary content corresponding to the embedded link.

[0038] Semantic tags are used to characterize the semantics of the content extracted from the links, and structured summary content is a summary of the content extracted from the links. After determining the predicted semantic tags, the link content corresponding to the embedded links is accessed. The predicted semantic tag `predicted_key` serves as a prompt to guide the data processing model to perform semantic parsing on the link content. The model then understands the semantic content associated with the predicted semantic tag `predicted_key` from the link content, summarizes the understood semantic content, and generates the corresponding summary semantic tag `summary_key` and its corresponding structured summary content.

[0039] Step S205: In response to the content insertion operation triggered for the structured summary content, the structured summary content is integrated into the embedding position of the embedded link in the target information based on the index relationship between the pre-built predicted semantic tags and the summary semantic tags.

[0040] The index relationship represents a mapping based on the association between the predicted semantic tag `predicted_key` and the summary semantic tag `summary_key`. That is, the predicted semantic tag `predicted_key` can be mapped to the summary semantic tag `summary_key`, and the summary semantic tag `summary_key` can be used to locate the predicted semantic tag `predicted_key`.

[0041] As described above, the summary semantic tag `summary_key` is generated based on the predicted semantic tag `predicted_key`. Therefore, there is a correlation between the predicted semantic tag `predicted_key` and the summary semantic tag `summary_key`. Based on this correlation, an index relationship can be built between the predicted semantic tag `predicted_key` and the summary semantic tag `summary_key`. Under this index relationship, the structured summary content corresponding to the summary semantic tag `summary_key` can be indexed through the predicted semantic tag `predicted_key`, and the context content corresponding to the predicted semantic tag `predicted_key` can also be located through the summary semantic tag `summary_key`.

[0042] Content insertion is a user-triggered operation that inserts structured summary content into the target information. For example... Figure 3As shown, target information 31 is displayed in the left area 301 of the interactive interface 30. An embedded link 32 is set within the target information 31. After the user triggers the embedded link 32, the structured summary content 33 corresponding to the summary semantic tag `summary_key` is obtained by combining the index relationship between the predicted semantic tag and the summary semantic tag, and this structured summary content 33 is displayed in the right area 302 of the interactive interface 30. Simultaneously, an insertion control 34 is provided in the right area 302, which the user can click to trigger a content insertion operation. Accordingly, the data processing tool can respond to the user's content insertion operation triggered by the structured summary content, integrating the structured summary content corresponding to the summary semantic tag `summary_key` into the embedded position within the target information, thus achieving readability of the embedded link and its context, as well as semantic visibility of the embedded link.

[0043] The link data processing method provided in this embodiment automatically generates predicted semantic tags for embedded links by identifying the context of their embedding positions in target information. Based on these predicted semantic tags, the link content is summarized, generating corresponding summary semantic tags and structured summary content. This achieves semantic visibility for embedded links, eliminating the need for users to repeatedly click to navigate and efficiently accessing the information carried by the embedded links. By constructing an index relationship between predicted and summary semantic tags, structured summary content is integrated into the embedding position of the embedded link in response to content insertion operations. This ensures the readability of the context between the embedded link and its embedding position, improves the information transparency of the embedded link, and facilitates knowledge accumulation based on the content of the embedded links.

[0044] This embodiment provides a method for processing linked data, which can be used in the aforementioned electronic devices, such as computers and tablets. Figure 4 This is a flowchart of a method for processing linked data according to an embodiment of this disclosure, such as... Figure 4 As shown, the process includes the following steps: Step S301: Display the target information to be processed, which includes at least one embedded link. For details, please refer to the relevant descriptions of the corresponding steps in the embodiments shown above; they will not be repeated here.

[0045] Step S302: In response to a triggering operation for any of the embedded links, identify the embedding position of the embedded link in the target information.

[0046] Specifically, step S302 includes: Step S3021: Based on the content structure of the target information, perform semantic segmentation on the content of the target information to obtain multiple semantic segments.

[0047] The content structure refers to the structure of the content contained in the target information, specifically including syntactic structure, paragraph structure, semantic density, and heading format structure (such as H1-H3 hierarchy). The data processing model, by invoking the segmentation scheduler Doc-split and guided by semantic segmentation prompts, parses the content structure of the target information. It then combines semantic vector similarity and content structure to perform heading understanding, semantic window recognition, and paragraph classification, resulting in multiple semantic segments. A unified title anchor node structure, TitleNode, is constructed for each semantic segment to facilitate the integration of predicted semantic tags (predicted_key) and summary semantic tags (summary_key) as semantic anchors. The semantic segmentation prompts are pre-defined prompts based on a prompting word mechanism, used to control the data processing model's execution of the semantic segmentation process.

[0048] In some optional implementations, step S3021 above includes: Step a1: Based on the content structure of the target information, the content of the target information is divided into multiple initial segments.

[0049] Step a2: Determine the semantic similarity between adjacent initial segments based on the segment content corresponding to each initial segment.

[0050] Step a3: If the semantic similarity exceeds the first preset threshold, then the adjacent initial segments are merged into a semantic segment.

[0051] The initial fragment is a segment obtained by initially dividing the target information according to its content structure. For example, if the target information is a requirements document, the initial fragment is a series of paragraphs obtained by initially dividing the requirements document. The first preset threshold is a pre-set minimum semantic similarity that represents the semantic consistency between adjacent initial fragments, such as 85% or 90%.

[0052] The data processing model divides the target information into multiple initial segments based on the syntactic structure, paragraph structure, and heading format structure (such as H1-H3 hierarchy) within the content structure. For each initial segment, the semantic similarity between adjacent initial segments is further calculated.

[0053] The calculated semantic similarity between each adjacent initial segment is compared with a first preset threshold to determine whether the semantic similarity exceeds the first preset threshold. If the semantic similarity exceeds the first preset threshold, it indicates that the adjacent initial segments have semantic consistency and are not suitable for being divided into two segments. In this case, the adjacent initial segments with semantic similarity exceeding the first preset threshold can be merged to obtain a semantic segment after merging the adjacent initial segments.

[0054] In the above implementation, the semantic similarity between adjacent initial segments is calculated to determine whether they should be merged into a single semantic segment, thereby improving the semantic consistency of the semantic segments.

[0055] Step S3022: Identify the content of each semantic segment, determine the target semantic segment with embedded links and the embedding position of the embedded links in the target semantic segment.

[0056] The algorithm iterates through each semantic segment sequentially, using a combination of regular expressions and the natural language processing capabilities of the data processing model to identify embedded links (such as HTTP links, image paths, SQL graphs, etc.) within each segment. If an embedded link is found, it is designated as the target semantic segment, and the embedding position of the embedded link is marked within the target semantic segment.

[0057] In some optional implementations, the above method further includes: Step b1: Determine whether the content of the target semantic segment exceeds the second preset threshold.

[0058] Step b2: If the content of the target semantic fragment exceeds a preset threshold, the target semantic fragment is split into multiple sub-fragments, and sub-tags corresponding to each sub-fragment are generated.

[0059] Step b3 involves nesting the original tags corresponding to the target semantic fragment with each sub-tag to generate a hierarchical nested structure between the target semantic fragment and each sub-fragment.

[0060] The second preset threshold is a pre-defined maximum content size allowed for a semantic fragment, such as 300 words or 2000 token characters. After merging adjacent semantic fragments, the content size of the merged target semantic fragment is checked to determine if it exceeds the second preset threshold. If the content size exceeds the threshold, the target semantic fragment is split based on the Markdown structure to obtain multiple sub-fragments. Simultaneously, a data processing model performs semantic understanding on the content of each sub-fragment to obtain sub-tags. Each sub-tag serves as a semantic supplement to the original tag corresponding to the target semantic fragment, forming a hierarchical nested structure between the original tag and the sub-tags, such as "## Original Tag → ###predicted_key Sub-tag". Each sub-tag acts as a semantic window segmentation point, through which the target semantic fragment is segmented into its sub-fragments.

[0061] In the above implementation, the Markdown structure is called to dynamically segment the content in the target semantic fragment and automatically generate sub-tags as semantic anchors to adapt to the segmentation of semantic fragments in different target information structures, which greatly reduces the burden of manual organization and lowers the cost of human intervention.

[0062] Step S303: Generate predicted semantic labels corresponding to the embedding position based on the context content corresponding to the embedding position.

[0063] Specifically, step S303 includes: Step S3031: Based on the context content corresponding to the embedded position, obtain the link type corresponding to the embedded link.

[0064] Based on the context content corresponding to the embedded location, a data processing model is used to infer the user's expected intent or target behavior when accessing the embedded link, and the link type corresponding to the embedded link is determined by combining the expected intent or target behavior.

[0065] Step S3032: Generate multiple candidate semantic tags at the embedding location according to the link type.

[0066] Candidate semantic tags are semantic tags generated by predicting the content of the embedding location. Specifically, different link types have different tag formats. After determining the tag format that matches the link type, candidate semantic tags are generated in one or more rounds according to the context content corresponding to the embedding location, based on the tag format and multiple semantic levels such as sentence, paragraph, and global.

[0067] In some optional implementations, step S3032 includes: performing multi-level semantic parsing of the context content based on the tag generation format corresponding to the link type to obtain multiple candidate semantic tags corresponding to the tag generation format.

[0068] The tag generation format is the tag format of the candidate semantic tags. Different tag generation formats correspond to different link types, as shown in Table 1. Based on the link type, the context content corresponding to the embedding position is semantically parsed according to the tag generation format. Candidate semantic tags are generated in one or more rounds from multiple semantic levels to ensure the accuracy of candidate semantic tag generation.

[0069] Table 1. Label Generation Format Table

[0070] Step S3033: Fuse multiple candidate semantic tags according to a preset fusion strategy to generate predicted semantic tags.

[0071] The preset fusion strategy is a pre-defined fusion strategy for candidate semantic tags. Specifically, it may include clustering and merging multiple candidate semantic tags based on semantic similarity; fusing each candidate semantic tag using its associated semantic confidence and contextual source; performing multi-vote fusion for each candidate semantic tag; generating group identifiers for candidate semantic tags using hash algorithms or clustering methods, and then indexing and classifying each candidate semantic tag using these group identifiers. No specific limitations are imposed on the fusion strategy used here.

[0072] Multiple candidate semantic labels will be identified and fused according to a preset fusion strategy. The final predicted semantic label will be determined by combining the fusion results of the candidate semantic labels.

[0073] Step S304: Based on the semantic parsing of the content corresponding to the embedded links using the predicted semantic tags, generate summary semantic tags and structured summary content corresponding to the embedded links. For details, please refer to the relevant descriptions of the steps in the embodiments shown above, which will not be repeated here.

[0074] Step S305: In response to the content insertion operation triggered for the structured summary content, based on the index relationship between the pre-built predicted semantic tags and the summarized semantic tags, the structured summary content is integrated into the embedding position of the embedded link in the target information. For details, please refer to the relevant descriptions of the corresponding steps in the embodiments shown above, which will not be repeated here.

[0075] In some alternative implementations, the method further includes generating a corrected target summary in response to a correction operation on the structured summary content.

[0076] As described above, after the structured summary content is generated, it will be displayed on the interactive interface 30 for the user to view. If the structured summary content is incorrect or inappropriate, the user can correct it to obtain the corrected target summary content.

[0077] Specifically, such as Figure 3 As shown, a correction control 35 is provided in the right-hand area 302 of the interactive interface 30. The user can click the correction control 35 to trigger a correction operation on the structured summary content. Correspondingly, the data processing tool can respond to the user's correction operation on the structured summary content, making the structured summary content editable. Then, the user can correct the structured summary content to obtain the corrected target summary content.

[0078] In the above implementation, it is possible to correct the structured summary content to ensure the accuracy of the summary content corresponding to the embedded link.

[0079] In some optional implementations, the above method further includes: Step c1: Extract the correction content from the target summary.

[0080] Step c2 involves updating the generation strategies for predicted semantic tags, summarized semantic tags, and structured summary content using the corrected content.

[0081] The correction content refers to the user's modifications to the structured summary content. This correction content can be determined by comparing the structured summary content with the target summary content. Since the correction content is determined by correcting inappropriate structured summary content, after the correction content is extracted, it can be used to update the generation strategies for predicted semantic tags, summarized semantic tags, and structured summary content. This ensures that accurate structured summary content is generated according to the index relationship between predicted and summarized semantic tags.

[0082] Specifically, the corrected target summary content is compared with the original structured summary content to determine the corrected content. The data processing tool collects the corrected content from users and provides it to the data processing model for learning. This allows the data processing model to retrain the prompt word mechanism and adjust the generation strategy. The data processing model then generates predicted semantic tags, summarized semantic tags, and structured summary content based on the optimized generation strategy, ensuring the semantic consistency and generation accuracy of the predicted semantic tags, summarized semantic tags, and structured summary content.

[0083] In the above implementation, by modifying the structured summary content, the generation strategy can be optimized using the modified content. This enables automatic regeneration or prompting for correction of poor-quality information predictions, thereby achieving continuous iterative improvement and learning evolution of the quality of the structured summary content.

[0084] In some optional implementations, the method further includes: in response to an operation of viewing relevant information for the structured summary content, displaying relevant information corresponding to the structured summary content, wherein the relevant information is used to indicate the generation of the structured summary content.

[0085] The "View Related Information" operation allows you to examine relevant information within the structured summary content, such as metadata probing and categorical search. For example... Figure 3As shown, when viewing related controls 36 are set in the right-hand area 302 of the interactive interface 30, the user can click on the related controls 36 to trigger the viewing of relevant information for the structured summary content. Correspondingly, the data processing tool can respond to the user's click on the related controls 36, jumping from the interactive interface 30 to the relevant information display page 37, and displaying the relevant information corresponding to the structured summary content on the relevant information display page 37. This relevant information indicates the generation of the structured summary content. This relevant information may include metadata probing, SQL statement generation, and caliber search, etc.

[0086] In the above embodiments, it is supported to view relevant information of the structured summary content, so as to determine the generation process of the structured summary content by viewing the relevant information, thereby enhancing the understanding of the generation of structured summary content and improving the transparency of the generation of structured summary content.

[0087] The link data processing method provided in this embodiment divides the semantic content of the target information into multiple semantic segments to determine the target semantic segments containing embedded links and the embedding positions of the embedded links within these segments, thereby achieving accurate identification of embedded links. Candidate semantic tags are generated for the link types corresponding to the embedded links, and these candidate semantic tags are fused to obtain corresponding predicted semantic tags. Based on the context, the link intent of the embedded links is automatically inferred, generating structured predicted semantic tags (i.e., predicted_keys). This avoids manual annotation, achieves context-driven link semantic understanding, significantly improves the readability and accessibility of link content, and enhances the data processing model's ability to understand and utilize unstructured link content.

[0088] This embodiment provides a method for processing linked data, which can be used in the aforementioned electronic devices, such as computers and tablets. Figure 5 This is a flowchart of a method for processing linked data according to an embodiment of this disclosure, such as... Figure 5 As shown, the process includes the following steps: Step S401: Display the target information to be processed, which includes at least one embedded link. For details, please refer to the relevant descriptions of the corresponding steps in the embodiments shown above; they will not be repeated here.

[0089] Step S402: In response to a trigger operation for any embedded link, identify the embedding position of the embedded link in the target information. For details, please refer to the relevant descriptions of the corresponding steps in the embodiments shown above, which will not be repeated here.

[0090] Step S403: Based on the context content corresponding to the embedding position, generate the predicted semantic label corresponding to the embedding position. For details, please refer to the relevant descriptions of the steps in the embodiments shown above, which will not be repeated here.

[0091] Step S404: Based on the semantic parsing of the content corresponding to the embedded link using the predicted semantic tags, generate the summary semantic tags and structured summary content corresponding to the embedded link.

[0092] Specifically, step S404 includes: Step S4041: Obtain summary generation prompt information.

[0093] The summary generation prompts are messages built upon a prompt word mechanism to generate summaries. Specifically, these prompts can be constructed using prompt word templates corresponding to the prompt word mechanism, and the data processing model can respond to the user's request to construct the prompts and obtain the corresponding summary generation prompts.

[0094] Step S4042: Using predicted semantic tags and summaries to generate prompt information, guide the semantic parsing process for the linked content, and extract at least one summary semantic tag corresponding to the embedded link and the structured summary content corresponding to each summary semantic tag from the linked content.

[0095] Based on the generated predicted semantic tag `predicted_key` and the summary generation prompt information constructed based on the prompt word mechanism, the data processing model is guided to access the link content in the embedded links and perform semantic parsing on the link content. The associated content of the predicted semantic tag `predicted_key` is extracted from the link content, and the associated content is semantically understood to obtain one or more summary semantic tags `summary_key` that match the associated content. At the same time, the associated content corresponding to each summary semantic tag `summary_key` is semantically summarized to obtain the structured summary content corresponding to each summary semantic tag `summary_key`.

[0096] For a single predicted semantic tag (predicted_key), there can be multiple summary semantic tags (summary_key). In other words, for a single predicted semantic tag (predicted_key), multiple summary semantic tags (summary_key) can coexist.

[0097] In some optional implementations, the above method further includes: dynamically updating the summary semantic tags based on the context content corresponding to the embedding location.

[0098] Since the summary semantic tag `summary_key` is generated based on the predicted semantic tag `predicted_key`, which in turn is predicted based on the context content, and the target information can be updated by the user, the context content at the embedding location of the embedded link can also be updated. Therefore, the predicted semantic tag `predicted_key` is regenerated based on the updated context content, and the summary semantic tag `summary_key` is dynamically updated based on the regenerated predicted semantic tag `predicted_key`. This allows for the tracking of the time-series evolution of the summary semantic tag `summary_key`, enabling the observation of trends in target information over time.

[0099] In some optional implementations, the above method further includes: Step d1: Obtain the semantic similarity between the predicted semantic label and the summarized semantic label.

[0100] In step d2, if the semantic similarity exceeds the similarity threshold, a semantic deviation label is generated, and the content in the target information is re-segmented to update the predicted semantic label.

[0101] Step d3: Regenerate the summary semantic labels based on the updated predicted semantic labels.

[0102] The predicted semantic label `predicted_key` and the summary semantic label `summary_key` are semantically vectorized to obtain predicted semantic label vectors and summary semantic label vectors. Semantic similarity is calculated by combining the predicted and summary semantic label vectors, and the calculated semantic similarity is compared with a similarity threshold to determine whether the semantic similarity exceeds the threshold.

[0103] If the semantic similarity exceeds a similarity threshold, it indicates a significant semantic deviation between the predicted semantic label and the summarized semantic label. In this case, deviation marker generation is automatically triggered, allowing the data processing model to re-segment the content of the target information. The re-segmented semantic fragments are then used to update and generate the predicted semantic labels. Subsequently, based on the updated predicted semantic labels, a prompt word mechanism is used to construct a summary and generate prompt information. Finally, the data processing model extracts the summary semantic label `summary_key` and its corresponding structured summary content from the linked content.

[0104] In the above implementation, based on the semantic deviation between the predicted semantic label and the summarized semantic label, the predicted semantic label is automatically updated, realizing an iterative mechanism that uses the summary semantic label as feedback to correct the predicted semantic label, so as to reconstruct the index relationship between the predicted semantic label and the summarized semantic label and realize semantic self-repair.

[0105] Step S405: In response to the content insertion operation triggered for the structured summary content, based on the index relationship between the pre-built predicted semantic tags and the summarized semantic tags, the structured summary content is integrated into the embedding position of the embedded link in the target information. For details, please refer to the relevant descriptions of the corresponding steps in the embodiments shown above, which will not be repeated here.

[0106] In some optional implementations, the above method further includes: Step e1: Align the predicted semantic labels and the summarized semantic labels based on the index relationship to generate a dual-label alignment structure.

[0107] Step e2: Store the target information based on the dual-label alignment structure.

[0108] Based on the association between the predicted semantic tag `predicted_key` and the summary semantic tag `summary_key`, an index perspective based on the predicted semantic tag `predicted_key` and an entity perspective based on the summary semantic tag `summary_key` are constructed, forming a dual-tag alignment structure. This dual-tag alignment structure supports a one-to-many mapping between the predicted semantic tag `predicted_key` and the summary semantic tag `summary_key` (one predicted semantic tag `predicted_key` can map to multiple summary semantic tags `summary_key`). Subsequently, the constructed dual-tag alignment structure is used as an information storage structure to perform storage of target information.

[0109] In the above implementation, a dual-label alignment structure is constructed to support semantic aggregation, version tracking, and semantic backtracking across documents and tasks, thereby providing support for the construction of knowledge graphs and long-term intelligent agent applications.

[0110] In some optional implementations, the above method further includes: Step f1: In response to a retrieval operation triggered for the target information, obtain retrieval request information.

[0111] Step f2 involves performing a multi-dimensional retrieval of the content in the target information based on the dual-label alignment structure to obtain the target content that matches the retrieval request information.

[0112] A retrieval operation is an action triggered by performing a search on the content of target information; retrieval request information is the information generated when the retrieval operation is triggered. Specifically, when a retrieval is triggered for target information, the retrieval request information required for the current retrieval is obtained, and a multi-dimensional retrieval is performed on the content of the target information according to the retrieval request information, that is, the content of the target information is retrieved from multiple dimensions to obtain the target content that matches the retrieval request information.

[0113] During the retrieval process, the predicted semantic tag (predicted_key) dominates the semantic indexing path, while the summary semantic tag (summary_key) serves as the fact feedback node, participating in the evolution of clustering, correction, and data processing tools for the predicted semantic tag (predicted_key).

[0114] In a specific example, the summary semantic tag `summary_key` corresponding to the predicted semantic tag `predicted_key` and the structured summary content can be determined based on the dual-tag alignment structure. Enhanced retrieval is then performed based on the predicted semantic tag `predicted_key` and its corresponding structured summary content.

[0115] In another specific example, the predicted semantic tag `predicted_key` and the context content corresponding to the summary semantic tag `summary_key` can be determined based on the dual-tag alignment structure. Therefore, knowledge question answering retrieval can be performed by combining the summary semantic tag `summary_key` and its corresponding context content.

[0116] In another specific example, the dual-label alignment structure can also be linked with the intelligent agent to enable the intelligent agent to perform tasks such as field completion, SQL generation, and link planning.

[0117] Of course, other dimensions of content retrieval can also be performed, which are not specifically limited here.

[0118] The above implementation method performs multi-dimensional content retrieval through a dual-label alignment structure, uses the structured predicted semantic label predicted_key as the query entry point, and uses the summary semantic label summary_key to provide highly reliable content entities, thereby improving the context awareness and retrieval efficiency in multi-dimensional retrieval, which is beneficial to improving the calling accuracy and retrieval coverage of downstream intelligent systems.

[0119] The link data processing method provided in this embodiment combines predicted semantic tags and summary generation prompts to guide the generation of summary semantic summaries and structured summary content, ensuring that the structured summary content has interpretability and semantic consistency, which facilitates intelligent question answering and knowledge enhancement for target information.

[0120] As a specific application embodiment of this disclosure, taking a business requirement document as an example, the data processing system of the data processing tool includes a document parsing module, a context understanding module, a content acquisition module, a summary generation module, a dual-key storage module, and a user interaction and feedback optimization module. The document parsing module automatically identifies all embedded links in the business requirements document and extracts the context content corresponding to each embedded link. The context understanding module constructs a dynamic context window, adaptively controlling the length of the content input to the data processing model based on the paragraph semantic density and content structure of the business requirements document to ensure semantic integrity. The content acquisition module asynchronously crawls link content from web pages, API interfaces, images, PDFs / Words, etc., and extracts effective information in a structured manner. The summary generation module generates a predicted semantic tag `predicted_key` based on the context content, and generates a summary semantic tag `summary_key` and structured summary content corresponding to the embedded links. The dual-key storage module constructs a knowledge graph structure with the predicted semantic tag `predicted_key` as the query index and the summary semantic tag `summary_key` as the content entity. The user interaction and feedback optimization module collects user confirmations or corrections regarding the generated summary content, and the feedback is used to optimize the prompt word mechanism and tag generation strategy of the data processing model.

[0121] The process of processing business requirements documents is as follows: Figure 6 As shown, the business requirements document is input into the data processing platform. The platform then uses a data processing model to identify embedded links and their corresponding contextual content within the document. Doc-Split is used to segment the document content, allowing for reasoning about the content at the embedding locations of the embedded links to obtain corresponding predicted semantic tags. These predicted semantic tags support feedback correction.

[0122] Furthermore, the content corresponding to the embedded links is crawled, and predicted semantic tags and summary generation prompts are used to guide the semantic understanding of the link content, generating summary semantic tags and their summary content for the link content. A dual-tag structure is constructed based on the predicted semantic tags and summary semantic tags for subsequent retrieval. At the same time, semantic similarity deviation detection is performed based on the predicted semantic tags and summary semantic tags to correct either the predicted semantic tags or the summary semantic tags.

[0123] This embodiment also provides a data processing apparatus for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0124] This embodiment provides a data processing device for linking data, such as... Figure 7 As shown, it includes: The information display module 501 is used to display target information to be processed, the target information including at least one embedded link.

[0125] The identification module 502 is used to identify the embedding position of the embedded link in the target information in response to a trigger operation for any embedded link.

[0126] The semantic prediction module 503 is used to generate predicted semantic tags corresponding to the embedding position based on the context content corresponding to the embedding position.

[0127] The semantic summary module 504 is used to perform semantic parsing of the content corresponding to the embedded links based on the predicted semantic tags, and to generate summary semantic tags and structured summary content for the embedded links.

[0128] The content integration module 505 is used to respond to a content insertion operation triggered for the structured summary content, and integrate the structured summary content into the embedding position of the embedded link in the target information based on the index relationship between the pre-built predicted semantic tags and the summary semantic tags.

[0129] In some alternative embodiments, the above-described apparatus further includes: The correction module is used to generate the corrected target summary content in response to correction operations on the structured summary content.

[0130] In some alternative embodiments, the above-described apparatus further includes: The extraction module is used to extract the corrected content from the target summary.

[0131] The strategy update module is used to update the generation strategies for predicted semantic tags, summarized semantic tags, and structured summary content using the corrected content.

[0132] In some alternative embodiments, the above-described apparatus further includes: The information viewing module is used to respond to the relevant information viewing operation for the structured summary content, and displays the relevant information corresponding to the structured summary content. The relevant information is used to indicate the generation of the structured summary content.

[0133] In some alternative implementations, the identification module 502 includes: The semantic segmentation unit is used to perform semantic segmentation on the content of the target information based on the content structure of the target information, and obtain multiple semantic segments.

[0134] The location determination unit is used to identify the content of each semantic segment, determine the target semantic segment with embedded links and the embedding position of the embedded links in the target semantic segment.

[0135] In some optional implementations, the semantic segmentation unit includes: The initial segmentation unit is used to divide the content of the target information into multiple initial segments based on the content structure of the target information.

[0136] The semantic similarity determination subunit is used to determine the semantic similarity between adjacent initial segments based on the segment content corresponding to each initial segment.

[0137] The merging subunit is used to merge adjacent initial segments into a semantic segment if the semantic similarity exceeds a first preset threshold.

[0138] In some alternative embodiments, the above-described apparatus further includes: The content quantity judgment module is used to determine whether the content quantity of the target semantic segment exceeds the second preset threshold.

[0139] The segment splitting module is used to split the target semantic segment into multiple sub-segments if the content size of the target semantic segment exceeds a preset threshold, and generate sub-tags corresponding to each sub-segment.

[0140] The nested module is used to nest the original tags corresponding to the target semantic fragment with each sub-tag, generating a hierarchical nested structure between the target semantic fragment and each sub-fragment.

[0141] In some optional implementations, the semantic prediction module 503 includes: The type determination unit is used to obtain the link type corresponding to the embedded link based on the context content corresponding to the embedding position.

[0142] The candidate generation unit is used to generate multiple candidate semantic tags at the embedding location according to the link type.

[0143] The fusion unit is used to fuse multiple candidate semantic tags according to a preset fusion strategy to generate predicted semantic tags.

[0144] In some optional implementations, the candidate generation unit is specifically used to perform multi-level semantic parsing of the context content based on the tag generation format corresponding to the link type, so as to obtain multiple candidate semantic tags corresponding to the tag generation format.

[0145] In some alternative implementations, the semantic summarization module 504 includes: The abstract prompt acquisition unit is used to acquire abstract generation prompt information.

[0146] The guiding unit is used to generate prompt information using predicted semantic tags and summaries to guide the semantic parsing process for the linked content, extracting at least one summary semantic tag corresponding to the embedded link and the structured summary content corresponding to each summary semantic tag from the linked content.

[0147] In some alternative embodiments, the above-described apparatus further includes: The update module is used to dynamically update the summary semantic tags based on the context content corresponding to the embedding position.

[0148] In some alternative embodiments, the above-described apparatus further includes: The tag similarity acquisition module is used to obtain the semantic similarity between the predicted semantic tags and the summarized semantic tags.

[0149] The first update module is used to generate semantic deviation tags if the semantic similarity exceeds the similarity threshold, and to re-segment the content in the target information and update the predicted semantic labels.

[0150] The second update module is used to regenerate the summary semantic labels based on the updated predicted semantic labels.

[0151] In some alternative embodiments, the above-described apparatus further includes: The alignment module is used to align the predicted semantic labels and the summarized semantic labels based on the index relationship, generating a dual-label alignment structure.

[0152] The storage module is used to store target information based on a dual-label alignment structure.

[0153] In some alternative embodiments, the above-described apparatus further includes: The information retrieval module is used to retrieve retrieval request information in response to retrieval operations triggered for target information.

[0154] The retrieval module is used to perform multi-dimensional retrieval of the content in the target information based on the dual-label alignment structure, and obtain the target content that matches the retrieval request information.

[0155] The link data processing apparatus provided in this disclosure can execute the link data processing method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of executing the method.

[0156] By identifying the context of embedded links within target information, predictive semantic tags are automatically generated for each embedded link. Based on these predictive semantic tags, the link content is summarized, generating corresponding summary semantic tags and structured summary content. This achieves semantic visibility for embedded links, eliminating the need for users to repeatedly click and navigate, and efficiently retrieving the information carried by the embedded links. By constructing an index relationship between predictive and summary semantic tags, structured summary content is integrated into the embedding location of the embedded link in response to content insertion operations. This ensures the readability of the context between the embedded link and its embedding location, improves the information transparency of embedded links, and facilitates knowledge accumulation regarding the content of embedded links.

[0157] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0158] Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure.

[0159] The following is a detailed reference. Figure 8 This diagram illustrates a suitable structural schematic for implementing an electronic device according to embodiments of the present disclosure. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 601, which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) 602 or a program loaded from memory 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of the electronic device. The processor 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0160] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 8 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0161] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a memory 608, or installed from a ROM 602. When the computer program is executed by the processor 601, it performs the functions defined in the linked data processing method of embodiments of this disclosure.

[0162] Figure 8 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0163] This disclosure also provides a computer-readable storage medium in which the methods described in this disclosure can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium may also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the linked data processing methods shown in the above embodiments.

[0164] A portion of this disclosure can be applied to computer program products, such as computer program instructions, which, when executed by a computer, can invoke or provide methods and / or technical solutions according to this disclosure through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, and installation package files. Accordingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions; the computer compiling the instructions and then executing the corresponding compiled program; the computer reading and executing the instructions; or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0165] Although embodiments of the present disclosure have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for processing linked data, characterized in that, The method includes: Display target information to be processed, the target information including at least one embedded link; In response to a triggering operation for any of the embedded links, the embedding position of the embedded link in the target information is identified; Based on the context content corresponding to the embedding position, a predicted semantic label corresponding to the embedding position is generated; Based on the semantic parsing of the content corresponding to the embedded link using the predicted semantic tags, a summary semantic tag and structured summary content corresponding to the embedded link are generated. In response to a content insertion operation triggered for the structured summary content, the structured summary content is integrated into the embedding position of the embedded link in the target information based on the index relationship between the pre-built predicted semantic tags and the summarized semantic tags.

2. The method according to claim 1, characterized in that, Also includes: In response to the correction operation on the structured summary content, a corrected target summary content is generated.

3. The method according to claim 2, characterized in that, Also includes: Extract the correction content from the target summary content; The generation strategy for the predicted semantic tags, the summarized semantic tags, and the structured summary content is updated using the corrected content.

4. The method according to claim 1 or 2, characterized in that, Also includes: In response to an operation to view relevant information for the structured summary content, relevant information corresponding to the structured summary content is displayed, and the relevant information is used to indicate the generation of the structured summary content.

5. The method according to claim 1, characterized in that, The process of identifying the embedding position of the embedded link in the target information includes: Based on the content structure of the target information, the content of the target information is semantically segmented to obtain multiple semantic segments; The fragment content in each of the semantic segments is identified to determine the target semantic segment containing the embedded link and the embedding position of the embedded link in the target semantic segment.

6. The method according to claim 5, characterized in that, Based on the content structure of the target information, the content of the target information is semantically segmented to obtain multiple semantic segments, including: Based on the content structure of the target information, the content of the target information is divided into multiple initial segments; Based on the content of each initial segment, the semantic similarity between adjacent initial segments is determined; If the semantic similarity exceeds a first preset threshold, then adjacent initial segments are merged into the semantic segment.

7. The method according to claim 5 or 6, characterized in that, Also includes: Determine whether the content size of the target semantic segment exceeds a second preset threshold; If the content of the target semantic segment exceeds the second preset threshold, the target semantic segment is split into multiple sub-segments, and sub-tags corresponding to each sub-segment are generated; The original tag corresponding to the target semantic fragment is nested with each of the sub-tags to generate a hierarchical nested structure between the target semantic fragment and each of the sub-fragments.

8. The method according to claim 1, characterized in that, The step of generating a predicted semantic label corresponding to the embedding position based on the context content corresponding to the embedding position includes: Based on the context content corresponding to the embedding location, the link type corresponding to the embedded link is determined; Multiple candidate semantic tags are generated at the embedding location according to the link type; The multiple candidate semantic tags are fused according to a preset fusion strategy to generate the predicted semantic tag.

9. The method according to claim 8, characterized in that, Based on the context content corresponding to the embedding location, multiple candidate semantic tags are generated at the embedding location according to the link type, including: Based on the tag generation format corresponding to the link type, the context content is subjected to multi-level semantic parsing to obtain multiple candidate semantic tags corresponding to the tag generation format.

10. The method according to claim 1, characterized in that, The semantic parsing of the content corresponding to the embedded link based on the predicted semantic tags, generating summary semantic tags and structured summary content corresponding to the embedded link, includes: Obtain summary generation prompt message; Using the predicted semantic tags and the summary to generate prompt information, the semantic parsing process for the link content is guided, and at least one summary semantic tag corresponding to the embedded link and the structured summary content corresponding to each summary semantic tag are extracted from the link content.

11. The method according to claim 10, characterized in that, Also includes: The summary semantic tags are dynamically updated based on the context content corresponding to the embedding position.

12. The method according to claim 10 or 11, characterized in that, Also includes: Obtain the semantic similarity between the predicted semantic label and the summarized semantic label; If the semantic similarity exceeds the similarity threshold, a semantic deviation marker is generated, and the content in the target information is re-segmented to update the predicted semantic label; The summarized semantic labels are regenerated based on the updated predicted semantic labels.

13. The method according to claim 1, characterized in that, Also includes: Based on the index relationship, the predicted semantic labels and the summarized semantic labels are aligned to generate a dual-label alignment structure; The target information is stored based on the dual-label alignment structure.

14. The method according to claim 13, characterized in that, Also includes: In response to a retrieval operation triggered for the target information, retrieval request information is obtained; Based on the dual-label alignment structure, a multi-dimensional retrieval is performed on the content in the target information to obtain the target content that matches the retrieval request information.

15. A data processing apparatus for linking data, characterized in that, The device includes: An information display module is used to display target information to be processed, the target information including at least one embedded link; The identification module is configured to identify the embedding position of the embedded link in the target information in response to a triggering operation for any one of the embedded links; The semantic prediction module is used to generate a predicted semantic label corresponding to the embedding position based on the context content corresponding to the embedding position. The semantic summary module is used to perform semantic parsing of the link content corresponding to the embedded link based on the predicted semantic tags, and generate summary semantic tags and structured summary content corresponding to the embedded link; The content integration module is used to respond to a content insertion operation triggered for the structured summary content, and integrate the structured summary content into the embedding position of the embedded link in the target information based on the index relationship between the pre-built predicted semantic tags and the summarized semantic tags.

16. An electronic device, characterized in that, include: A memory and a processor are communicatively connected, the memory storing computer instructions, and the processor executing the computer instructions to perform the method for processing linked data according to any one of claims 1 to 14.

17. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the method for processing linked data according to any one of claims 1 to 14.

18. A computer program product, characterized in that, Includes computer instructions for causing a computer to perform the method for processing linked data according to any one of claims 1 to 14.

Images