Document feedback content processing method and device, equipment and storage medium

Abstract

The invention discloses a document feedback content processing method and device, equipment and a storage medium, and relates to the technical field of document operation software. The method comprises the steps of obtaining feedback content for a target document; generating summary content based on the feedback content; and associating the summarized content to the target document. Through the technical means, the feedback content of the document can be summarized and associated to the target document, a user is assisted in rapidly understanding and processing the feedback content of the document, and the problem that in the prior art, the feedback content processing efficiency is low is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of document software, and in particular to a document feedback content processing method and device, equipment and a storage medium. BACKGROUND

[0002] In the context of the increasing popularity of digital collaborative office work, multi-person collaboration scenarios such as product document review, legal contract review, academic paper writing, and project approval processes have become the core link of work in various industries. In such scenarios, participants provide feedback and suggestions for modification through document annotation functions, which is a key way to achieve information exchange and work collaboration. The quality and efficiency of annotations determine the speed and quality of collaborative tasks.

[0003] In the prior art, participants read the annotation content of the document one by one and reply to the annotation content or revise the corresponding document content. However, as the scale of collaboration expands and the complexity of content increases, document annotations gradually reveal a series of common problems such as large number, scattered location, repeated occurrence of similar problems, and complex semantics, making it difficult for manual processing mode to cope with the problem, affecting the efficiency of annotation processing. SUMMARY

[0004] The present application provides a document feedback content processing method, device, equipment and storage medium to summarize and associate the feedback content of the document to the target document, assisting users in quickly understanding and processing the feedback content of the document, solving the problem of low efficiency of feedback content processing in the prior art.

[0005] In a first aspect, the present application provides a document feedback content processing method, comprising: obtaining feedback content for a target document; generating summary content based on the feedback content; associating the summary content to the target document.

[0006] In a second aspect, the present application provides a document feedback content processing device, comprising: a feedback obtaining module configured to obtain feedback content for a target document; a summary generating module configured to generate summary content based on the feedback content; a summary associating module configured to associate the summary content to the target document.

[0007] In a third aspect, the present application provides a document feedback content processing device, comprising: one or more processors; The memory stores one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method for processing feedback content of a document according to the first aspect.

[0008] In a fourth aspect, the present application provides a storage medium containing computer executable instructions for executing the method for processing feedback content of a document according to the first aspect when executed by a computer processor.

[0009] In the present application, the feedback content for the target document is obtained, the summary content is generated based on the feedback content, and the summary content is associated with the target document. Through the above technical means, the feedback content of the target document distributed discretely can be summarized and a unified summary content is generated, so as to briefly summarize the massive and complex feedback content through the summary content, reduce the artificial processing burden caused by a series of common problems such as large quantity, dispersed position, repeated occurrence of similar problems and complex semantics, and facilitate the user to quickly understand and process the feedback content of the document. By establishing the association relationship between the summary content and the target document, the user can master the core feedback information through the summary content without traversing the target document, the processing mode is changed from checking one by one to grasping the whole, the user's reading and processing of the feedback content is saved, the feedback processing efficiency is improved from the processing mode, and the problem of low feedback processing efficiency in the prior art is solved. BRIEF DESCRIPTION OF DRAWINGS

[0010] Figure 1 is a flowchart of a method for processing feedback content of a document provided by an embodiment of the present application; Figure 2 is a schematic diagram of a session process of instant messaging provided by an embodiment of the present application; Figure 3 is a schematic diagram of a content display interface of a target document provided by an embodiment of the present application; Figure 4 is a schematic diagram of a content display interface of a target document provided by an embodiment of the present application; Figure 5 is a schematic diagram of a content display interface of a target document provided by an embodiment of the present application; Figure 6 is a schematic diagram of a document space interface of a target document provided by an embodiment of the present application; Figure 7 is a schematic diagram of a content display interface of a target document provided by an embodiment of the present application; Figure 8 is a schematic diagram of a content display interface of a target document provided by an embodiment of the present application; Figure 9Fig. 6 is a schematic diagram of a content display interface of a target document according to an embodiment of the present application; Figure 10 Fig. 7 is a flowchart of another method for processing feedback content of a document according to an embodiment of the present application; Figure 11 Fig. 8 is a flowchart of processing feedback content of a target document according to an embodiment of the present application; Figure 12 Fig. 9 is a schematic diagram of a device for processing feedback content of a document according to an embodiment of the present application; Figure 13 Fig. 10 is a schematic diagram of a device for processing feedback content of a document according to an embodiment of the present application. DETAILED DESCRIPTION

[0011] In order to make the objects, technical solutions and advantages of the present application clearer, the specific embodiments of the present application are further described in detail below with reference to the drawings. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application. In addition, it should be noted that only the parts related to the present application are shown in the drawings and not all parts. Before discussing the example embodiments in more detail, it should be mentioned that some example embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when the operations are completed, but can also have additional steps not included in the drawings. The process can correspond to a method, function, procedure, subroutine, etc.

[0012] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of a kind and do not limit the number of objects, for example, the first object can be one or more. In addition, the specification and claims "and / or" indicate at least one of the connected objects, and the character " / ", generally indicates that the objects before and after are in a "or" relationship.

[0013] In an implementation manner, the participants feed back opinions and make modification suggestions through a document annotation function to realize information interaction and work collaboration. Specifically, the participants read the annotation content of the document item by item, and reply to the annotation content or revise the corresponding document content. However, as the collaboration scale expands and the content complexity increases, the document annotation gradually exposes a series of common problems such as large quantity, scattered location, repeated occurrence of similar problems, and complex semantics, and the artificial processing mode item by item is difficult to cope with, affecting the annotation processing efficiency.

[0014] To solve the above problems, the embodiment provides a processing method of document feedback content to summarize the feedback content of the document and associate the feedback content to a target document, thereby assisting a user in quickly understanding and processing the feedback content of the document.

[0015] The processing method of document feedback content provided in the embodiment can be executed by a document feedback content processing device. The document feedback content processing device can be implemented in a software and / or hardware manner, and can be composed of two or more physical entities or one physical entity. For example, the document feedback content processing device can be a terminal device installed with a document application software, such as a computer, a tablet, and a mobile phone. The document application software is an application program or a browser that can open and process a document.

[0016] In the embodiment, the feedback content of the document includes at least one of annotation content, conversation content, and comment content of the document. The annotation content is annotation content of a specific position in the document, which focuses on accurate opinion annotation of specific paragraphs, data, or formats in the document. The conversation content is communication and discussion of participants of the document around the document, and the communication and discussion modes include email, instant messaging, conference, and AI (Artificial Intelligence) conversation. The comment is a macroscopic viewpoint expression on the whole or a part of the document. The annotation content, the conversation content, and the comment content of the document jointly constitute a transmission system of feedback opinions in collaboration. Since the feedback content of the document involves the annotation content, the conversation content, and the comment content, the document application software can also be a document application plug-in embedded in an interactive application software such as a conversation and comment interaction, and the interactive application software realizes editing of the document through the embedded document application plug-in. Of course, similarly, the interactive application software can also be a plug-in of the document application software. The document application software can be a traditional document application software or a smart document application software, and the smart document application software supports processing of the feedback content of the document through a pre-trained intelligent model.

[0017] Furthermore, document application software can also call the service interfaces provided by interactive application software to query document-related conversation content and / or comment content. In addition, multi-user collaborative document application software can also support annotating documents, initiating conversations and discussions on documents, and leaving corresponding discussions on documents. That is, multi-user collaborative document application software can directly query the conversation content and discussion content of documents in the conversation function module and discussion function module.

[0018] The document feedback processing device is equipped with at least one type of operating system, including but not limited to Android, Linux, and Windows. The device may have at least one application installed on the operating system; this application can be a built-in application or an application downloaded from a third-party device or server. In this embodiment, the device has at least one application capable of executing document feedback processing methods. This application can be an application for opening and editing documents or a browser.

[0019] For ease of understanding, this embodiment uses a document application software as the main body for describing the document feedback content processing method.

[0020] Figure 1 A flowchart of a document feedback content processing method provided in an embodiment of this application is given. (Reference) Figure 1 The specific methods for handling the feedback content in this document include: S110. Obtain feedback content for the target document.

[0021] For example, when the document application software is a regular document application software that only supports document feedback through annotations, the annotations in the target document are the only feedback content existing in the target document, and one annotation in the target document can be obtained as feedback content. In addition to obtaining the annotation content as feedback content, the text content referenced by the annotation content can also be obtained as the target document content associated with the annotation content. At least one of the following can be used as the basic information for the corresponding feedback content: the annotation time, the annotation author, and the position of the target document content in the target document.

[0022] When a document application supports commenting on documents via a comment interaction application, the target publication record corresponding to the target document can be determined in the comment interaction application based on the document's name or identifier. The comment content of the target publication record is then used as feedback content for the target document. The comment interaction application can be a plugin for the document application, or the document application can be a plugin for the comment interaction application. In addition to obtaining the comment content as feedback, at least one of the following can be obtained as basic information for the corresponding feedback content: commenter, comment time, comment channel, number of supporting comments, and number of opposing comments.

[0023] Optionally, some comments may be irrelevant to the target document or may not be helpful in modifying the target document. These comments are considered invalid and can be ignored or deleted. The comments that relate to the modification suggestions for the target document should be retained as the feedback content for the target document. This avoids analyzing invalid content and affecting the accuracy and efficiency of the feedback summary.

[0024] Furthermore, to facilitate reading of comments on the target document, semantically related document content can be identified within the document and used as the target document content for the comments. For example, keywords in the target comment can be matched with keywords in the target document, and the paragraph containing those keywords can be used as the target document content. Alternatively, the target comment and the document can be input into a large model for processing to obtain semantically related document content. The input to the large model can be identifiers within the document or the document content itself. Of course, the target comment can also be set as annotations to the corresponding content in the target document. Subsequently, the commenter and comment time of the target comment can be used as the annotator and annotation time of the corresponding annotation.

[0025] When a document application supports document discussions via a conversational application, the target conversation for that document can be located within the conversational application based on the document's name or identifier. The content of this conversation can then be used as feedback for the target document. The conversational application can be a plugin for the document application, or vice versa. For example, when collaborators on a target document communicate their revision suggestions via email, the corresponding email can be identified based on the document's name, and the email's body can be used as feedback for the target document. The target document can be an email attachment, or a link to the target document can be used as the email body. In addition to obtaining the conversation content as feedback, basic information such as the participants, conversation time, conversation type, and conversation keywords can also be obtained as feedback.

[0026] However, a single conversation may involve more than just the target document; it may also raise other related questions that are actually unrelated to the processing of the target document. In this regard, it is possible to accurately obtain relevant information about the target document within the conversation, eliminate conversational interference unrelated to the target document, and improve the accuracy of the feedback.

[0027] In one embodiment, a target session for the target document is determined, wherein the target session is a session initiated based on the target document, or the target session is a session that accesses the target document during the session; target session content related to the target document is obtained from the session content of the target session as feedback content of the target document.

[0028] For example, Figure 2 This is a schematic diagram of the instant messaging session process provided in an embodiment of this application. For example... Figure 2 As shown, personnel A, B, C, and D initiated an instant messaging discussion meeting about Project A's content document. Personnel B, C, and D offered their respective opinions on the document. When personnel A edited the content document of Project A using the chat application, the application used Project A's content document as the target document. Based on the document's terminology, it searched for the Project A discussion meeting and designated it as the target meeting for the target document. Then, within the target meeting's content, it retrieved the chat messages from personnel B, C, and D, all of which involved their suggestions for modifying the target document. This chat content was then used as the target meeting content related to the target document and subsequently as feedback for the target document.

[0029] Alternatively, if multiple participants in an online meeting (A) open and discuss a document related to project A, and then open other documents for further discussion, the session software application will record the names of the documents opened in online meeting A, and convert the audio from online meeting A into text. Then, when the session software application uses a document related to project A as the target document, it will look up the document name to determine if it was accessed during online meeting A, and then designate online meeting A as the target session, using its text as the session content. Finally, based on the meeting time when the target document was accessed in online meeting A, it will quickly retrieve the relevant text from the meeting text to provide feedback for the target document.

[0030] This embodiment breaks through the traditional limitation of only collecting comments within the document, by including the conversation content discussing the target document in the feedback content. This avoids missing important opinions generated during the collaboration process and ensures the accuracy of target document processing. Furthermore, by identifying the target conversation and related target conversation content, interference from conversations unrelated to the target document is eliminated, ensuring that the feedback content only contains valid discussion information about the target document, thus improving the accuracy and relevance of the feedback content.

[0031] Furthermore, besides directly using target session content related to the target document as feedback content, it's also possible to generate annotations for the document based on the target session content, and then use these annotations as feedback content. For example, this involves retrieving target session content related to the target document from the target session's session content; identifying target document content associated with the target session content within the target document; and setting annotations for the target document content generated based on the target session content.

[0032] refer to Figure 2 After determining that the Project A discussion meeting is the target session for the content document of Project A, the chat content from the Project A discussion meeting, containing modification opinions on the content document of Project A from personnel B, C, and D, is used as the target session content. Further, semantically related document content within the document can be identified as the target document content for the target session. This embodiment does not limit the specific target document content matching method. For example, keywords in the target session content can be matched with keywords in the target document, and the paragraph containing the matched keywords can be used as the target document content associated with the target session content. Specifically, for example, if the keyword in personnel B's chat content is "launch date," this keyword will match the launch date in the target document, and thus the paragraph describing the launch date in the target document can be used as the target document content for personnel B's chat content. After determining the target session content and associated target document content, the target session content is created as an annotation for that target document content. Subsequently, the participants and the time of the session can also be used as the annotator and annotation time for the corresponding annotation.

[0033] It should be noted that the target session content related to the target document within the target session content may be discretely or continuously distributed. However, this embodiment matches the target session content corresponding to the document within the target session content based on the semantics of the document content or the document name. Therefore, regardless of whether the target session content is discretely or continuously distributed, the session content discussing the target document will be considered as the target session content. For example, in online meeting A, the content document of project A is discussed first, then the listing precautions of project B are discussed, and finally the content document of project A is discussed again. Therefore, online meeting A will have two discretely distributed session content segments discussing the content document of project A, and these two discretely distributed session content segments will be used as the target session content of project A's content document, respectively. This embodiment matches the target session content corresponding to the document within the target session content based on the semantics of the document content or the document name. Even if the target session content is discretely distributed within the session content, all target session content related to the target document can be compiled, thereby summarizing and processing the discretely distributed target session content and improving the convenience for users to view the session content related to the target document.

[0034] This embodiment accurately locates the target document content associated with the target session content within the target document, and creates annotations for the target session content within the target document content. This establishes a precise association between the session content and the document content, allowing users to directly see the relevant opinions in the session when viewing the target document, without having to switch back and forth between the session record and the target document, thus improving the convenience and efficiency of feedback processing.

[0035] When the document application is a collaborative application that supports document annotation, discussion, and comments, it can retrieve the annotation content, discussion content, and comments of the target document as feedback content. For example, the document application stores the sessions initiated by the document, the sessions accessing the document during discussions, and the document's publication records in the document's relevant record space. When document processing is triggered, it retrieves the target session and target publication record related to the document from the relevant record space, and obtains the target session content from the target session and the document's comments from the target publication record.

[0036] Alternatively, the document application software can support calling the service interfaces of the session application software and / or the discussion interaction application software to query the target session and / or the publication record of the target document. Then, it can retrieve the target session content of the target document from the target session returned by the session application software and / or the target discussion content of the target document from the publication record returned by the discussion interaction application software. For example, the document application software calls the service interface of the session application software to pass the document name of the target document to the session application software. The session application software queries all session records for the target session involving the target document and returns the session content of the target session to the document application software through the service interface. The document application software receives the session content of the target session from the service interface and retrieves the target session content of the target document from the session content. And / or, the document application software calls the service interface of the discussion interaction application software to pass the document name of the target document to the discussion interaction application software. The discussion interaction application software queries all publication records for the target publication record of the target document and returns the target publication record of the target document to the document application software through the service interface. The document application software receives the target publication record of the target document from the service interface and retrieves the target discussion content of the target document from the comment section of the target publication record.

[0037] When feedback content includes any of the following types: annotations, conversation content, and comments, it suffers from problems such as discrete distribution, large quantity, scattered locations, and repetition of similar feedback. When feedback content includes at least two of these types, it may even be distributed across applications. For users, directly processing this feedback content is very time-consuming. This embodiment aims to summarize and process this feedback content into a summary. The summary does not suffer from the aforementioned problems of discrete distribution, large quantity, scattered locations, and repetition of similar feedback, making it easier for users to intuitively understand and process the feedback content, thus improving the convenience of feedback processing.

[0038] S120. Generate summary content based on feedback content.

[0039] The summary can be a summary of all feedback from the target document, or it can be a summary of only some of the feedback from the target document.

[0040] For example, when summarizing all feedback content of the target document, a natural language processing algorithm can be called to preprocess all feedback content to remove redundant expressions, and a keyword extraction algorithm can be used to extract the core opinions of all feedback content, such as "release date postponed" and "interface version error". The core opinions can be integrated and summarized to obtain the summary content: release date postponed and interface version error.

[0041] When summarizing the feedback from the target document, natural language processing algorithms are used to preprocess the feedback to remove redundant expressions. Keyword extraction algorithms are used to extract the core opinions from the feedback, and these core opinions are then integrated and summarized to obtain the summary of the feedback.

[0042] Optionally, when summarizing partial feedback, the feedback content of the target document can be categorized to generate at least one feedback set; corresponding summary content is then generated based on the feedback content in each set. For example, when the feedback content includes annotations, conversation content, and comments, all annotations can be assigned to an annotation feedback set, all conversation content to a conversation feedback set, and all comments to a comment feedback set. Summarizing the annotations in the annotation feedback set yields the corresponding annotation summary content; summarizing the annotations in the conversation feedback set yields the corresponding conversation summary content; and summarizing the comments in the comment feedback set yields the corresponding comment summary content. It is understood that annotators, conversation participants, and commenters of the target document will offer opinions from different perspectives. Categorizing the feedback content of the target document according to its source into annotation feedback sets, conversation feedback sets, and comment feedback sets allows the summary content of each feedback set to intuitively reflect the opinions of annotators, conversation participants, and commenters on the target document, improving the convenience for staff in processing the feedback content.

[0043] This embodiment categorizes feedback content to generate corresponding summary content for each categorized feedback set. This avoids irrelevant feedback from being mixed in, which would affect the relevance and comprehensiveness of the summary content, thus improving the accuracy of the summary. Categorized summaries allow users to quickly locate feedback content related to their areas of interest, improving feedback processing efficiency.

[0044] Furthermore, in addition to grouping feedback content based on its source, feedback content can also be categorized based on preset dimensions to generate multiple feedback sets corresponding to those dimensions. Preset dimensions include at least one of location, author, time, and type. Location refers to the position of the feedback content within the target document. This position can be the content itself, or the location of the content within a paragraph, chapter, or page. Author refers to the author of the feedback content, such as the annotator, participant, or commenter. Time refers to the time the feedback content was generated, such as the annotation time, session time, or comment time. Feedback types include errors, questions, and suggestions, and these types can be identified based on the semantic information of the feedback content.

[0045] For example, when classifying feedback content based on the location of the target document content associated with the feedback content, feedback content of the same paragraph of the target document can be grouped into the same feedback set, or feedback content of the target document content under the same title can be grouped into the same feedback set. Then, the feedback content of each feedback set is summarized to obtain the corresponding set summary content. This embodiment classifies feedback content based on the location of the target document content to group continuously distributed feedback content into the same feedback set, making it easier for users to process feedback in batches according to the distribution area of ​​the document, such as paragraphs and titles, avoiding the tedious operation of processing feedback across locations. At the same time, when the same problem frequently occurs in a certain paragraph or title of the document, the feedback content of these same problems can be effectively summarized into a single summary content, further improving the efficiency of feedback processing.

[0046] When categorizing feedback content based on the author, feedback from the same author can be grouped into the same feedback set. Then, the feedback content from each set is summarized to obtain a summary of the corresponding feedback set. It's understood that different collaborators on the target document have different tasks and therefore provide different feedback. This embodiment categorizes feedback content based on the author to summarize the core feedback from each author's feedback set. This allows users to specifically understand the core needs of each collaborator through the summary of each feedback set, improving the efficiency of processing feedback. Furthermore, after dividing feedback sets based on the author, the proportion of feedback content in each set relative to the total amount of feedback content can be calculated. This proportion is then correlated with the summary of the set to indicate the importance of each set's summary to the user.

[0047] When categorizing feedback based on its time, feedback from the same period (e.g., one day, half a day, one hour, etc.) can be grouped into the same feedback set. Then, the feedback from each set is summarized to obtain a summary of the corresponding feedback set. It can be understood that feedback from different periods reflects the focus of feedback at different stages. This embodiment categorizes feedback based on its time to summarize the focus of feedback for each stage through the corresponding feedback set, allowing users to understand the focus of feedback at each stage through the summary, thus improving the efficiency of processing feedback. Furthermore, after grouping feedback sets based on their time, the priority of each set can be determined according to its time; the closer the time is to the current time, the higher the priority. The priority of each feedback set is then linked to the summary content to prompt users to prioritize the most recent feedback, ensuring the timeliness of feedback processing.

[0048] When classifying feedback content based on its type, a preset content classification model can be used to identify and categorize the feedback content into errors, questions, and suggestions. Feedback of type error is grouped into an error feedback set, feedback of type question into a question feedback set, and feedback of type suggestion into a suggestion feedback set. Then, the feedback content in the error feedback set is summarized to obtain error summary content, the feedback content in the question feedback set is summarized to obtain question summary content, and the feedback content in the suggestion feedback set is summarized to obtain suggestion summary content. It can be understood that the type of feedback content reflects the processing method. In this embodiment, classifying feedback sets based on the type of feedback content effectively summarizes the processing direction of the feedback set: correcting errors, answering questions, or accepting and rejecting suggestions. This allows users to quickly understand the meaning of the feedback set and clarify the feedback processing method based on the summary content, improving the efficiency of users in processing feedback content. Furthermore, after dividing the feedback sets based on the type of feedback content, the proportion of feedback content in each feedback set to the total number of feedback contents can be calculated. The proportion of feedback in each feedback set can be correlated with the summary content of the set, so as to prompt the user with the core processing method of the target document through the feedback proportion.

[0049] Document application software can categorize feedback content according to preset dimensions, and can also categorize feedback content of a target document according to the target categorization dimension selected by the user within the preset dimensions. For example, Figure 3 This is one of the schematic diagrams of the content display interface of the target document provided in the embodiments of this application. For example... Figure 3 As shown, the content display interface displays a toolbar 11 and a document area 10. The document content of the target document is displayed in the document area 10. The toolbar 11 has a feedback category control 12. When the user clicks the feedback category control 12, a category dimension window 13 is opened, where the target category dimension can be selected. When the user selects chapter and type, the chapter and type are used as the target category dimensions, thus dividing the feedback content according to chapter and type.

[0050] In another embodiment, the document application software can also classify the feedback content into multiple levels according to various preset dimensions. For example, it can first classify by paragraph, and then further classify by author within the same paragraph category.

[0051] This embodiment explicitly defines the classification dimensions of feedback content as location, author, time and / or type to meet diverse classification needs. Users can directly select preset dimensions for classification without customizing classification rules, simplifying the classification process of feedback content and improving classification efficiency.

[0052] Optionally, when generating the summary content, in addition to extracting the core opinions of the feedback content based on keyword extraction algorithms and integrating them into the summary content, the summary content can also be generated based on the semantic information of the feedback content to further improve the accuracy of the summary content. Specifically, semantic analysis is performed on at least two feedback content sets to generate the summary content, which includes at least one of the following: a summary of the feedback content and statistical values ​​of the feedback content.

[0053] For example, prompts for the large language model are constructed based on all feedback content of the target document. For instance, a prompt could be "Summarize the above feedback content to outline similar feedback content, count the number of similar feedback contents, and output a summary of the above feedback content." The constructed prompts are then input into a pre-defined large language model. Guided by the prompts, the model performs semantic analysis on all feedback content of the target document, summarizes similar feedback content, counts the number of similar feedback contents, and finally outputs a summary of all feedback content. For example, the final output of the large language model might be: "This document contains 8 annotations, of which 5 indicate unclear logic, 2 suggest supplementing citations, and 1 is a risk warning."

[0054] When all feedback content of the target document is categorized into multiple feedback sets, prompt words for a large language model can be constructed from all the feedback content of each set. These prompt words are then input into a pre-defined large language model, which outputs a summary of the feedback sets under the guidance of these prompt words. For example, when feedback sets are categorized based on authors, feedback set A includes feedback content from author A, and feedback set B also includes feedback content from author A. Inputting the feedback content from feedback set A into the large language model yields a summary of "Mainly concerned about the legality of payment terms (6 items in total)", while inputting the feedback content from feedback set B yields a summary of "Proposing unified format standards (4 items in total)".

[0055] Furthermore, to clarify which feedback content the summary points to, a requirement can be added to the prompt words to specify the feedback content associated with the output summary. This ensures that the summary output by the large language model includes not only the summary and statistical values, but also the corresponding feedback content associated with the summary.

[0056] Besides directly generating summaries of feedback content using large language models, semantic analysis can be performed on multiple feedback items to obtain semantic vectors. These semantic vectors can then be quantified, statistically analyzed, and integrated with similar semantics to generate a summary containing both abstracts and statistical values. For example, semantic analysis models such as BERT, Transformer encoder, or Sentence-BERT can be used to extract semantic vectors from the feedback content. These vectors are then input into a summarization model, which integrates similar semantic vectors to obtain corresponding abstracts, counts the number of similar semantic vectors used for integration, and outputs the summary content. The summarization model can be a statistical summarization model or an NLP summarization model. It should be noted that each abstract in the summary output by the summarization model is also associated with its corresponding feedback content, allowing for the location and processing of the relevant feedback content through the abstract. For example, when the feedback set is categorized based on document chapters, the feedback set includes feedback content from Chapter 3. Semantic analysis is performed on the feedback content from Chapter 3 to obtain semantic vectors. The semantic vectors of the feedback content from Chapter 3 are then quantified and statistically analyzed, and similar semantics are integrated to generate a summary of "Flowchart location error (3 items), suggestions to add examples (5 items), unclear terminology explanation (2 items)".

[0057] This embodiment utilizes semantic analysis algorithms to deeply mine the core semantics of feedback content, avoiding omissions of key information or biases in the summary. It supports summaries including both abstracts and statistical values, allowing for a more intuitive presentation of core opinions through abstracts and quantitative information through statistical values, thus enriching the presentation of the summary. For semantically complex long text feedback content, semantic analysis can effectively extract the core content and generate concise and clear summaries, facilitating intuitive user understanding and improving the processing efficiency of feedback content.

[0058] The summary content mentioned above refers to summarizing the core opinions of the feedback in a textual form. However, besides presenting the core opinions in textual form, the summary content can also be presented in list form. For example, feedback can be categorized into multiple feedback sets, and the feedback content from each set can be summarized to obtain a set summary. Finally, a summary list can be generated from the set summaries of all feedback sets.

[0059] Optionally, in addition to classifying feedback content based on preset dimensions and content sources, semantic similarity analysis can be performed on the feedback content to obtain grouped feedback content and generate summary content for each group; based on the summary content of each group, a summary list is formed. For example, the semantic vector of each feedback content is extracted through a semantic analysis model, the semantic similarity between any two feedback contents is calculated based on the semantic vectors of each feedback content, and the semantic similarity between the feedback contents is used to group semantically similar feedback contents into the same group. The semantic vectors of the feedback contents in the same group are summarized and the number of feedbacks is counted to obtain the summary content of that group, and the summary content of each group is combined into a summary list. For example, if three pieces of feedback in the target document involve incorrect flowchart placement, five pieces of feedback suggest adding examples, and two pieces of feedback involve unclear terminology, then the three pieces of feedback regarding incorrect flowchart placement are assigned to group A, the five pieces of feedback suggesting adding examples are assigned to group B, and the two pieces of feedback regarding unclear terminology are assigned to group C. The semantics of each group are then summarized, and the number of feedback points is counted. The summary for group A is "3 incorrect flowchart placements," the summary for group B is "5 suggestions to add examples," and the summary for group C is "2 unclear terminology points." The summaries for groups A, B, and C are then combined into a summary category.

[0060] This embodiment utilizes the semantic similarity of feedback content to group feedback content that differs in expression but shares the same semantic information into the same group, and generates summary content for each group. This facilitates users' quick understanding of similar feedback content and improves the processing efficiency of similar feedback. The summary content of each group is compiled into a summary list to present the summary content in a structured manner, allowing users to quickly browse the summary content of each group and improving the ease of processing feedback. Furthermore, after grouping similar feedback, the statistical frequency in the summary content of each group can intuitively show the frequency of the corresponding feedback issue, allowing users to prioritize high-frequency feedback and improve the efficiency of collaborative tasks.

[0061] Furthermore, when the document is long, the feedback content of each chapter can be categorized to obtain feedback sets for each chapter. Semantic similarity analysis can then be performed on the feedback content in each chapter's feedback set to further divide the feedback content into groups. Based on the summary content corresponding to the group of feedback content, a summary list for the feedback set of that chapter can be generated. Thus, the summary list of each chapter can intuitively present the summary and statistical values ​​of various types of feedback content in that chapter.

[0062] S130. Link the summary content to the target document.

[0063] For example, after generating a summary of all or part of the feedback content of the target document, the summary can be displayed in the target document's content display interface or in the target document's document space interface. Regardless of whether the summary is text or a list, it can be displayed in either the target document's content display interface or document space interface.

[0064] For example, in the content display interface, the document content of the target document is expanded and displayed in the document area, and the summary content can be displayed in the summary area of ​​the content display interface. The summary area can be displayed in the content display interface as annotations, a sidebar, or a floating window, and the summary area can be hidden and expanded. For example, Figure 4 This is a second schematic diagram of the content display interface of the target document provided in the embodiments of this application. For example... Figure 4 As shown, the summary area is displayed in the content display interface as a floating window 14. The summary text of the target document can be displayed in the floating window 14, and the floating window 14 will also slide when the user turns the page. Figure 5 This is the third schematic diagram of the content display interface of the target document provided in the embodiments of this application. For example... Figure 5 As shown, the summary area is displayed as a sidebar 15 on one side of the document area 10. The summary list 16 of the target document can be displayed in the sidebar 15, and the summary list 16 will also slide when the user turns the page.

[0065] In the document space interface, the target document is displayed in thumbnail form. Summary content can be displayed in a summary area associated with the target document's thumbnail. This summary area is displayed in the document space interface as a floating window or notes. For example, Figure 6 This is a schematic diagram of the document space interface of the target document provided in the embodiments of this application. For example... Figure 6 As shown, the summary area 22 associated with the thumbnail 21 of the target document is displayed as a floating window in the document space interface 20, and the summary content of the target document can be displayed in the summary area 22.

[0066] It should be noted that both the summary area of ​​the content display interface and the summary area of ​​the document space interface can display all summary content of the target document. That is, when the feedback content of the target document is divided into multiple feedback sets, the summary content corresponding to all feedback sets is displayed in the summary area; or when the feedback content of the target document is divided into multiple groups, the summary content corresponding to all groups is displayed in the summary area. When the summary content of feedback sets is displayed in the summary area, the corresponding category dimension of the feedback set can also be added. For example, if feedback set A is the feedback content of author A, the summary area will display "Author A's feedback summary is: mainly concerned about the legality of payment terms (6 items in total)". The summary area can also support the simultaneous display of summary content of feedback sets corresponding to various category dimensions, such as simultaneously displaying the summary content of feedback sets divided by author and the summary content of feedback sets divided by chapter.

[0067] Furthermore, when the feedback content of the target document is divided into feedback sets based on chapters or paragraphs, the summary content of each feedback set can be associated with the corresponding chapter or paragraph in the target document. This allows users to quickly understand the corresponding feedback content through the summary content associated with the chapter or paragraph, improving the convenience and efficiency of feedback processing. For example, the summary content can be displayed in the summary area of ​​the corresponding chapter or paragraph in the target document, and the summary area can be displayed in the content display interface in the form of annotations, sidebars, or floating windows.

[0068] After associating the summary content with the target document, the summary content and the source entry can be displayed. The source entry points to the feedback content corresponding to the summary content; user actions applied to the source entry are received; and in response to user actions, the feedback content pointed to by the source entry is displayed.

[0069] For example, each summary corresponds to a source tracing entry, displayed as a clickable "View Details" link. This entry is associated with the identifiers of all feedback content corresponding to the summary. When a user clicks the source tracing entry, they can access the identifiers of all feedback content corresponding to that entry. Based on these identifiers, the system retrieves the relevant feedback content from the database and displays all feedback content related to the summary in a pop-up window.

[0070] Alternatively, a summary of the content can correspond to a source tracing entry, displayed as a clickable "View Details" link. This entry is associated with the identifiers of all feedback content corresponding to the summary. For example, Figure 7 This is the fourth schematic diagram of the content display interface of the target document provided in the embodiments of this application. Figure 7As shown, the summary content and the source tracing entry are displayed in the floating window 14. The summary content includes two summaries. The first summary is a summary of three feedback contents that are missing from the login process description based on semantic information. The source tracing entry (view details) corresponding to this summary is associated with the identifiers of these three feedback contents. When the user clicks the source tracing entry, the identifiers of these three feedback contents can be obtained, and the corresponding feedback contents can be queried in the database based on the identifiers of these three feedback contents. The queried feedback contents are then displayed below the first summary in the floating window 14.

[0071] Furthermore, when the feedback content is annotated in the form of comments within the target document, the source tracing entry can point to the feedback content that exists as comments within the target document. When a user clicks the source tracing entry, they are redirected to the target document content display page, simultaneously showing both the target document content and the corresponding feedback content. For example, the source tracing entry for the summary content or its abstract is displayed as a clickable link to "View Original Annotations." The number of source tracing entries displayed corresponds to the number of feedback content items corresponding to the summary content or its abstract, with each source tracing entry associated with a specific section of the target document content associated with a summary or feedback content. For example, Figure 8 This is the fifth schematic diagram of the content display interface of the target document provided in the embodiments of this application. Figure 8 As shown, the summary content and three source tracing entry points are displayed in the floating window 14. The summary content is derived from the semantic information of three feedback items missing from the login process description. These three feedback items correspond to three source tracing entry points (view original annotations). The order of the three source tracing entry points can be consistent with the annotation order of the corresponding feedback content in the document. When the user clicks the first source tracing entry point, the content display interface jumps to the annotation of the feedback content pointed to by the first entry point, and displays the document page of the target document content associated with that feedback content in the document area. This allows the user to modify the corresponding target document content based on the feedback content or reply to the feedback content based on the corresponding target document content, improving feedback processing efficiency.

[0072] Furthermore, the floating window 14 that displays the summary content will slide along with the document. When the document area jumps to the document page of the target document content associated with the first feedback content, the floating window 14 will also be displayed. After the user has finished processing the first feedback content, they can click on the source entry of the second feedback content in the floating window 14 to jump to the document page of the target document content associated with the second feedback content, which improves the convenience of feedback processing.

[0073] In addition, the floating window 14 can display only the source entry corresponding to the summary content or the source entry corresponding to the abstract. This source entry points to the annotation of the first feedback content corresponding to the summary content or the annotation of the first feedback content corresponding to the abstract. When the source entry is clicked, it jumps to the document page of the annotation of the first feedback content and the associated target document content. The annotation of the feedback content is equipped with a jump control. When the jump control is clicked, it jumps to the document page of the annotation of the second feedback content and the associated target document content, realizing the jump processing of feedback content and improving the convenience of feedback processing.

[0074] This embodiment displays a summary and a source tracing entry point, allowing users to quickly view feedback content through the source tracing entry point, avoiding information inaccuracies caused by overly simplified summaries. Furthermore, the source tracing entry point allows users to reverse-check the authenticity and completeness of the summary content, enhancing user trust in it. The source tracing entry point locates the feedback content and / or target document content, eliminating the need for users to manually search for the problem text within the target document, thus improving the convenience and efficiency of feedback processing.

[0075] When generating a summary list, the document application software can display the summary content of each group and the source entry point for the feedback content in the corresponding group; it can receive user actions applied to the source entry point; and in response to user actions, it can display the feedback content pointed to by the source entry point. For example, Figure 9 This is the sixth schematic diagram of the content display interface of the target document provided in the embodiments of this application. Figure 9 As shown, the summary content of each group and the source entry for the feedback content within that group are displayed in the same row in summary list 16. The source entry is presented as a clickable "View Details" link, and each source entry is associated with all feedback content of the group in a one-to-many manner. When a user clicks "View Details," all feedback content of the corresponding group will be displayed in the summary list.

[0076] In addition, the source tracing entry also features a clickable link to "View Original Annotations," linking the source tracing entry one-to-one with the annotations of the feedback content in that group. When a user clicks "View Original Annotations," the content display interface jumps to the annotations of the corresponding feedback content, and the document page of the target document content associated with that feedback content is displayed in the document area.

[0077] It is understood that the source tracing entry point of the summary list has the same function and implementation method as the source tracing entry point displayed corresponding to the summary content above. You can refer to the source tracing entry point displayed corresponding to the summary content above to determine the specific function and implementation method of the source tracing entry point of the summary list. No detailed limitation is made here.

[0078] In this embodiment, a corresponding tracing entry point is configured for each group. Users can directly locate the feedback content corresponding to the group through the tracing entry point, thus eliminating the confusion of cross-group tracing.

[0079] Furthermore, when the feedback content is conversation content or comment content, it originates from the target conversation or target posting record. After viewing the feedback content through the source tracing entry, if the feedback content belongs to conversation content or comment content, the feedback content and the data source entry can be displayed together. The data source entry points to the target conversation to which the conversation content belongs or the target posting record to which the comment content belongs. When the data source entry triggers a user action, the target conversation or target posting record pointed to by the data source entry is displayed in response to the user action. For example, when displaying feedback content through a pop-up window, a summary area, or annotations, the data source entry can be displayed together (if the feedback content is conversation content or comment content). The data source entry for conversation content is displayed as a clickable link to "View Conversation Details," and the data source entry for comment content is displayed as a clickable link to "View Posting Details." The clickable link to "View Conversation Details" is associated with the conversation identifier of the target conversation; when the user clicks this link, a display window for the corresponding target conversation pops up. The clickable link to "View Posting Details" is associated with the record identifier of the target posting record; when the user clicks this link, a display window for the corresponding target posting record pops up.

[0080] Because the target document is a collaborative document, when one collaborator is modifying it, other collaborators are adding new feedback in real time. To ensure the timeliness of this feedback, the document application software's server synchronizes the new feedback to the target document being processed by the application software in real time, allowing users to address the new feedback promptly. The document application software can monitor the target document for new feedback in real time. If new feedback is detected, the summary content can be updated accordingly to include the feedback received.

[0081] Furthermore, updating the summary of all feedback content in the target document takes time, and the resulting summary may be too concise, omitting the core information of the new feedback, leading to users not processing it promptly and affecting its timeliness. To address this, all feedback content in the target document can be categorized to determine the feedback set for each type of feedback. Upon detecting new feedback in the target document, the corresponding feedback set is determined; then, a new summary is generated based on the feedback content of each set. For example, if the current feedback set is based on the author, the new feedback can be assigned to the appropriate set based on its author. If this is the first time the author has provided feedback, the new feedback is assigned to a new set. Then, a new summary is generated based on the feedback set to which the new feedback belongs, and the summary associated with the target document is updated. When a user sees an updated summary, they can process the new feedback based on that summary, ensuring timely processing of new feedback. This embodiment categorizes newly added feedback content into corresponding feedback sets and updates the summary content of each set, avoiding the need to update the summary content of all feedback content across the entire document. This improves the efficiency of updating summary content and ensures the timeliness of processing new feedback content. Furthermore, since the summary content covers a smaller number of feedback items, its abstract can highlight the core opinions of the new feedback, preventing these core opinions from being overlooked.

[0082] In summary, the document feedback processing method provided in this application involves obtaining feedback content for a target document; generating summary content based on the feedback content; and associating the summary content with the target document. Through these technical means, feedback content from discretely distributed target documents can be aggregated and a unified summary content can be generated. This summary content concisely summarizes massive and complex feedback content, reducing the manual processing burden caused by common problems such as large quantity, scattered locations, recurring similar issues, and semantic complexity. It facilitates users' quick understanding and processing of document feedback content. By establishing the association between the summary content and the target document, users can grasp the core feedback information through the summary content without traversing the target document. This achieves a shift in processing mode from item-by-item review to overall understanding, eliminating the need for users to read and process feedback content item by item. This improves feedback processing efficiency and solves the problem of low feedback processing efficiency in existing technologies.

[0083] In another embodiment, Figure 10 This is a flowchart illustrating another method for processing document feedback content provided in an embodiment of this application. For example... Figure 10 As shown, the steps for processing the feedback content of this document include: S210. Obtain feedback content for the target document.

[0084] Step S210 can be referred to step S110.

[0085] S220. Classify the feedback content of the target document to generate at least one feedback set.

[0086] For example, feedback content can be categorized based on its source to generate sets of annotation feedback, conversation feedback, and comment feedback. Alternatively, feedback content of the target document can be categorized based on preset dimensions to generate multiple feedback sets corresponding to those dimensions; preset dimensions include at least one of location, author, time, and type. Furthermore, semantic similarity analysis can be performed on the feedback content to group semantically similar feedback content into the same feedback set.

[0087] S230. Generate corresponding set feedback content based on the feedback set.

[0088] For example, the aggregated feedback content can be a combination of all feedback content in the feedback set. That is, the aggregated content directly presents the full text of the corresponding feedback content, rather than a summary of the feedback content. The summary of all feedback content in the feedback set can be in list form or text form.

[0089] In addition, the set of feedback content can also be a summary of all feedback content in the feedback set. The summary content is obtained by summarizing all feedback content in the feedback set and is a summary text of the feedback content. Step S120 describes the generation method of the summary content, which can be referred to in step S120. It will not be repeated here.

[0090] S240. Associate the collected feedback content with the target document.

[0091] For example, the collected feedback content can be displayed in the content display interface of the target document, or the summary content can be displayed in the document space interface of the target document. Regardless of whether the collected feedback content is a summary or a combination of content, it can be displayed in either the content display interface or the document space interface of the target document.

[0092] If the feedback content is a summary, you can refer to the specific implementation process of associating the summary content with the target document in step S110, which will not be repeated here.

[0093] When the feedback content is a combined set, the batch injection portals for each piece of feedback in the set can be associated with the corresponding feedback content in the combined set. The batch injection portals point to the annotations of the feedback content in the target document. The system receives user actions primarily for the batch injection portals and responds by displaying the annotations corresponding to the feedback content in the target document, along with the target document content itself. For example, the batch injection portal for a feedback content might be displayed as a clickable link to "View Original Annotations." The batch injection portal is associated with the location of the target document content linked to the feedback content. When the user clicks the batch injection portal, the content display interface jumps to the annotations of the feedback content pointed to by the batch injection portal, and the document page of the target document content associated with that feedback content is displayed in the document area.

[0094] Furthermore, when the feedback content in the combined content belongs to conversation content or comment content, the data source entry of each piece of feedback content in the feedback set can be associated with the corresponding feedback content in the combined content for display. The data source entry points to the target conversation to which the conversation content belongs or the target posting record to which the comment content belongs. When the data source entry triggers a user action, the target conversation or target posting record pointed to by the data source entry is displayed in response to the user action. For example, the data source entry for conversation content is displayed as a clickable link "View Conversation Details," and the data source entry for comment content is displayed as a clickable link "View Posting Details." The clickable link "View Conversation Details" is associated with the conversation identifier of the target conversation; when the user clicks this link, a display window for the corresponding target conversation pops up. The clickable link "View Posting Details" is associated with the record identifier of the target posting record; when the user clicks this link, a display window for the corresponding target posting record pops up.

[0095] In summary, the document feedback content processing method provided in this application involves: acquiring feedback content for a target document; classifying the feedback content of the target document to generate at least one feedback set; generating corresponding set feedback content based on the feedback set; and associating the set feedback content with the target document. Through these technical means, the discrete and disorganized feedback content of the target document can be classified according to certain rules to generate set feedback content, making the feedback content more organized and facilitating targeted processing by users. Users can process various types of feedback content sequentially according to their categories, avoiding cross-interference between different types of feedback content. By establishing the association between the set feedback content and the target document, users can grasp feedback information through the set feedback content without traversing the target document, thereby improving feedback processing efficiency and solving the problem of low feedback content processing efficiency in the prior art.

[0096] Building upon the above embodiments, whether classifying or summarizing feedback content, document application software can intelligently process the feedback content based on the associated target document content, thereby improving processing efficiency. For example,Figure 11 This is a flowchart illustrating the processing of feedback content based on the target document content, provided in an embodiment of this application. For example... Figure 11 As shown, the steps for processing the feedback content based on the target document content specifically include S310-S330: S310. Obtain the target document content that is associated with the feedback content in the target document.

[0097] For example, when the feedback content has corresponding annotations in the target document, the text referenced by the annotations in the feedback content is the content of the target document associated with it. When the feedback content does not have corresponding annotations in the target document, keywords in the target document can be matched based on the keywords in the feedback content, and the paragraphs, sentences, or containers containing the matched keywords can be used as the target document content associated with the feedback content. Containers include blocks containing elements such as tables, lists, images, videos, columns, and highlighted blocks. Alternatively, semantically related paragraphs, sentences, containers, or images in the target document can be matched based on the semantic information of the feedback content, and the matched paragraphs, sentences, containers, or images can be used as the target document content associated with the feedback content.

[0098] S320. Based on the target document content and the feedback content, generate revised content for the target document content or a response to the feedback content.

[0099] The revised content refers to the content of the target document modified based on the feedback. The response content is used to reply to the feedback. For example, there are two ways to process feedback: one is to revise the target document content associated with the feedback, and the other is to reply to the feedback. The appropriate processing method can be selected according to the type of feedback. For example, when the feedback points out an error or provides suggestions in the target document content, the target document content is revised based on the feedback; when the feedback indicates a question about the target document content, a response to the feedback is generated based on the target document content.

[0100] Optionally, prompts can be constructed based on the target document content and feedback content. These prompts are then input into a pre-defined large language model, which determines the appropriate processing method for the feedback content and generates the corresponding processing result. For example, the prompt could be: "Text 1 is the opinion of Material 2. Please understand the content of Text 1 and Material 2 and determine whether the opinion is reasonable. If reasonable, modify Material 2 or reply to Text 1, and output the modified Material 2 or the reply to Text 1. If unreasonable, provide the reason why it is unreasonable." Here, Text 1 in the prompt refers to the feedback content, and Material 2 refers to the target document content. Based on the prompts, the large language model first determines whether the logic of the feedback content is correct. If the logic is correct, it determines whether the current processing method is to modify the target document content or reply to the feedback content. If the processing method is to modify the target document content, the large language model generates and outputs the revised target document content. If the processing method is to reply to the feedback content, the large language model generates and outputs the reply to the feedback content.

[0101] Furthermore, contextual information about the target document content can be added to the prompt words to help the large language model accurately understand the semantics of the target document content.

[0102] The aforementioned feedback processing method can be triggered without prior operation, offering flexibility but potentially fluctuating accuracy. To address this, this embodiment proposes predicting future user actions on related feedback based on the user's actions on that feedback. This allows for the execution of corresponding processing logic on the relevant feedback, ensuring the accuracy of intelligent feedback processing and reducing repetitive processing of related feedback by the user.

[0103] For example, after a modification operation is triggered on the target document content associated with the feedback content, the relevant feedback content is determined from the remaining feedback content based on the initial feedback. Based on the content before and after the modification operation, revised content of the target document content associated with the relevant feedback content is generated. The modification operation can be a manual modification of the target document content by the user, or an operation where the user accepts an intelligently predicted revised content. Both operations reflect the user's intention to modify the target document content of the feedback content. The content before modification is the target document content associated with the feedback content, and the content after modification is the revised content of the target document content.

[0104] After responding to a modification operation on the target document content associated with the feedback content, semantically similar feedback content can be identified as relevant feedback content based on the semantic information of the feedback content, and / or, based on the structural attributes of the target document content associated with the feedback content, feedback content with the same structural attributes can be identified as relevant feedback content. Then, the differences between the content before and after the modification operation are compared, and the target document content associated with the relevant feedback content is adjusted based on these differences to obtain the revised content. For example, if the difference is a format change, the format of the target document content associated with the relevant feedback content is adjusted to match the format of the modified content, and the format-adjusted target document content is used as the revised content. If the difference is a change in a word, the word in the target document content associated with the relevant feedback content is adjusted to match the word in the modified content, and the word-adjusted target document content is used as the revised content.

[0105] Optionally, prompt words can be constructed from the feedback content, the original content, the revised content, other feedback content, and the associated target document content. These prompt words guide the large language model to identify relevant feedback content from the remaining feedback content based on the original and revised content, and then modify the associated target document content based on the original and revised content to output the revised target document content. Alternatively, prompt words can be constructed from the original and revised content, and the associated target document content, guiding the large language model to modify the associated target document content based on the original and revised content to output the revised target document content.

[0106] This embodiment, based on triggering modification operations on the target document content associated with the feedback content, predicts that the user has a unified intention to modify the target document content associated with the relevant feedback content based on the modification operations, and thus modifies the target document content associated with the relevant feedback content, reducing repeated modification operations on the target document content associated with the relevant feedback content by the user, and also ensuring the accuracy of feedback processing.

[0107] After the target document triggers the first response to the feedback content, relevant feedback content is determined from the remaining feedback content based on that initial response. Then, based on the response content of the first response, the corresponding response content is generated. The first response can be either a manual response from the user or an action where the user accepts intelligently predicted feedback content; both actions reflect the user's intention to respond to the feedback.

[0108] After the first response to the feedback content, based on the semantic information of the feedback content, semantically similar feedback content among the remaining feedback content can be used as relevant feedback content, and / or, based on the structural attributes of the target document content associated with the feedback content, feedback content of the target document content with the same structural attributes among the remaining feedback content can be used as relevant feedback content. Then, the response content of the first response operation is directly used as the response content for the relevant feedback content.

[0109] Optionally, prompt words can be constructed from the feedback content, response content, other feedback content, and associated target document content. These prompt words guide the large language model to determine relevant feedback content from the remaining feedback content based on the feedback content and associated target document content. The response content is then adaptively modified based on the feedback content and relevant feedback content to output a response containing the relevant feedback content. Alternatively, prompt words can be constructed from the feedback content, response content, and related feedback content. These prompt words guide the large language model to adaptively modify the response content based on the feedback content and relevant feedback content to output a response containing the relevant feedback content.

[0110] This embodiment, based on the first response operation triggered on the feedback content, predicts that the user has a unified intention to respond to the relevant feedback content based on the first response operation, thereby generating unified response content for the relevant feedback content, reducing the user's repeated response operations on the relevant feedback content, and also ensuring the accuracy of feedback processing.

[0111] Furthermore, when generating summary content based on the feedback from the target document, a response can be made to the summary content. Correspondingly, after the target document triggers a second response to the summary content, a response to the feedback content corresponding to the summary content is generated based on the response content of the second response. The second response can be a manual response from the user to the summary content, or an action where the user accepts a response to intelligently predicted summary content; both actions reflect the user's intention to respond to the summary content. The summary content is a summary of the feedback content, and the user's intention to respond to the summary content also represents their intention to respond to the feedback content. Therefore, this embodiment generates the corresponding response to the feedback content based on the response to the summary content. For example, when the summary content includes only one abstract, it indicates that the feedback content corresponding to the summary content is semantically similar and related. The response content to the summary content is equivalent to the response content to each individual feedback content or to a portion of the feedback content within the summary content. Therefore, the response content to the summary content can be used as the response content to each individual feedback content corresponding to the summary content, or as the response content to a portion of the related feedback content within the summary content. When the summary content includes multiple abstracts, it indicates that the response content to the summary content is a combination of responses to each abstract. Therefore, the response content to each abstract can be extracted from the response content to the summary content. Since the feedback content corresponding to each abstract is semantically similar and related, the response content to the abstract is equivalent to the response content to each individual feedback content within it. Therefore, the response content to each abstract can be used as the response content to each individual feedback content corresponding to the abstract.

[0112] When extracting the response content for each summary from the responses to the summary content, either sequential matching or keyword matching can be used. For example, when using sequential matching, the first response sentence in the summary content's response content is taken as the response content for the first summary, and so on, to determine the response content for each summary. Alternatively, the keywords of each response sentence in the summary content's response content can be matched with the keywords of each summary content's keywords; if the keywords of the response sentence match the keywords of the summary, the response sentence is taken as the response content for the summary.

[0113] Optionally, prompt words can be constructed from the summary content, reply content, and each feedback content. The prompt word prompts the large language model to generate and output the reply content for each feedback content based on the summary content and related reply content.

[0114] This embodiment automatically generates a response for each piece of feedback after the summary is replied to, ensuring that a corresponding response is generated for each piece of feedback. This avoids omissions in the processing of feedback and also avoids requiring users to reply to each piece of feedback individually, thus improving the efficiency and accuracy of feedback processing.

[0115] S330. Associate the revised content or response content with the target document.

[0116] For example, the revised content of the target document can be displayed in the target document as a revision. When the user accepts the revision, the original content of the target document can be replaced with the revised content. Alternatively, the target document content and the revision can be compiled into a summary table, which can then be displayed in the document space interface of the target document.

[0117] When feedback content has annotations in the target document, the corresponding response will be displayed as an annotation in the annotation box. When feedback content does not have annotations in the target document, the response will be displayed in association with the corresponding summary content or collection of feedback content. Alternatively, the feedback content and responses can be compiled into a response summary table, which will then be displayed in the document space interface of the target document.

[0118] This embodiment automatically modifies the target document content or automatically replies to the feedback content based on the feedback content and the associated target document content, providing users with specific reference solutions for processing feedback content. This avoids processing errors caused by users' misunderstanding of the feedback content. Moreover, users do not need to start from scratch to conceive the processing logic of the feedback content. They can make fine adjustments based on the revised text or reply text, which greatly reduces the processing time and effort required for the feedback content and significantly improves the processing efficiency of the feedback content.

[0119] Based on the above embodiments, Figure 12 This is a schematic diagram of a document feedback content processing device provided in an embodiment of this application. (Reference) Figure 12 The document feedback processing device provided in this embodiment specifically includes: a feedback acquisition module 31, a summary generation module 32, and a summary association module 33.

[0120] Among them, the feedback acquisition module 31 is configured to acquire feedback content for the target document; The summary generation module 32 is configured to generate summary content based on feedback content; The summary association module 33 is configured to associate summary content with the target document.

[0121] Based on the above embodiments, the feedback content includes session content related to the target document. The feedback acquisition module 31 includes: a target session determination submodule, configured to determine the target session of the target document before acquiring the feedback content for the target document, wherein the target session is a session initiated based on the target document, or the target session is a session that accesses the target document during the session; and a feedback acquisition submodule, configured to acquire target session content related to the target document from the session content of the target session, as the feedback content of the target document.

[0122] Based on the above embodiments, the feedback acquisition submodule includes: a target session content acquisition unit, configured to acquire target session content related to the target document from the session content of the target session; a target document content determination unit, configured to determine the target document content associated with the target session content in the target document; and an annotation generation unit, configured to set annotations generated based on the target session content for the target document content.

[0123] Based on the above embodiments, the summary association module 33 includes: a first source tracing entry display submodule, configured to display the summary content and the source tracing entry after associating the summary content with the target document, wherein the source tracing entry points to the feedback content corresponding to the summary content; a first source tracing entry access submodule, configured to receive user operations applied to the source tracing entry; and a first source tracing operation response submodule, configured to respond to user operations and display the feedback content pointed to by the source tracing entry.

[0124] Based on the above embodiments, the document feedback content processing device further includes: a feedback classification module, configured to classify the feedback content of the target document and generate at least one feedback set after obtaining the feedback content for the target document; a set feedback generation module, configured to generate corresponding set feedback content based on the feedback set; and a set feedback association module, configured to associate the set feedback content with the target document.

[0125] Based on the above embodiments, the summary generation module 32 includes: a feedback classification submodule, configured to classify the feedback content of the target document to generate at least one feedback set; and a set summary generation submodule, configured to generate corresponding set summary content based on the feedback content in the feedback set.

[0126] Based on the above embodiments, the feedback classification submodule includes: a classification unit, configured to classify the feedback content of the target document based on a preset dimension, and generate multiple feedback sets corresponding to the preset dimension; the preset dimension includes at least one of location, author, time and type.

[0127] Based on the above embodiments, the summary generation module 32 includes: a semantic analysis unit configured to perform semantic analysis on at least two feedback contents to generate summary content, the summary content including at least one of the following: a summary of the feedback content and statistical values ​​of the feedback content.

[0128] Based on the above embodiments, the summary generation module 32 includes: a semantic grouping unit, configured to perform semantic similarity analysis on the feedback content, obtain feedback content groups, and generate summary content for each group; and a summary list generation unit, configured to combine the summary content of each group into a summary list.

[0129] Based on the above embodiments, the summary association module 33 includes: a second traceability entry display submodule, configured to display the summary content of each group and the traceability entry of the feedback content in the corresponding group in the summary list after the summary content is associated with the target document; a second traceability entry access submodule, configured to receive user operations applied to the traceability entry; and a second traceability operation response submodule, configured to respond to user operations and display the feedback content pointed to by the traceability entry.

[0130] Based on the above embodiments, the document feedback content processing device further includes: a target document content acquisition module, configured to acquire target document content associated with the feedback content in the target document after acquiring feedback content for the target document; a feedback content processing module, configured to generate revised content of the target document content or response content of the feedback content based on the target document content and the feedback content; and a processing result association module, configured to associate the revised content or response content with the target document.

[0131] Based on the above embodiments, the feedback content processing module includes: a first relevant feedback determination submodule, configured to determine relevant feedback content from other feedback content based on the feedback content after the target document triggers a modification operation on the target document content associated with the feedback content; and a revision content generation submodule, configured to generate revised content of the target document content associated with the relevant feedback content based on the content before and after the modification operation.

[0132] Based on the above embodiments, the feedback content processing module includes: a second relevant feedback determination submodule, configured to determine relevant feedback content from the remaining feedback content after the target document triggers a first reply operation to the feedback content; and a first reply content generation submodule, configured to generate reply content for the relevant feedback content based on the reply content of the first reply operation.

[0133] Based on the above embodiments, the feedback content processing module includes: a second reply content generation submodule, configured to generate reply content corresponding to the feedback content of the summary content based on the reply content of the second reply operation after the target document triggers a second reply operation on the summary content.

[0134] Based on the above embodiments, the document feedback content processing device further includes a feedback monitoring module, configured to determine the feedback set corresponding to the new feedback content when new feedback content of the target document is detected; and to regenerate the corresponding set summary content based on the feedback content of the feedback set.

[0135] The document feedback content processing apparatus provided in this application embodiment can be used to execute the document feedback content processing method provided in the above embodiment, and has corresponding functions and beneficial effects.

[0136] Figure 13 This is a schematic diagram of the structure of a document feedback content processing device provided in an embodiment of this application, with reference to... Figure 13The document feedback processing device includes a processor 41, a memory 42, a communication device 43, an input device 44, and an output device 45. The number of processors 41 and the number of memories 42 in the document feedback processing device can be one or more. The processor 41, memory 42, communication device 43, input device 44, and output device 45 of the document feedback processing device can be connected via a bus or other means.

[0137] The memory 42, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as program instructions / modules corresponding to the document feedback content processing method in any embodiment of this application (e.g., the feedback acquisition module 31, summary generation module 32, and summary association module 33 in the document feedback content processing apparatus). The memory 42 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the device, etc. Furthermore, the memory 42 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0138] The communication device 43 is used for data transmission.

[0139] The processor 41 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 42, that is, it implements the above-mentioned document feedback content processing method.

[0140] Input device 44 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device. Output device 45 may include display devices such as a display screen.

[0141] The document feedback processing device provided above can be used to execute the document feedback processing method provided in the above embodiments, and has corresponding functions and beneficial effects.

[0142] This application also provides a storage medium containing computer-executable instructions. When executed by a computer processor, the computer-executable instructions are used to perform a method for processing document feedback content. The method for processing document feedback content includes: obtaining feedback content for a target document; generating summary content based on the feedback content; and associating the summary content with the target document.

[0143] Storage medium – any type of memory device or storage device. The term “storage medium” is intended to include: mounting media, such as CD-ROM, floppy disk, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (e.g., hard disk or optical storage); registers or other similar types of memory elements, etc. Storage medium may also include other types of memory or combinations thereof. Furthermore, storage medium may reside in a first computer system in which the program is executed, or it may reside in a different second computer system connected to the first computer system via a network (such as the Internet). The second computer system can provide program instructions to the first computer for execution. The term “storage medium” can include two or more storage media residing in different locations (e.g., in different computer systems connected via a network). Storage medium may store program instructions (e.g., specifically implemented as a computer program) executable by one or more processors.

[0144] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the document feedback content processing method described above, but can also execute related operations in the document feedback content processing method provided in any embodiment of this application.

[0145] The document feedback content processing apparatus, storage medium, and document feedback content processing device provided in the above embodiments can execute the document feedback content processing method provided in any embodiment of this application. For technical details not described in detail in the above embodiments, please refer to the document feedback content processing method provided in any embodiment of this application.

[0146] The above description is merely a preferred embodiment and the technical principles employed in this application. This application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions that can be made by those skilled in the art will not depart from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application. The scope of this application is determined by the scope of the claims.

Claims

1. A method for processing document feedback content, characterized in that, include: Obtain feedback content for the target document; A summary is generated based on the feedback. The summary content is then associated with the target document.

2. The method according to claim 1, characterized in that, The feedback content includes session content related to the target document, and before obtaining the feedback content for the target document, it also includes: Determine the target session of the target document, wherein the target session is a session initiated based on the target document, or the target session is a session that accesses the target document during the session; The process of obtaining feedback content for the target document includes: Obtain target session content related to the target document from the session content of the target session, and use it as feedback content for the target document.

3. The method according to claim 2, characterized in that, The step of obtaining target session content related to the target document from the session content of the target session, and using it as feedback content for the target document, includes: Obtain the target session content related to the target document from the session content of the target session; Determine the target document content that is associated with the target session content in the target document; Set annotations for the target document content based on the target session content.

4. The method according to claim 1, characterized in that, Following the association of the summary content with the target document, the method further includes: The summary content and the source tracing entry are displayed, wherein the source tracing entry points to the feedback content corresponding to the summary content; Receive user operations applied to the source tracing entry point; In response to the user's action, the feedback content pointed to by the tracing entry point is displayed.

5. The method according to claim 1, characterized in that, After obtaining the feedback content for the target document, the process also includes: The feedback content of the target document is categorized to generate at least one feedback set; Generate corresponding set feedback content based on the aforementioned feedback set; The feedback content of the collection is associated with the target document.

6. The method according to claim 1, characterized in that, The process of generating summary content based on the feedback includes: The feedback content of the target document is categorized to generate at least one feedback set; Generate corresponding summary content based on the feedback content in the feedback set.

7. The method according to claim 5 or 6, characterized in that, The step of classifying the feedback content of the target document to generate at least one feedback set includes: The feedback content of the target document is classified based on preset dimensions to generate multiple feedback sets corresponding to the preset dimensions; the preset dimensions include at least one of location, author, time and type.

8. The method according to claim 1, characterized in that, The process of generating summary content based on the feedback includes: Semantic analysis is performed on at least two of the feedback contents to generate summary content, which includes at least one of the following: a summary of the feedback content and statistical values ​​of the feedback content.

9. The method according to claim 1, characterized in that, The summary content includes a summary list, and the generation of summary content based on the feedback content includes: Semantic similarity analysis is performed on the feedback content to obtain feedback content groups, and summary content for each group is generated. Based on the summary content of each group, a summary list is formed.

10. The method according to claim 9, characterized in that, Following the association of the summary content with the target document, the method further includes: The summary list displays the summary content of each group and the source entry point for the feedback content in the corresponding group; Receive user operations applied to the source tracing entry point; In response to the user's action, the feedback content pointed to by the tracing entry point is displayed.

11. The method according to claim 1, characterized in that, After obtaining the feedback content for the target document, the process also includes: Obtain the target document content that is associated with the feedback content from the target document; Based on the target document content and the feedback content, generate revised content of the target document content or a response to the feedback content; The revised content or the response content is associated with the target document.

12. The method according to claim 11, characterized in that, The step of generating revised content for the target document or a response to the feedback content based on the target document content and the feedback content includes: After the target document triggers a modification operation on the target document content associated with the feedback content, relevant feedback content is determined from the remaining feedback content based on the feedback content. Based on the content before and after the modification operation, the revised content of the target document associated with the relevant feedback content is generated.

13. The method according to claim 11, characterized in that, The step of generating revised content for the target document or a response to the feedback content based on the target document content and the feedback content includes: After the target document triggers the first response operation to the feedback content, relevant feedback content is determined from the remaining feedback content based on the feedback content. Based on the response content of the first response operation, generate the response content for the relevant feedback content.

14. The method according to claim 11, characterized in that, The step of generating revised content for the target document or a response to the feedback content based on the target document content and the feedback content includes: After the target document triggers a second response operation to the summary content, the response content corresponding to the summary content is generated based on the response content of the second response operation.

15. The method according to claim 6, characterized in that, The method further includes: Upon detecting new feedback content in the target document, determine the feedback set corresponding to the new feedback content; Based on the feedback content of the feedback set, the corresponding summary content of the set is regenerated.

16. A document feedback content processing device, characterized in that, include: The feedback acquisition module is configured to acquire feedback content for the target document; The summary generation module is configured to generate summary content based on the feedback content; The summary association module is configured to associate the summary content with the target document.

17. A document feedback content processing device, characterized in that, include: One or more processors; A memory that stores one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the document feedback content processing method as described in any one of claims 1-15.

18. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the document feedback content processing method as described in any one of claims 1-15.

Images