Electronic book content review method and device and electronic equipment

Abstract

The invention provides an electronic book content review method and device and electronic equipment. The method comprises the following steps: obtaining the whole content of the electronic book; the content of the electronic book is concentrated into corresponding abstracts according to different numbers of chapters; and according to all the abstracts, expanding the content of the whole electronic book into a plot summary and additional information. According to the method, the content of the whole electronic book is divided into chapters and sections, the content of each chapter and section is concentrated into the corresponding abstract, and all abstracts are expanded into the plot overview and the additional information, so that the review content can completely cover the key venation of the plot, the logic association between the chapters and between the chapters is kept, the selection of hot chapters and sections is avoided, and the review efficiency is improved. The reader can clearly understand the logic venation of the plot after review, so that the review efficiency of the reader is effectively improved, the reading experience is improved, and the book abandoning rate is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to an electronic book content review method and device and an electronic device. BACKGROUND

[0002] Under the background of rapid development of digital reading industry, electronic books have become one of the mainstream reading methods of the public due to their portability and large storage capacity. Especially in the field of network novels, readers often face three typical reading scenarios: first, some works have a long update cycle, which easily leads to a situation of discontinuous updates, making it difficult for readers to catch up with the plot when they resume reading; second, the habit of accumulating chapters before concentrated reading is common, and long intervals between reading sessions can easily cause forgetting of key points such as character relationships and key events; third, the phenomenon of parallel reading of multiple books is common, and readers follow multiple works with different styles and settings at the same time, further exacerbating the confusion of remembering the plot logic, exclusive terminology (such as skill names, world setting), and character attributes (such as skills, wealth) of each book. The above problems directly lead to a decline in the reading experience of readers, and even cause them to give up reading due to not understanding or not remembering, resulting in a high abandonment rate of electronic book works, which not only affects the reading satisfaction of readers, but also restricts the user retention and work dissemination of content platforms.

[0003] To address readers' need for content recall, some existing technologies have emerged related to e-book plot review, but all have significant shortcomings and fail to meet readers' actual needs. For example, patent application number 202411074535.3, entitled "Electronic Reading Interactive Method, Electronic Device, and Storage Medium," focuses on pushing plot prediction tasks to readers. After the author updates a chapter, the accuracy of the reader's predictions is compared with the actual plot. This solution essentially focuses on the matching degree between reader predictions and the plot, failing to effectively support the core pain point of content recall. It only indirectly assists readers in paying attention to new plot points through comparison results, lacking the function of sorting out forgotten historical plots, character relationships, and other key points. Its technological innovation is weak, and it cannot solve the problem of recalling content after a break in reading. Another patent application, number 202310791125.X, entitled "A Method for Generating Novel Plot Recall, Electronic Device, and Storage Medium," employs a chapter popularity-based filtering logic, extracting the content introductions of popular chapters and sorting them chronologically to form plot recall content. This approach has two major flaws: First, the determination of chapter popularity is often related to non-core plot factors such as readers' comments and interesting discussions, which cannot accurately reflect the importance of a chapter in the overall plot. This may lead to the review content focusing on unimportant fragments while missing the core nodes that drive the plot forward. Second, generating a review simply by sorting the popular chapters severs the narrative coherence and logical connection between chapters. When readers are faced with fragmented review content, it may exacerbate their confusion about the plot and prevent them from achieving the goal of quickly connecting the reading flow.

[0004] In summary, current technologies have not yet developed a solution for recalling key points that balances content integrity and logical coherence, addressing the core pain point of e-book readers' tendency to forget what they read. Summary of the Invention

[0005] This invention provides a method, apparatus, and electronic device for reviewing e-book content, in order to solve the shortcomings of existing technologies where the reviewed content is not complete or logically coherent due to readers' interruptions and forgetting.

[0006] This invention provides a method for reviewing e-book content, comprising:

[0007] Get the full content of the ebook;

[0008] The e-book content is condensed into corresponding summaries according to different numbers of chapters;

[0009] Based on all the summaries, the full content of the ebook is expanded into a plot summary and additional information as a review.

[0010] According to the e-book content review method provided by the present invention, after expanding the entire e-book content into a plot summary and additional information, it further includes:

[0011] In response to reader instructions, and based on those instructions, the review content is adjusted;

[0012] The reader instructions include: instructions to end the review early and instructions on the scope of the review.

[0013] According to the e-book content review method provided by the present invention, the plot summary includes: multiple plot summaries corresponding to different reading interruptions;

[0014] The various plot summaries include: a plot summary for a single chapter, a total plot summary for three consecutive chapters, a plot summary for a large chapter, a plot summary for multiple consecutive large chapters, and an outline corresponding to the entire e-book; the large chapter consists of at least four chapters.

[0015] According to the e-book content review method provided by the present invention, the step of dividing the e-book content into chapters and condensing it into corresponding summaries according to different numbers of chapters includes:

[0016] Extract the content of a single chapter and condense it into a first summary; the first summary should have 150-250 words.

[0017] Extract content from multiple consecutive chapters and condense it into a second summary; the second summary should be 1000-1500 words.

[0018] Extract the full content of the e-book and condense it into a third summary; the third summary should be 5000-10000 words long.

[0019] The plot summary includes: multiple plot summaries corresponding to different reading intervals;

[0020] The various plot summaries include: a plot summary for a single chapter, a general plot summary for three consecutive chapters, a plot summary for a large chapter, a plot summary for multiple consecutive large chapters, and an outline corresponding to the entire ebook; each large chapter consists of at least four chapters.

[0021] The process of expanding the entire ebook content into a plot summary based on all the summaries includes:

[0022] Retrieve reader ID, the time interval since the reader last read, the latest chapter read by the reader, and book attributes;

[0023] Input the reader ID, time interval, latest chapter, and book attributes into a pre-trained plot review model, and output a plot summary corresponding to the time interval based on the model;

[0024] The pre-trained plot review model takes the following inputs: reader ID, time interval, latest chapter, and book attributes, and outputs the review range corresponding to the time interval.

[0025] According to the e-book content review method provided by the present invention, the pre-trained plot review model is trained using the following loss function:

[0026] The loss function for a single chapter is as follows:

[0027] Loss1=crossE(yssum,ssum)+k1*crossE(Dlg,Ydlg);

[0028] Where yssum is the predictive summary generated autonomously by the model for the content of a single chapter, ssum is the condensed real summary of a single chapter; Dlg is the real dialogue related to the key logic of the plot in a single chapter; Ydlg is the predictive dialogue extracted by the model from a single chapter; crossE is the cross-entropy function, and k1∈[1.5,4] is the adjustable weight;

[0029] The loss functions for multiple chapters are as follows:

[0030] Loss2 = crossE(ymsum,msum);

[0031] Where, msum is the true summary corresponding to the large chapter; msum is the predictive summary corresponding to the large chapter;

[0032] The loss function for the full ebook is as follows:

[0033] Loss3 = crossE(bsum, ybsum);

[0034] Where bsum is the true summary of the entire book, and ybsum is the predictive summary of the entire book.

[0035] According to the e-book content review method provided by the present invention, the additional information includes: the main character, the main character's participation, and the items owned by the main character; the items include: accessories, skills, attributes, and wealth.

[0036] According to the e-book content review method provided by the present invention, the following formula is used to determine whether there are multiple main characters:

[0037] Z = n + m * 3

[0038] Z>Z1 / K

[0039] Where n is the number of times the character is mentioned; m is the number of chapters in which the character appears; Z1 is the first protagonist; k is the type coefficient, k>1.

[0040] According to the e-book content review method provided by the present invention, based on all the summaries, the full content of the e-book is expanded to include additional information, including:

[0041] Based on the specific information in the attachment, select the chapters that contain that specific information from the chapters already read by the reader;

[0042] Extract the content of the first few chapters based on the frequency of the specific information appearing;

[0043] The content of the preceding chapters is condensed into a summary corresponding to the specific information.

[0044] The present invention also provides an e-book content review device, comprising:

[0045] The acquisition unit is used to acquire the full content of the e-book.

[0046] The condensation unit condenses the e-book content into corresponding summaries according to different numbers of chapters;

[0047] The expansion unit expands the entire ebook content into a plot summary and additional information as a review, based on all the summaries.

[0048] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the e-book content review method as described above.

[0049] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the e-book content review method as described above.

[0050] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the e-book content review method as described above.

[0051] This invention provides an e-book content review method, apparatus, and electronic device. By first breaking down the entire e-book content into chapters and condensing the content of each chapter into corresponding summaries, and then expanding all summaries into plot overviews and additional information, the review content can fully cover the key plot lines and maintain the logical connections between chapters and sections, rather than being a selection of popular chapters. After reviewing, readers can clearly understand the logical connections of the plot, thereby effectively improving the reader's review efficiency, enhancing the reading experience, and reducing the book abandonment rate. Attached Figure Description

[0052] Figure 1 A schematic diagram of the e-book content review method provided by the present invention;

[0053] Figure 2 A schematic diagram of the structure of the e-book content review device provided by the present invention;

[0054] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0056] Figure 1 This is a flowchart illustrating the e-book content review method provided by the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0057] Step 101: Obtain the full content of the e-book.

[0058] Specifically, a complete ebook refers to an ebook work whose content covers all chapters from the beginning to the currently updated / published chapters, including all textual information such as plot, character dialogue, item descriptions, world-building, and proprietary terminology (e.g., cultivation techniques, attributes). This invention can extract all publicly available chapter texts of a target ebook from the official content library of an ebook platform (such as an online novel platform or digital reading app). The complete ebook content also includes the chapter number, publication date, and chapter title for each chapter.

[0059] Step 102: Divide the e-book content into corresponding summaries according to the number of chapters.

[0060] Specifically, this application processes each chapter independently according to the chapter structure of the ebook. Using natural language processing technology, the core content of each chapter is extracted and summaries of different granularities are generated. For example, the 2000-3000 words of each chapter are condensed into a small summary of 150-250 words, retaining key information while meeting the reader's need for quick review.

[0061] Step 103 expands the entire ebook content into a plot summary and additional information as a review based on all the summaries.

[0062] Specifically, based on the generated summaries, a plot overview system is further integrated, covering multiple levels from chapters to the entire book. At the same time, key supplementary information, such as protagonist information and item attributes, is extracted to provide readers with comprehensive content review support. For example, in fantasy novels, the protagonist's skill system and growth trajectory are clearly defined, helping readers quickly understand complex character relationships and world-building.

[0063] The method provided in this application first breaks down the entire content of the e-book into chapters and condenses the content of each chapter into a corresponding summary. Then, all the summaries are expanded into a plot overview and additional information. This allows the review content to fully cover the key plot lines and maintain the logical connections between chapters and sections, rather than being a selection of popular chapters. After reviewing the content, readers can clearly understand the logical connections of the plot (such as why characters move from location A to location B and why events develop from state X to state Y). This avoids the confusion of knowing only a popular scene but not understanding the causes and consequences, thereby effectively improving the reader's review efficiency, enhancing the reading experience, and reducing the book abandonment rate.

[0064] Preferably, after expanding the entire content of the e-book to a plot summary and additional information, the method further includes: responding to a reader's instruction and adjusting the review content based on the instruction; the reader's instruction includes: an instruction to end the review early and an instruction to cover the review range.

[0065] Specifically, regarding the instruction to end the review early, when the system receives the reader's instruction to end the review early, it will simultaneously perform two key operations:

[0066] 1. Immediately stop displaying the current plot summary (and additional information) and jump directly to the last chapter the reader last read (or the next chapter of the main content) to avoid redundant review content taking up the reader's time;

[0067] 2. Automatically store the effective coverage of this review, including the chapters / major sections actually browsed by the reader. This record will serve as the basis for the next review (for example, after the reader stops reading, the system can prioritize pushing the subsequent chapters that were not reviewed last time or extended content based on this record).

[0068] Once the system receives a review scope instruction from the reader, it also allows the reader to manually select the review scope before or during the review process, and the system will generate corresponding review content based on that selection. For example, if the reader still cannot recall all the content after reading the current review content, they can expand the review scope or "manually select the chapters or large sections to review."

[0069] Preferably, the present invention condenses the content of an e-book into a corresponding summary, specifically including: extracting the content of a single chapter and condensing it into a first summary; the first summary has 150-250 words; extracting the content of multiple consecutive chapters and condensing them into a second summary; the second summary has 1000-1500 words; extracting the entire content of the e-book and condensing it into a third summary; the third summary has 5000-10000 words.

[0070] Specifically, this application ensures that key plot points in each chapter are not omitted by condensing the core content of a single chapter into a small-grained summary of 150-250 words. For example, in suspense novels, foreshadowing, clues, and other details in each chapter can be accurately captured through a small-grained summary, providing a foundation for the coherence of subsequent plot developments. By condensing multiple consecutive chapters (such as long chapters or 100 chapters) into medium-grained summaries of 1000-1500 words, the development of local plot lines can be fully presented. For example, in long historical novels, medium-grained summaries can cover the rise and fall of a dynasty or the storyline of three generations of a family, ensuring tight logic within the chapters. Finally, by condensing the entire book into a large-grained outline of 5000-10000 words, a macro perspective can be provided, ensuring that both core plot points (such as main quests and key conflicts) and subplots (such as the growth of supporting characters and background setup) are covered.

[0071] The method provided in this application focuses on details through small-grained summaries, connects local plot points through medium-grained summaries, and constructs the overall framework of the book through large-grained outlines. This creates a progressive logical chain from chapters to sections to the whole, ensuring that the plot summary presented to the reader not only fully covers the key plot lines but also strictly maintains the logical connections between chapters and sections. This approach significantly improves the accuracy of e-book reviews and the user experience, and is particularly suitable for long, complex, or multi-narrative works.

[0072] Preferably, the plot summary includes multiple plot summaries corresponding to different reading intervals. These multiple plot summaries include: a plot summary for one chapter, a total plot summary for three consecutive chapters, a plot summary for a large chapter, plot summaries for multiple consecutive large chapters, and an outline corresponding to the entire ebook; each large chapter consists of at least four chapters.

[0073] Specifically, a chapter's plot summary is based on a single chapter, condensing the core plot of that chapter (such as the cause of key events, development nodes, conflicts and results within the chapter). The length is usually around 150-250 words, suitable for scenarios where you can quickly recall the memory of a single chapter after a short break from reading (such as an interval of 1-2 days).

[0074] The overall plot summary for three consecutive chapters integrates the content of three adjacent chapters and extracts the plot connections between the three chapters (such as the coherent development of events and the logical connection of characters' actions). It is slightly longer than a single chapter summary (about 450-750 words) and is suitable for quickly sorting out short-term continuous plot scenes after a 3-7 day break from reading.

[0075] The extended chapter synopsis is a unit consisting of at least four chapters, condensing the core theme, core conflict, changes in the relationships between main characters, and chapter-level plot twists. It is about 1,000-1,500 words long and is suitable for reviewing the mid-story framework after a 1-2 week break from reading.

[0076] A summary of the plot of several consecutive long chapters integrates the content of 2-3 adjacent long chapters, focusing on the logical connections between chapters (such as the impact of the resolution of the conflict in the previous chapter on the next chapter, and the advancement of the main plot across chapters). The length is about 2000-4500 words. It is suitable for scenarios where you need to quickly connect the long-term plot after a break of more than 1 month.

[0077] The outline corresponding to the entire e-book covers the whole book, condensing the core main plot, key subplots, the growth arc of the main characters, and the connecting role of key props / events. The length corresponds to 5,000-10,000 words, which is suitable for scenarios where you pick up the book again after a break of several months or after reading multiple books in parallel, so as to fully recall the memory of the whole book.

[0078] The method provided by this invention, through a hierarchical plot summary, allows readers to choose a matching review level based on their reading intervals, without being forced to read excessively long or short content. For example, if a reader hasn't read for one day, they only need to read a single chapter summary (1-2 minutes), and if they haven't read for three months, they can directly read the entire book outline (10-15 minutes). This avoids losing patience due to excessively long content or failing to recall memories due to excessively short content. Moreover, readers do not need to manually filter review content (such as searching through hundreds of chapters for previously seen key plot points); they only need to select the corresponding level based on their reading interval to quickly obtain the required information. This is especially suitable for scenarios where multiple books are read in parallel (such as reading three novels simultaneously, where different levels can be used to quickly recall memories of each book).

[0079] Based on all the abstracts, this application expands the entire content of the ebook into a plot summary, specifically including the following solutions:

[0080] Obtain the reader ID, the time interval since the reader last read, the latest chapter read by the reader, and the book attributes; input the reader ID, time interval, latest chapter, and book attributes into a pre-trained plot review model, and based on the model, output a plot overview corresponding to the time interval; the pre-trained plot review model takes the reader ID, time interval, latest chapter, and book attributes as input, and outputs the review range corresponding to the time interval.

[0081] Specifically, different readers require different amounts of review material when reading the same book. For example, if reader A reviews more material than other readers, it indicates that reader A's memory is not as good as other readers, and therefore more review material needs to be shown to that reader. Similarly, if a reader forgets historical and literary books quickly but remembers novelty books well, then more review material should be shown to that reader for historical and literary books, and relatively less for novelty books. Therefore, different review scopes need to be determined based on different reader IDs.

[0082] The so-called book attribute refers to the difficulty level of the book, which is categorized as follows: Classical Chinese > Suspense / Intrigue > Curiosity > Thrilling Read. In other words, if a book is written in classical Chinese, more content will be reviewed; while if it's a thrilling read, the depth and scope of the content will be less.

[0083] Simultaneously, the system automatically adjusts the scope of review content based on the length of time the reader has interrupted reading (e.g., 1 day, 1 week, or 1 month). For example, if the reader interrupts reading for 1 week, the model may push a medium-granular summary (1000-1500 words) of the last 3 chapters to awaken short-term memory; if the interruption is 3 months, it will push a large-granular outline of the entire book (5000-10000 words) to help reconstruct the overall plot framework. This application uses the latest chapter read by the reader as input to the model because the model can accurately locate the reader's memory breakpoint based on this. For example, if the reader interrupts at a key plot point (such as the protagonist encountering a crisis), the model will only push a summary of the plot up to that point, without spoiling anything in advance.

[0084] In summary, this application determines the scope of the review content based on the reader ID, the time interval since the reader last read, the latest chapter read by the reader, and the book attributes. This allows the returned content to be more tailored to the reader's personalized needs, effectively improving the reader's reading experience.

[0085] The pre-trained plot review model provided in this application takes the following inputs as: reader ID, time interval, latest chapter, and book attributes, and outputs the review range corresponding to the time interval.

[0086] Specifically, historical data is first collected, and then a plot recap model is trained based on this historical data. The collected historical data includes:

[0087] Reader ID: A unique identifier for a reader, associated with personalized data such as the reader's historical review habits (e.g., whether they need to review multiple chapters) and memory ability (e.g., whether they belong to the "poor memory" group);

[0088] The time interval between the reader's last reading: quantifying the reading interval (e.g., 1 day, 1 week, 3 months) is the core basis for judging the degree of memory decay;

[0089] The latest chapter read by the reader: Locate the reader's memory breakpoint (such as when they last read Chapter 120), ensure that the plot summary does not spoil unread content in advance, and accurately connect the plot of the read chapter;

[0090] Book attributes: Reflect the difficulty of understanding the book.

[0091] The model inputs the above four types of input data into the plot review model. By learning the mapping relationship between "input parameters → optimal review range" in historical data, the model finally determines and outputs the best review range that is suitable for the current scene.

[0092] The method provided in this application links reader IDs to historical habits, providing customized content for readers with different memory abilities and reading preferences. For example, for readers with poor memory, the model automatically expands the scope of review. Furthermore, it dynamically adjusts review strategies for books of varying difficulty. This personalized adaptation ensures that every reader and every type of book receives just the right amount of review content.

[0093] Preferably, the pre-trained plot recap model provided in this application is trained using the loss function Loss1 for a single chapter, the loss function Loss2 for multiple chapters, and the loss function Loss3 for the entire ebook. The loss function for a single chapter is as follows:

[0094] Loss1=crossE(yssum,ssum)+k1*crossE(Dlg,Ydlg)

[0095] Wherein, yssum is the predictive summary generated autonomously by the model for the content of a single chapter, ssum is the condensed real summary of a single chapter; Dlg is the real dialogue related to the key logic of the plot in a single chapter; Ydlg is the predictive dialogue extracted by the model from a single chapter; crossE is the cross-entropy function, and k1∈[1.5,4] are adjustable weights.

[0096] Specifically, the single-chapter summary model trained by the loss function Loss1 for each chapter can provide readers who need to read briefly with accurate and detailed review content without redundancy, avoiding reading confusion caused by missing details.

[0097] The loss functions for multiple chapters are as follows:

[0098] Loss2 = crossE(ymsum, msum)

[0099] Where, msum is the true summary corresponding to the large chapter; msum is the predictive summary corresponding to the large chapter.

[0100] Specifically, Loss2 addresses the logical breaks caused by excerpts from current technical hot topics through direct constraints between the true and predicted summaries of large sections (crossE(ymsum,msum)). This provides readers with coherent, uninterrupted, and logically clear review content for medium- to long-term reading.

[0101] The loss function for the full ebook is as follows:

[0102] Loss3 = crossE(bsum, ybsum)

[0103] Where bsum is the true summary of the entire book, and ybsum is the predictive summary of the entire book.

[0104] Loss3's training objective is book-wide summary matching. The model automatically identifies and retains the main plot (such as the protagonist searching for a magical artifact) and key subplots (such as the impact of a supporting character's family feud on the main plot), avoiding the problem of existing technologies focusing only on the main plot and thus forgetting subplots. Furthermore, it adapts to multi-book parallel memorization scenarios, helping readers quickly distinguish the worldviews of different books and reducing the risk of memory confusion. Therefore, the loss function corresponding to the entire ebook provides readers who engage in long-term, intermittent reading or multi-book parallel reading with a clear global framework and well-defined core settings for review, effectively reducing the book abandonment rate due to forgetting the overall context.

[0105] Preferably, the additional information includes: the main character, the main character's involvement, and the items the main character possesses; the items include: accessories, skills, attributes, and wealth.

[0106] In the e-book content review system, supplementary information consists of additional review content built upon key elements that readers may easily forget, in addition to the core plot summary. Among these, the protagonist is selected as a core character using a quantitative formula (Z=n+m*3, where n is the number of times the character is mentioned and m is the number of chapters in which the character appears). If multiple characters have a Z value greater than that of the first protagonist, then... 1 The ratio of the type coefficient k (k>1) can be used to determine whether a story has multiple protagonists (such as ensemble novels or works with two male / female protagonists).

[0107] The events in which the protagonist participates refer to the events in which the protagonist leads or plays a key role in the chapters that have been read, including main events (such as the protagonist teaming up to explore a secret realm), side events (such as the protagonist helping a supporting character solve a family crisis), and key turning point events (such as the protagonist discovering the secret of his own origins).

[0108] The protagonist's possessions refer to items / abilities that the protagonist possesses that have plot significance or functional attributes. These can be further divided into accessories (such as jade pendants with special significance or commonly used weapons), skills (such as the cultivation technique Nine Yang Manual or the unlocked special ability teleportation), attributes (such as character level, attack power, spiritual power, and other quantitative indicators), and wealth (such as the amount of gold coins, owned industries, and precious resources).

[0109] The method provided by this invention extracts and structures the above information from the full text of an e-book to form an independent supplementary information database. Readers can quickly retrieve the corresponding content through keyword search (such as searching for the protagonist's skills) without having to search through the plot summary one by one, thereby achieving accurate review.

[0110] Preferably, this application expands the entire content of the e-book into additional information based on all the abstracts, specifically including the following schemes:

[0111] Based on the specific information in the attachment, select chapters containing that specific information from the chapters already read by the reader; extract the content of the first few chapters according to the frequency of the specific information; and condense the content of the first few chapters into a summary corresponding to the specific information.

[0112] Specifically, this application selects chapters containing specific information, counts the frequency of the target information in each chapter (such as the number of times it is mentioned, the length of the description, and the depth of its connection with the plot), and prioritizes extracting the top few chapters with the highest frequency (usually the top 10-20 chapters, which can be dynamically adjusted according to the importance of the information). Then, for the extracted core chapter content, the system calls a natural language processing model (such as deepseek, GPT, etc.) to integrate and condense the information from multiple chapters into a dedicated summary corresponding to the additional information.

[0113] Existing technologies (such as generating reviews based on chapter popularity) often push content that is irrelevant to the reader's needs (e.g., popular chapters may be minor character rants that readers don't care about). This application, however, selects chapters containing specific information from those already read by the reader, and extracts the content of the first few chapters according to the frequency of the specific information to condense the corresponding summaries. This ensures that the summaries corresponding to the specific information can focus only on the additional information that the reader currently needs, thereby ensuring that the review content is supplied on demand.

[0114] Furthermore, by selecting chapters containing specific information from those already read by the reader, this application ensures that all expanded additional information comes from the plot scope already known to the reader, thus preventing premature disclosure of the content of unread chapters and protecting the reader's reading experience.

[0115] Figure 2 A schematic diagram of the structure of the e-book content review device provided by the present invention is shown below. Figure 2 As shown, the device includes:

[0116] Acquisition unit 201 is used to acquire the full content of the e-book;

[0117] The condensation unit 202 is used to condense the e-book content into corresponding summaries according to different numbers of chapters;

[0118] The extension unit 203 is used to expand the full content of the e-book into a plot summary and additional information as review content based on all the summaries.

[0119] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330, and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other via the communication bus 340. The processor 310 can call logical instructions in the memory 330 to execute an e-book content review method, which includes:

[0120] Get the full content of the ebook;

[0121] The e-book content is condensed into corresponding summaries according to different numbers of chapters;

[0122] Based on all the summaries, the full content of the ebook is expanded into a plot summary and additional information as a review.

[0123] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0124] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute the e-book content review method provided by the above methods, the method comprising:

[0125] Get the full content of the ebook;

[0126] The e-book content is condensed into corresponding summaries according to different numbers of chapters;

[0127] Based on all the summaries, the full content of the ebook is expanded into a plot summary and additional information as a review.

[0128] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the e-book content review method provided by the above methods, the method comprising:

[0129] Get the full content of the ebook;

[0130] The e-book content is condensed into corresponding summaries according to different numbers of chapters;

[0131] Based on all the summaries, the full content of the ebook is expanded into a plot summary and additional information as a review.

[0132] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0133] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for reviewing e-book content, characterized in that, include: Get the full content of the ebook; The e-book content is condensed into corresponding summaries according to different numbers of chapters; Based on all the summaries, the full content of the ebook is expanded into a plot summary and additional information as a review.

2. The e-book content review method according to claim 1, characterized in that, After expanding the entire ebook content to include a plot summary and additional information, it also includes: In response to reader instructions, and based on those instructions, the review content is adjusted; The reader instructions include: instructions to end the review early and instructions on the scope of the review.

3. The e-book content review method according to claim 2, characterized in that, The process of dividing the e-book content into chapters and condensing it into corresponding summaries according to different numbers of chapters includes: Extract the content of a single chapter and condense it into a first summary; the first summary should have 150-250 words. Extract content from multiple consecutive chapters and condense it into a second summary; the second summary should be 1000-1500 words. Extract the full content of the e-book and condense it into a third summary; the third summary should be 5000-10000 words long.

4. The method for reviewing e-book content according to any one of claims 1-3, characterized in that, The plot summary includes: multiple plot summaries corresponding to different reading intervals; The various plot summaries include: a plot summary for a single chapter, a general plot summary for three consecutive chapters, a plot summary for a large chapter, a plot summary for multiple consecutive large chapters, and an outline corresponding to the entire ebook; each large chapter consists of at least four chapters. The process of expanding the entire ebook content into a plot summary based on all the summaries includes: Retrieve reader ID, the time interval since the reader last read, the latest chapter read by the reader, and book attributes; Input the reader ID, time interval, latest chapter, and book attributes into a pre-trained plot review model, and output a plot summary corresponding to the time interval based on the model; The pre-trained plot review model takes the following inputs: reader ID, time interval, latest chapter, and book attributes, and outputs the review range corresponding to the time interval.

5. The e-book content review method according to claim 4, characterized in that, The pre-trained plot recap model is trained using the following loss function: The loss function for a single chapter is as follows: Loss1=crossE(yssum,ssum)+k1*crossE(Dlg,Ydlg); Where yssum is the predictive summary generated autonomously by the model for the content of a single chapter, ssum is the condensed real summary of a single chapter; Dlg is the real dialogue related to the key logic of the plot in a single chapter; Ydlg is the predictive dialogue extracted by the model from a single chapter; crossE is the cross-entropy function, and k1∈[1.5,4] is the adjustable weight; The loss functions for multiple chapters are as follows: Loss2 = crossE(ymsum,msum); Where, msum is the true summary corresponding to the large chapter; msum is the predictive summary corresponding to the large chapter; The loss function for the full ebook is as follows: Loss3 = crossE(bsum, ybsum); Where bsum is the true summary of the entire book, and ybsum is the predictive summary of the entire book.

6. The method for reviewing e-book content according to any one of claims 1-5, characterized in that, The additional information includes: the main character, the main character's involvement, and the items the main character possesses; the items include: accessories, skills, attributes, and wealth.

7. The e-book content review method according to claim 6, characterized in that, Determine whether there are multiple main characters using the following formula: Z = n + m * 3 Z>Z1 / K Where n is the number of times the character is mentioned; m is the number of chapters in which the character appears; Z1 is the first protagonist; k is the type coefficient, k>1.

8. The e-book content review method according to claim 6, characterized in that, Based on all the summaries, the full content of the ebook is expanded to include additional information, including: Based on the specific information in the attachment, select the chapters that contain that specific information from the chapters already read by the reader; Extract the content of the first few chapters based on the frequency of the specific information appearing; The content of the preceding chapters is condensed into a summary corresponding to the specific information.

9. An e-book content review device, characterized in that, include: The acquisition unit is used to acquire the full content of the e-book. The condensation unit condenses the e-book content into corresponding summaries according to different numbers of chapters; The expansion unit expands the entire ebook content into a plot summary and additional information as a review, based on all the summaries.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the e-book content review method as described in any one of claims 1 to 9.

Images