Demonstration document generation method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a presentation document generation method and device, electronic equipment and a storage medium, and belongs to the technical field of document processing. The method comprises the following steps: generating a demonstration document according to a demonstration document generation prompt word to obtain an original demonstration document; determining a modification intention for the original presentation document according to the modified presentation document; generating cue words according to the modification intention and the demonstration document to construct a knowledge graph to obtain a joint semantic graph; calculating a semantic matching degree between the presentation document modification prompt word and each semantic node, and performing screening according to the semantic matching degree to obtain a candidate keyword set and semantic weights of candidate keywords; screening the candidate keyword set according to the semantic weight to obtain a target keyword set, and combining keywords in the target keyword set to generate a target prompt word; and performing presentation document updating on the modified presentation document based on the target prompt word to obtain a target presentation document. According to the embodiment of the invention, the generation accuracy of the presentation document can be improved.

Description

Technical Field

[0001] The present application relates to the field of document processing technology, which is applied to the field of financial technology, and in particular to a method and device for generating a presentation document, an electronic device, and a storage medium. Background Art

[0002] A presentation document is a document used for demonstrations. For example, in financial marketing scenarios, business personnel can use a presentation document to present product solutions, product research, or data findings to clients. Technologies for generating presentation documents typically automatically generate presentation documents based on user input. However, these generated presentation documents often deviate from the user's actual needs, requiring users to repeatedly adjust the presentation document content. Therefore, improving the accuracy of presentation document generation has become a pressing issue. Summary of the Invention

[0003] The main purpose of the embodiments of the present application is to propose a method and device for generating a presentation document, an electronic device, and a storage medium, aiming to improve the accuracy of presentation document generation.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application provides a method for generating a presentation document, the method comprising:

[0005] Generate a presentation document according to the presentation document generation prompt words to obtain an original presentation document;

[0006] Performing modification intention analysis on the original presentation document according to the modified presentation document to determine the modification intention; wherein the modified presentation document is obtained by modifying the original presentation document;

[0007] A knowledge graph is constructed based on the modification intention and the prompt words generated in the presentation document to obtain a joint semantic graph;

[0008] Calculating the semantic matching degree between the presentation document modification prompt word and each semantic node in the joint semantic graph, and screening according to the semantic matching degree to obtain a candidate keyword set and the semantic weight of the candidate keyword;

[0009] Screening the candidate keyword set according to the semantic weight to obtain a target keyword set, and combining the keywords in the target keyword set to generate a target prompt word;

[0010] The modified presentation document is updated based on the target prompt word to obtain a target presentation document.

[0011] In some embodiments, generating a presentation document according to a presentation document generation prompt word to obtain an original presentation document includes:

[0012] Performing semantic analysis on the presentation document generation prompt words to obtain a presentation document content outline; wherein the presentation document content outline includes a parent title and a subtitle;

[0013] Generate a cover page according to the parent title and the subtitle, and generate a directory page according to the parent title and the subtitle;

[0014] Generate a text page according to the subtitle;

[0015] The cover page, the directory page and the text page are spliced ​​together to form a presentation document to obtain the original presentation document.

[0016] In some embodiments, performing modification intent analysis on the original presentation document according to the modified presentation document to determine the modification intent includes:

[0017] Performing a presentation document comparison with the original presentation document according to the modified presentation document to determine a presentation document modification record; the presentation document modification record includes a text modification record, a graphic modification record, and a table modification record;

[0018] Extracting keywords from the text modification record to obtain modification keywords;

[0019] Performing operation classification on the graphic modification record to obtain an image modification type;

[0020] Performing behavior recognition on the table modification record to obtain the table modification behavior;

[0021] The modification intention is determined according to the modification keyword, the image modification type and the table modification behavior.

[0022] In some embodiments, updating the modified presentation document based on the target prompt word to obtain a target presentation document includes:

[0023] updating elements of the modified presentation document according to the target prompt word to obtain an updated presentation document;

[0024] The updated presentation document is adjusted according to the target prompt word to obtain the target presentation document.

[0025] In some embodiments, updating elements of the modified presentation document according to the target prompt word to obtain an updated presentation document includes:

[0026] By using an attention model, performing attention calculation on presentation document elements in the modified presentation document according to the target prompt word to obtain element contribution;

[0027] Screening the presentation document elements in the modified presentation document according to the element contribution to obtain the original presentation document elements;

[0028] The original presentation document elements in the modified presentation document are replaced according to the material presentation document elements to obtain the updated presentation document.

[0029] In some embodiments, replacing the original presentation document elements in the modified presentation document according to the material presentation document elements to obtain the updated presentation document includes:

[0030] Performing semantic vectorization on the material presentation document elements to obtain a material presentation document vector, and performing semantic vectorization on the target prompt word to obtain a semantic prompt word vector;

[0031] Calculating the similarity between the material presentation document vector and the semantic prompt word vector to obtain semantic similarity, and screening the preset presentation document material according to the semantic similarity to obtain a target presentation document element;

[0032] The original presentation document element of the modified presentation document is updated according to the target presentation document element to obtain the updated presentation document.

[0033] In some embodiments, updating the original presentation document element of the modified presentation document according to the target presentation document element to obtain the updated presentation document includes:

[0034] Screening the target presentation document elements according to the types of the original presentation document elements to obtain selected presentation document elements;

[0035] Calculating the similarity between the selected presentation document element and the original presentation document element, and screening the selected presentation document element based on the similarity to obtain a replacement presentation document element;

[0036] The original presentation document element in the updated presentation document is replaced according to the replacement presentation document element to obtain the updated presentation document.

[0037] To achieve the above-mentioned purpose, a second aspect of an embodiment of the present application provides a presentation document generation device, the device comprising:

[0038] A presentation document generation module is used to generate a presentation document according to the presentation document generation prompt words to obtain an original presentation document;

[0039] an intention parsing module, configured to parse the original presentation document for modification intent based on the modified presentation document to determine the modification intent; wherein the modified presentation document is obtained by modifying the original presentation document;

[0040] A graph construction module is used to construct a knowledge graph based on the modification intention and the prompt words generated in the presentation document to obtain a joint semantic graph;

[0041] A keyword screening module is used to calculate the semantic matching degree between the presentation document modification prompt word and each semantic node in the joint semantic graph, and screen according to the semantic matching degree to obtain a set of candidate keywords and the semantic weights of the candidate keywords;

[0042] a prompt word combination module, configured to screen the candidate keyword set according to the semantic weight to obtain a target keyword set, and combine the keywords in the target keyword set to generate a target prompt word;

[0043] The presentation document updating module is used to update the modified presentation document based on the target prompt word to obtain a target presentation document.

[0044] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the method described in the first aspect when executing the computer program.

[0045] To achieve the above-mentioned purpose, the fourth aspect of the embodiments of the present application proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the method described in the first aspect.

[0046] The present application proposes a method and device for generating a presentation document, an electronic device, and a storage medium, which generates an original presentation document based on a presentation document generation prompt word, and then analyzes the modification intention of the original presentation document based on the modified presentation document, thereby determining the modification direction expected by the user; then, a knowledge graph is constructed by combining the modification intention with the generation prompt word to obtain a joint semantic graph that can reflect the semantic structure of the presentation document, and representative candidate keywords and their corresponding semantic weights are screened out through semantic matching calculation; on this basis, the candidate keywords are further screened and combined according to the semantic weight to generate a target prompt word that is highly matched with the user's intention; finally, the modified presentation document is updated based on the target prompt word to obtain a target presentation document with a complete semantic structure and accurate content expression. By combining semantic graph construction, keyword screening, and prompt word reconstruction, the present application effectively improves the semantic adaptability and content rationality in the process of automatic modification of presentation documents, and solves the problem of deviation between generated content and user intention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 This is a flowchart of a method for generating a presentation document provided by an embodiment of the present application;

[0048] Figure 2 yes Figure 1 Flowchart of step S101 in FIG.

[0049] Figure 3 yes Figure 1 Flowchart of step S102 in FIG.

[0050] Figure 4 yes Figure 1 Flowchart of step S106 in FIG.

[0051] Figure 5 yes Figure 4 Flowchart of step S401 in FIG.

[0052] Figure 6 yes Figure 5 Flowchart of step S503 in FIG.

[0053] Figure 7 yes Figure 6 Flowchart of step S603 in FIG.

[0054] Figure 8 It is a structural diagram of the presentation document generating device provided in an embodiment of the present application;

[0055] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0057] It should be noted that although the device schematics illustrate functional module divisions and the flowcharts illustrate logical sequences, in certain circumstances, the steps shown or described may be performed in a sequence that differs from the module divisions in the device or the sequence in the flowcharts. The terms "first," "second," and so on, in the specification, claims, and drawings, are used to distinguish similar items and are not necessarily used to describe a specific sequence or precedence.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0059] A presentation document is a document used for demonstrations. For example, in financial marketing scenarios, business personnel can use a presentation document to present product solutions, product research, or data findings to clients. Technologies for generating presentation documents typically automatically generate presentation documents based on user input. However, these generated presentation documents often deviate from the user's actual needs, requiring users to repeatedly adjust the presentation document content. Therefore, improving the accuracy of presentation document generation has become a pressing issue.

[0060] Based on this, the embodiments of the present application provide a method and device for generating a presentation document, an electronic device, and a storage medium, aiming to improve the accuracy of presentation document generation.

[0061] The embodiments of the present application provide a method and device for generating a presentation document, an electronic device, and a storage medium, which are specifically described through the following embodiments. First, the method for generating a presentation document in the embodiments of the present application is described.

[0062] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0063] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0064] The method for generating a presentation document provided in the embodiment of the present application relates to the field of document processing technology and is applied to the field of financial technology. The method for generating a presentation document provided in the embodiment of the present application can be applied in a terminal, can be applied in a server side, or can be software running in a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or as a server cluster or distributed system composed of multiple physical servers, or as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the method for generating a presentation document, etc., but is not limited to the above forms.

[0065] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments in which tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media, including storage devices.

[0066] It should be noted that in each specific embodiment of the present application, when it comes to the need to perform relevant processing based on data related to the user's identity or characteristics, such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of such data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present application needs to obtain the user's sensitive personal information, the user's separate permission or consent will be obtained through a pop-up window or by jumping to a confirmation page. After clearly obtaining the user's separate permission or consent, the necessary user-related data for the normal operation of the embodiment of the present application will be obtained.

[0067] Figure 1 This is an optional flowchart of the method for generating a presentation document provided in an embodiment of the present application. Figure 1 The method may include but is not limited to steps S101 to S106.

[0068] Step S101, generating a presentation document according to the presentation document generation prompt words to obtain an original presentation document;

[0069] Step S102, analyzing the modification intention of the original presentation document according to the modified presentation document to determine the modification intention; wherein the modified presentation document is obtained by modifying the original presentation document;

[0070] Step S103, constructing a knowledge graph based on the modification intention and the prompt words generated in the presentation document to obtain a joint semantic graph;

[0071] Step S104, calculating the semantic matching degree between the presentation document modification prompt word and each semantic node in the joint semantic graph, and screening according to the semantic matching degree to obtain a candidate keyword set and the semantic weight of the candidate keyword;

[0072] Step S105, screening the candidate keyword set according to the semantic weight to obtain the target keyword set, and combining the keywords in the target keyword set to generate the target prompt word;

[0073] Step S106: updating the modified presentation document based on the target prompt word to obtain a target presentation document.

[0074] In steps S101 to S106 shown in the embodiment of the present application, an original presentation document is generated based on the prompt words generated by the presentation document, and then the modification intention of the original presentation document is parsed based on the modified presentation document, so as to determine the modification direction expected by the user; then, a knowledge graph is constructed by combining the modification intention and the generated prompt words to obtain a joint semantic graph that can reflect the semantic structure of the presentation document, and representative candidate keywords and their corresponding semantic weights are screened out through semantic matching calculation; on this basis, the candidate keywords are further screened and combined based on the semantic weights to generate target prompt words that are highly matched with the user's intention; finally, the modified presentation document is updated based on the target prompt words to obtain a target presentation document with a complete semantic structure and accurate content expression. This embodiment effectively improves the semantic adaptability and content rationality in the process of automatic modification of presentation documents by combining semantic graph construction, keyword screening and prompt word reconstruction, and solves the problem of deviation between generated content and user intention.

[0075] See also Figure 2 In some embodiments, step S101 may include but is not limited to steps S201 to S204:

[0076] Step S201: semantically analyze the prompt words generated by the presentation document to obtain the content outline of the presentation document; wherein the content outline of the presentation document includes a parent title and a subtitle;

[0077] Step S202: Generate a cover page based on the parent title and subtitle, and generate a directory page based on the parent title and subtitle;

[0078] Step S203, generating a text page according to the subtitle;

[0079] Step S204: splice the cover page, the table of contents page and the body page into a presentation document to obtain an original presentation document.

[0080] In steps S201 to S204, illustrated in the embodiment of the present application, semantic analysis is performed on the presentation document generation prompt words to obtain a presentation document content outline, including the parent title and subtitles. Based on the content outline, a cover page, a table of contents page, and a body page are generated. Ultimately, these pages are spliced ​​together to obtain a structurally complete original presentation document. This converts the presentation document generation prompt words into a semantically clear and hierarchical content outline, and automatically generates various pages based on the content outline. This ensures that the generated original presentation document has a unified logical structure and complete information organization, significantly improving the practicality and accuracy of automatic presentation document generation.

[0081] In step S201 of some embodiments, the presentation document generation prompt refers to text input information used to indicate the generation of a target presentation document. The presentation document generation prompt can be one or more phrases, keyword groups, question-like input, or declarative statements. For example, when a user enters "carbon neutrality technology path analysis" as the presentation document generation prompt, it indicates a desire to generate a presentation document analyzing technology paths around the theme of carbon neutrality.

[0082] Semantic parsing refers to the process of performing a structured analysis on the semantics contained in the prompt words generated by the presentation document. The implementation of semantic parsing may include but is not limited to syntactic analysis and semantic role labeling of prompt words based on a pre-trained language model, further extracting the core concept phrases that represent the subject content, and matching them with the preset semantic structure template through the subject extraction algorithm to obtain a structured semantic output. For example, for "Carbon Neutrality Technology Path Analysis", the semantic parsing process can identify "carbon neutrality" as the core subject word, "technology path" as the discussion dimension, and "analysis" as the content intention, thereby generating a presentation document content outline including the parent title "Carbon Neutrality" and subtitles such as "Energy Structure Adjustment", "Industrial Upgrading Strategy", and "Emission Monitoring Means".

[0083] The content outline of a presentation document is a set of structured texts used to guide the organization of the content of a presentation document page. The content outline of a presentation document contains at least one parent title and multiple sub-titles that are semantically associated with the parent title, and are arranged in a logical order. For example, when the prompt word generated by the presentation document is "Urban Traffic Congestion Governance Plan", the content outline of the presentation document can include "Current Status of Urban Traffic Governance" as the parent title, and "Analysis of Causes of Congestion", "Evaluation of Current Policies", "Application of Intelligent Transportation Technology", and "Future Governance Trends" as the corresponding sub-title sets.

[0084] A parent title is a first-level heading in the presentation document's content outline that represents the overall theme or main line of logic. For example, "Overview of AI Applications" in "Applications of AI in Healthcare" could be a parent title. A sub-topic is a subtopic corresponding to a parent title, reflecting the specific content of each branch within that topic. For example, sub-topics could include "Medical Image-Assisted Diagnosis," "Construction of an Intelligent Diagnosis System," and "Clinical Data Mining Methods."

[0085] In step S202 of some embodiments, the cover page is located at the beginning of the original presentation document and is used to display the subject information of the presentation document. The directory page is the page following the cover page and is used to list the title structure of the main text pages and provide content guidance.

[0086] The cover page and table of contents are generated using a pre-trained page generation model. For example, given the parent title "Smart City Construction Pathway" and a set of sub-titles including "Digital Infrastructure," "Intelligent Transportation System Scheduling," and "Environmental Monitoring Network Deployment," the model will generate the main title "Smart City Construction Pathway" on the cover page, and the table of contents will list the sub-titles.

[0087] In step S203 of some embodiments, the body page refers to a set of pages generated based on the subheadings in the presentation document's content outline. The body page is located after the cover page and the table of contents, and is generated page by page according to the order in which the subheadings are arranged in the content outline. The content contained in the body page typically includes, but is not limited to, subheading text, explanatory paragraphs related to the subheadings, bullet point information, chart data, citations, image information, or other structured presentation content.

[0088] The method of generating the main text page is the same as the principle of generating the cover page and the directory page, so I will not go into details here.

[0089] In step S204 of some embodiments, the presentation document splicing of the cover page, the directory page, and the body page refers to integrating the generated pages into a presentation document set with a continuous structure according to a preset page sequence.

[0090] It should be noted that the reason why the presentation document generation prompt words are split into the presentation document content outline, and on this basis, the cover page, table of contents page and main text page are generated according to the parent title and subtitle respectively, and then the pages are spliced ​​together, is because in the process of generating the presentation document content based on the model, directly using the prompt words as input to generate the entire presentation document can easily lead to problems such as lack of thematic focus, chaotic structure and semantic jumps in the page content. By structuring the prompt words into a presentation document content outline including parent titles and subtitles, the original semantic information can be converted into a set of titles with hierarchical relationships and content boundaries, forming a structural constraint input, and providing clear and unambiguous semantic guidance for subsequent model generation. When processing the structured outline input, the model can generate content page by page based on the title level, and construct the semantic blocks required for different pages respectively, thereby reducing the risk of information drift, improving the semantic coherence and structural consistency between the content of each page, and thus improving the overall quality of the generated presentation document.

[0091] See also Figure 3 In some embodiments, step S102 may include but is not limited to steps S301 to S305:

[0092] Step S301, performing a comparison between the modified presentation document and the original presentation document to determine a presentation document modification record; the presentation document modification record includes a text modification record, a graphic modification record, and a table modification record;

[0093] Step S302: extract keywords from the text modification record to obtain modification keywords;

[0094] Step S303, classifying the graphics modification records to obtain the image modification type;

[0095] Step S304: performing behavior recognition on the table modification record to obtain the table modification behavior;

[0096] Step S305: Determine the modification intention based on the modification keyword, the image modification type, and the table modification behavior.

[0097] In steps S301 to S305, the modified presentation document is compared with the original presentation document to extract multi-dimensional modification record information, including text modification records, graphic modification records, and table modification records. Semantic features are extracted and classified for each type of modification record, and then integrated to obtain a unified modification intention. In this way, by dividing the modified content into three dimensions: text, graphics, and tables and processing them separately, the user's modification intention for different content types can be captured more specifically, improving the accuracy of the analysis of presentation document modification behavior.

[0098] In step S301 of some embodiments, presentation document comparison refers to the process of identifying content changes between the modified presentation document and the original presentation document based on a log-based tracing method. Logging refers to recording each user's editing operations during the presentation document editing process, generating structured log data containing information such as the operation time, operation object, operation type, and the state before and after the operation.

[0099] A text modification record is a collection of records generated by editing text information during the editing process of a presentation document. This includes adding, deleting, or modifying text content such as paragraph headings, body text, list items, or comments. For example, if the phrase "Improve user satisfaction" in a body paragraph in a presentation document is changed to "Optimize user experience," the corresponding text modification record is a text replacement operation.

[0100] Graphic modification records refer to edit operations on graphic elements within a presentation document. These records include graphic insertion, deletion, color adjustment, resizing, and relocation. For example, if a user deletes a flowchart and adds a bar chart in the original presentation document, the graphic modification record will contain two entries: one for the deleted graphic and one for the added graphic.

[0101] Table modification records refer to the information generated by structural adjustments or data updates to table elements within a presentation document. These records may include operations such as adding or deleting rows and columns, merging or splitting cells, replacing numbers, revising text, adjusting borders, and changing background colors. For example, if a user expands a three-column data table to five columns and updates some sales data values ​​in a presentation document, the table modification record will include both the column structure expansion and the cell data modification.

[0102] In step S302 of some embodiments, keyword extraction means: first extracting keywords from the modified text in the text modification record, and then combining the modified keywords and the operation type to form an operation-keyword pair.

[0103] For example, if "data processing capability improvement" is changed to "system data processing capability optimization", keyword extraction is used to identify "system" and "optimization" as new keywords. At the same time, operation-keyword pairs are formed based on "addition": "add-system", "add-optimization", and "delete-improvement".

[0104] In step S303 of some embodiments, operation classification involves extracting graphic modification operations and associating these operations with specific graphic objects to form structured data of operation-graphic pairs. Operation types include, but are not limited to, adding, deleting, scaling, moving, rotating, changing color, and adjusting style.

[0105] For example, when a user adds a flowchart representing a business process in a modified presentation document, the operation type "add" is extracted and the associated graphic type "flowchart" is formed to form an operation-graphic pair: "add-flowchart".

[0106] For example, when a user scales an existing bar chart and adjusts its width to 1.5 times the original size, the operation type "zoom" is extracted and the associated graphic type "bar chart" is associated, forming an operation-graphic pair: "zoom-bar chart".

[0107] In step S304 of some embodiments, action recognition involves extracting table operations and associating them with specific table objects to form structured data for operation-table pairs. Operation types include, but are not limited to, adding columns, deleting columns, merging cells, splitting cells, scaling the entire table, modifying cell content, and adjusting table styles.

[0108] For example, when a user adds two columns, "quarterly growth rate" and "annual cumulative value", to an original sales data table in a modified presentation document, the operation type "add column" is extracted and the table object is associated to form an operation-table pair.

[0109] In step S305 of some embodiments, determining the modification intention refers to performing intention pattern recognition and classification based on operation-keyword pairs, operation-graphic pairs, and operation-table pairs to infer the modification purpose reflected by the user in the process of editing the presentation document. Specifically, the identification of the modification intention can be achieved by:

[0110] In one embodiment, for different types of operation pairs, corresponding modification intention rule sets are preset, and the extracted operation pairs are matched with rules to determine the user's intention type when modifying the presentation document. For example, when it is detected that there are a large number of operation types of "new" and the keywords contain "background information", "preconditions" and other content, it can be determined that the user's intention is to "add pre-instructions". For example, when it is detected that there is an operation type of "new" and the graphic type is "icon", it can be determined that the user's intention is to "add auxiliary instructions to the icon". For example, when it is detected that the operation type is "modify" and the keywords involve words related to easing the tone (such as "must" is changed to "suggested", "must be completed" is changed to "can be adjusted appropriately"), it can be inferred that the user's intention is to "ease the expression tone".

[0111] In another embodiment, various types of operation pairs are used as input to train an intent classification model, and the model outputs the corresponding modification intention category. For example, if there are multiple text additions in the input operation pair, and the newly added keywords are mainly explanatory or background words, the model prediction result is "adding explanatory content". If the input operation pair is mostly composed of graphic additions, and the graphic types are concentrated in flow charts, annotation icons, etc., the model prediction result is "enriching visual expression". If the input operation pair is mainly composed of fine-tuning of text content and weakening of tone-related modifications, the model prediction result is "adjusting the expression tone".

[0112] It should be noted that in the generation of presentation documents, by performing refined feature extraction for the text, graphic, and table dimensions respectively, the details of changes in various content elements during the editing process are more fully preserved. Content elements of different dimensions have different expression functions, forms of expression, and modification methods in presentation documents. If a unified abstract processing is adopted, it is easy to cause confusion between features of various dimensions and fail to accurately capture the user's adjustment intentions at the specific content level. Therefore, through multi-dimensional processing, semantic changes can be identified in the text dimension, visual layout adjustments can be identified in the graphic dimension, and structural reconstruction and data updates can be identified in the table dimension, forming a highly targeted and clearly layered modification feature system.

[0113] In step S103 of some embodiments, it is first necessary to explain the sequential relationship between the presentation document generation prompt and the modification intention. The presentation document generation prompt is text information input by the user in the initial stage to guide the generation of the original presentation document. The modification intention is obtained by analyzing the differences between the modified presentation document and the original presentation document after the user performs modification operations based on the generation results after the original presentation document is generated. The presentation document generation prompt is formed before the original presentation document is generated, while the modification intention is reversely deduced through the user's editing operations after the original presentation document is generated and displayed to the user.

[0114] During the knowledge graph construction process, using the pre-defined domain knowledge graph as the basic framework, we first search for the corresponding topic nodes and their associated nodes in the pre-defined domain knowledge graph based on the prompt words generated by the presentation document, extracting and forming a preliminary subgraph. Subsequently, based on the modification intent obtained through analysis, we make adjustments to the preliminary subgraph, including but not limited to:

[0115] Add new nodes corresponding to the modification intention;

[0116] Adjust to strengthen or weaken the semantic connection between existing nodes;

[0117] Update node attributes based on intent to obtain a joint semantic graph.

[0118] By combining presentation document generation prompts and modification intentions in the construction of a joint semantic graph, the user's full-process expression intention in the presentation document generation task can be comprehensively reflected. On the one hand, presentation document generation prompts serve as the user's initial input, reflecting the user's desired topic direction and focus before content generation. On the other hand, modification intentions are formed by the user's manual editing after the original presentation document is generated, and actually reflect the user's acceptance of the generated content, that is, the user's subjective satisfaction with the content generation results and adjustment demands. Combining the two for semantic graph construction can more comprehensively cover the user's real needs than building a map based on only a single input dimension.

[0119] In step S104 of some embodiments, the semantic matching degree is calculated by calculating the semantic matching degree between the presentation document modification prompt word and each semantic node in the joint semantic graph one by one. The semantic matching degree can be calculated by cosine similarity or Euclidean distance.

[0120] The following methods can be used to obtain the candidate keyword set and the semantic weight of the candidate keyword based on the semantic matching degree:

[0121] First, the modification prompts of the demonstration document are vectorized with the semantic nodes in the combined semantic graph. Subsequently, cosine similarity or Euclidean distance is used to obtain the matching scores of each group of modification prompts of the demonstration document and the semantic nodes. Then, screening is performed according to the score magnitude to select a set of semantic nodes with high semantic relevance to the modification prompts of the demonstration document as the candidate keyword set.

[0122] The screening method can be based on a preset threshold or take the top N after sorting.

[0123] Furthermore, the semantic weight is used to represent the importance degree of the candidate keywords in terms of semantic relevance. The value can be a continuous value within a preset range, such as a decimal between 0 and 1. The higher the weight value, the closer the semantic association between the keyword and the modification prompt of the demonstration document. The semantic weight can be the semantic matching degree calculated above or a preset weight value.

[0124] In step S105 of some embodiments, the candidate keyword set screening is to screen the candidate keyword set according to the semantic weight. First, weight sorting and screening processing are performed on each candidate keyword to select keywords that meet the preset weight threshold requirements or satisfy the relative ranking conditions, forming the target keyword set.

[0125] Furthermore, combination processing is performed on the keywords in the target keyword set. The specific implementation method is as follows: Based on the semantic weight of the target keywords, according to the set combination rules, several target keywords with higher weights are arranged and connected in an orderly manner to generate a new prompt.

[0126] For example, several keywords with the highest semantic weights in the target keyword set can be selected and combined according to fixed connectors (such as natural language connectors like "of", "and", "as well as", etc.) to generate a new prompt that conforms to the language expression habit. For example, when the target keyword set includes "intelligent manufacturing", "data collection", "edge computing", and the weight order is "edge computing", "data collection", "intelligent manufacturing" in sequence, the keywords can be arranged in order according to the weight sorting rule, and connectors are inserted to combine and generate the target prompt "Data collection and intelligent manufacturing based on edge computing".

[0127] Please refer to Figure 4 , in some embodiments, step S106 may include but is not limited to steps S401 to S402:

[0128] Step S401, update the elements of the modified demonstration document according to the target prompt to obtain the updated demonstration document;

[0129] Step S402, adjust the demonstration document according to the target prompt for the updated demonstration document to obtain the target demonstration document.

[0130] In steps S401 and S402 of the embodiment of the present application, elements in the modified presentation document are first updated based on the target prompt word to obtain an updated presentation document. Subsequently, a target presentation document is further generated based on the target prompt word. This achieves a two-stage optimization of the presentation document content and structure in conjunction with the target prompt word, resulting in a target presentation document that better meets the user's actual needs in terms of theme expression, semantic coherence, and layout arrangement.

[0131] See also Figure 5 In some embodiments, step S401 includes but is not limited to steps S501 to S503:

[0132] Step S501: Using an attention model, attention is calculated on presentation document elements in the modified presentation document according to the target prompt word to obtain element contribution.

[0133] Step S502, screening the presentation document elements in the modified presentation document according to the element contribution to obtain the original presentation document elements;

[0134] Step S503: replacing the original presentation document elements in the modified presentation document according to the material presentation document elements to obtain an updated presentation document.

[0135] In steps S501 to S503 shown in the embodiment of the present application, attention is calculated for each presentation document element in the modified presentation document according to the target prompt word through the attention model, and the element contribution of the importance of each element under the current semantic requirements is obtained. Subsequently, the presentation document elements in the modified presentation document are screened according to the element contribution, and the original presentation document elements with a high correlation with the target prompt word and playing a major role in the semantic expression are screened out. Furthermore, according to the material presentation document elements, the original presentation document elements obtained by screening are replaced to generate an updated presentation document containing the updated content. In this way, the importance of each presentation document element in the modified presentation document can be quantitatively identified, and the specific element objects that need to be retained or replaced can be accurately determined during the update process, thereby improving the pertinence and rationality of the presentation document content update.

[0136] In step S501 of some embodiments, attention calculation refers to determining the importance evaluation of each presentation document element in the modified presentation document based on a pre-trained attention model and according to the target prompt word, and obtaining the element contribution that represents the degree of contribution of each presentation document element under the semantics of the current target prompt word.

[0137] The attention model can employ a deep neural network with a Transformer structure. By calculating attention weights between the target cue word and the feature vectors of the presentation document elements, it determines the importance of each element in representing the semantic meaning of the target cue word. Other deep neural networks known in the art may also be employed.

[0138] The element contribution is usually expressed in numerical form, for example, ranging from 0 to 1, where a larger numerical value indicates a higher contribution of the element in semantic matching.

[0139] In step S502 of some embodiments, the presentation document elements in the modified presentation document are screened according to the element contribution, and the original presentation document elements are obtained by screening according to a preset threshold or a relevant ranking.

[0140] See also Figure 6 In some embodiments, step S503 includes but is not limited to steps S601 to S603:

[0141] Step S601: semantically vectorize the material presentation document elements to obtain a material presentation document vector, and semantically vectorize the target prompt word to obtain a semantic prompt word vector;

[0142] Step S602: Calculate the similarity between the material presentation document vector and the semantic prompt word vector to obtain semantic similarity, and filter the preset presentation document materials based on the semantic similarity to obtain the target presentation document element;

[0143] Step S603: updating the original presentation document element of the modified presentation document according to the target presentation document element to obtain an updated presentation document.

[0144] In the steps S601 to S603 shown in the embodiment of the present application, first, semantic vectorization is performed on the material presentation document elements, and the material presentation document elements are mapped as material presentation document vectors in the semantic space, and the target prompt words are subjected to the same semantic vectorization to obtain semantic prompt word vectors. Subsequently, based on the vectorized representation, the semantic similarity between the material presentation document vectors and the semantic prompt word vectors is calculated. Further, according to the semantic similarity, the preset material presentation document elements are screened, and the material presentation document elements that are highly relevant to the target prompt word semantics are screened out to form a target presentation document element set. Finally, according to the target presentation document elements, the original presentation document elements in the modified presentation document are updated at the element level to generate an updated presentation document. In this way, material screening and replacement operations based on semantic understanding are realized, ensuring that the updated content introduced into the modified presentation document is highly consistent with the target prompt word semantically, and improving the selection and replacement accuracy of the presentation document material.

[0145] In step S601 of some embodiments, semantic vectorization refers to mapping the material presentation document elements to vector representations through a preset semantic coding model to obtain a material presentation document vector; at the same time, the target prompt word is subjected to the same semantic coding process to obtain a semantic prompt word vector.

[0146] In step S602 of some embodiments, similarity calculation involves comparing the material presentation document vector with the semantic cue word vector, using a similarity measurement algorithm to calculate the degree of correlation in the semantic space to obtain semantic similarity. Semantic similarity can be calculated using cosine similarity, Euclidean distance, or other vector space measurement methods, and the value is typically set between 0 and 1, with larger values ​​indicating stronger semantic relevance.

[0147] The screening method may be to select material presentation document elements whose semantic similarity with the target prompt word is higher than a set threshold as the target presentation document element set.

[0148] See also Figure 7 In some embodiments, step S603 may include but is not limited to steps S701 to S703:

[0149] Step S701, screening target presentation document elements according to the types of original presentation document elements to obtain selected presentation document elements;

[0150] Step S702, calculating the similarity between the selected presentation document element and the original presentation document element, and screening the selected presentation document element based on the similarity to obtain a replacement presentation document element;

[0151] Step S703: Replace the original presentation document element in the updated presentation document according to the replacement presentation document element to obtain an updated presentation document.

[0152] In steps S701 to S703 shown in the embodiment of the present application, the target presentation document elements are first screened according to the type of the original presentation document elements to ensure that the selected presentation document elements are consistent with the original presentation document elements in terms of element type, and a set of selected presentation document elements is obtained. Subsequently, the content similarity between the selected presentation document elements and the original presentation document elements is calculated, and the selected presentation document elements are further screened based on the similarity to obtain replacement presentation document elements that meet the content matching requirements. Finally, the original presentation document elements in the updated presentation document are replaced according to the replacement presentation document elements to generate an updated presentation document. In this way, through a dual screening mechanism based on element type and content similarity, the accuracy of the updated content and the consistency of expression of the presentation document elements are guaranteed.

[0153] In step S701 of some embodiments, screening the target presentation document elements according to the type of the original presentation document elements means screening out a set of elements of the same type from the target presentation document elements using the type information of the original presentation document elements to be updated in the modified presentation document as a screening basis.

[0154] In step S702 of some embodiments, similarity calculation refers to matching preset indicators to obtain the similarity between the two. For example, the indicators include: the layout position of the element (such as being located in the title area, body area, or graphic area), the content module to which the element belongs (such as title text, body text, graphic, table), and the basic size category of the element (such as large size, medium size, small size).

[0155] Specifically, the selected presentation document element can be judged by checking whether its layout partitions, content module types, and size categories are consistent with those of the original presentation document element. Different similarity scores can be assigned based on the degree of match. For example, a score of 1 is assigned when all features are consistent, a score of 0.5 is assigned when only some features are consistent, and a score of 0 is assigned when they are inconsistent. Finally, a replacement presentation document element with features closest to those of the original presentation document element can be selected based on the similarity score.

[0156] In step S402 of some embodiments, the updated presentation document is adjusted according to the target prompt word. Obtaining the target presentation document means inputting the target prompt word into a pre-trained page modification model to modify each page of the updated presentation document.

[0157] In one embodiment, the present embodiment receives a user input of a presentation document generation prompt word "Smart City Construction Path," generates an original presentation document based on the presentation document generation prompt word, and includes several pages of content, such as modules such as "Smart Transportation System Construction," "Smart Grid Deployment," and "Urban Governance Platform Construction."

[0158] Subsequently, the user edits and adjusts the original presentation document based on actual needs to generate a modified presentation document. For example, the page about "Traditional Infrastructure Maintenance" in the original presentation document is deleted, and the page content related to "Green Building Application Demonstration Project" is added. This embodiment compares the content of the original and modified presentation documents to analyze the corresponding modification intentions, such as strengthening the direction of "Green Building Application" and weakening the content of "Traditional Infrastructure Construction".

[0159] After confirming the modification intent, the user further enters the presentation document modification prompt, "Integrated Application of Green Building and Intelligent Energy-Saving Systems." This embodiment constructs a knowledge graph based on the modification intent and the prompt generated in the presentation document, resulting in a joint semantic graph. This joint semantic graph includes nodes such as "Green Building Design," "Intelligent Energy Management," and "Renewable Energy Utilization," and establishes semantic associations between these nodes.

[0160] Subsequently, based on the presentation document modification prompt words, the semantic match between the presentation document modification prompt words and each semantic node in the joint semantic graph is calculated. This match is then filtered based on the semantic match to obtain a set of candidate keywords and their corresponding semantic weights. Furthermore, the candidate keyword set is filtered based on the semantic weights to obtain a target keyword set. Keywords in the target keyword set are then combined to generate a target prompt word, such as "Green Building Intelligent Energy Consumption Integrated Solution." Finally, based on the target prompt word, the modified presentation document is updated, adjusting its page content and structure to produce a target presentation document that meets the user's latest needs.

[0161] See also Figure 8 The present application also provides a device for generating a presentation document, which can implement the above-mentioned method for generating a presentation document. The device includes:

[0162] A presentation document generation module 801 is configured to generate a presentation document according to a presentation document generation prompt word to obtain an original presentation document;

[0163] The intention analysis module 802 is used to analyze the modification intention of the original presentation document according to the modified presentation document to determine the modification intention; wherein the modified presentation document is obtained by modifying the original presentation document;

[0164] A graph construction module 803 is used to construct a knowledge graph based on the modification intention and the prompt words generated in the presentation document to obtain a joint semantic graph;

[0165] The keyword screening module 804 is used to calculate the semantic matching degree between the presentation document modification prompt word and each semantic node in the joint semantic graph, and screen according to the semantic matching degree to obtain a candidate keyword set and the semantic weight of the candidate keyword;

[0166] The prompt word combination module 805 is used to screen the candidate keyword set according to the semantic weight to obtain the target keyword set, and combine the keywords in the target keyword set to generate the target prompt word;

[0167] The presentation document updating module 806 is configured to update the modified presentation document based on the target prompt word to obtain a target presentation document.

[0168] The specific implementation of the presentation document generating device is substantially the same as the specific embodiment of the presentation document generating method described above, and will not be described in detail herein.

[0169] The present application also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-mentioned method for generating a presentation document. The electronic device can be any smart terminal including a tablet computer, an in-vehicle computer, or the like.

[0170] See also Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0171] The processor 901 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0172] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program codes are stored in the memory 902 and are called by the processor 901 to execute the presentation document generation method of the embodiments of this application.

[0173] Input / output interface 903, used to implement information input and output;

[0174] Communication interface 904, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0175] Bus 905 , which transmits information between various components of the device (e.g., processor 901 , memory 902 , input / output interface 903 , and communication interface 904 );

[0176] The processor 901 , the memory 902 , the input / output interface 903 and the communication interface 904 are connected to each other in communication within the device via a bus 905 .

[0177] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned presentation document generation method is implemented.

[0178] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0179] The embodiments of the present application provide a presentation document generation method, a presentation document generation device, an electronic device, and a storage medium, which generate an original presentation document based on the presentation document generation prompt words, and then parse the modification intention of the original presentation document based on the modified presentation document, thereby determining the modification direction expected by the user; then, a knowledge graph is constructed by combining the modification intention and the generation prompt words to obtain a joint semantic graph that can reflect the semantic structure of the presentation document, and representative candidate keywords and their corresponding semantic weights are screened out through semantic matching calculation; on this basis, the candidate keywords are further screened and combined according to the semantic weights to generate target prompt words that are highly matched with the user's intentions; finally, the modified presentation document is updated based on the target prompt words to obtain a target presentation document with a complete semantic structure and accurate content expression. By combining semantic graph construction, keyword screening, and prompt word reconstruction, the present application effectively improves the semantic adaptability and content rationality in the process of automatic modification of presentation documents, and solves the problem of deviation between generated content and user intentions.

[0180] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0181] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0182] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0183] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0184] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0185] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0186] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0187] The units described above as separate components may or may not be physically separate, and 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 these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0188] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0189] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0190] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A method for generating a presentation document, characterized in that: The method comprises: Generate a presentation document according to the presentation document generation prompt words to obtain an original presentation document; Performing modification intention analysis on the original presentation document according to the modified presentation document to determine the modification intention; wherein the modified presentation document is obtained by modifying the original presentation document; A knowledge graph is constructed based on the modification intention and the prompt words generated in the presentation document to obtain a joint semantic graph; Calculating the semantic matching degree between the presentation document modification prompt word and each semantic node in the joint semantic graph, and screening according to the semantic matching degree to obtain a candidate keyword set and the semantic weight of the candidate keyword; Screening the candidate keyword set according to the semantic weight to obtain a target keyword set, and combining the keywords in the target keyword set to generate a target prompt word; The modified presentation document is updated based on the target prompt word to obtain a target presentation document.

2. The method according to claim 1, characterized in that The step of generating a presentation document according to the presentation document generation prompt word to obtain an original presentation document includes: Performing semantic analysis on the presentation document generation prompt words to obtain a presentation document content outline; wherein the presentation document content outline includes a parent title and a subtitle; Generate a cover page according to the parent title and the subtitle, and generate a directory page according to the parent title and the subtitle; Generate a text page according to the subtitle; The cover page, the directory page and the text page are spliced ​​together to form a presentation document to obtain the original presentation document.

3. The method according to claim 1, characterized in that The step of parsing the modification intention of the original presentation document according to the modified presentation document to determine the modification intention includes: Performing a presentation document comparison with the original presentation document according to the modified presentation document to determine a presentation document modification record; the presentation document modification record includes a text modification record, a graphic modification record, and a table modification record; Extracting keywords from the text modification record to obtain modification keywords; Performing operation classification on the graphic modification record to obtain an image modification type; Performing behavior recognition on the table modification record to obtain the table modification behavior; The modification intention is determined according to the modification keyword, the image modification type and the table modification behavior.

4. The method according to any one of claims 1 to 3, characterized in that The updating of the modified presentation document based on the target prompt word to obtain the target presentation document includes: updating the elements of the modified slideshow according to the target prompt word to obtain an updated presentation document; The updated presentation document is adjusted according to the target prompt word to obtain the target presentation document.

5. The method according to claim 4, characterized in that The step of updating elements of the modified presentation document according to the target prompt word to obtain an updated presentation document includes: By using an attention model, performing attention calculation on presentation document elements in the modified presentation document according to the target prompt word to obtain element contribution; Screening the presentation document elements in the modified presentation document according to the element contribution to obtain the original presentation document elements; The original presentation document elements in the modified presentation document are replaced according to the material presentation document elements to obtain the updated presentation document.

6. The method according to claim 5, characterized in that The replacing the original presentation document elements in the modified presentation document according to the material presentation document elements to obtain the updated presentation document includes: Performing semantic vectorization on the material presentation document elements to obtain a material presentation document vector, and performing semantic vectorization on the target prompt word to obtain a semantic prompt word vector; Calculating the similarity between the material presentation document vector and the semantic prompt word vector to obtain semantic similarity, and screening the preset presentation document material according to the semantic similarity to obtain a target presentation document element; The original presentation document element of the modified presentation document is updated according to the target presentation document element to obtain the updated presentation document.

7. The method according to claim 6, characterized in that The updating of the original presentation document element of the modified presentation document according to the target presentation document element to obtain the updated presentation document includes: Screening the target presentation document elements according to the types of the original presentation document elements to obtain selected presentation document elements; Calculating the similarity between the selected presentation document element and the original presentation document element, and screening the selected presentation document element based on the similarity to obtain a replacement presentation document element; The original presentation document element in the updated presentation document is replaced according to the replacement presentation document element to obtain the updated presentation document.

8. A presentation document generating device, characterized in that: The device comprises: A presentation document generation module is used to generate a presentation document according to the presentation document generation prompt words to obtain an original presentation document; an intention parsing module, configured to parse the original presentation document for modification intent based on the modified presentation document to determine the modification intent; wherein the modified presentation document is obtained by modifying the original presentation document; A graph construction module is used to construct a knowledge graph based on the modification intention and the prompt words generated in the presentation document to obtain a joint semantic graph; A keyword screening module is used to calculate the semantic matching degree between the presentation document modification prompt word and each semantic node in the joint semantic graph, and screen according to the semantic matching degree to obtain a set of candidate keywords and the semantic weights of the candidate keywords; a prompt word combination module, configured to screen the candidate keyword set according to the semantic weight to obtain a target keyword set, and combine the keywords in the target keyword set to generate a target prompt word; The presentation document updating module is used to update the modified presentation document based on the target prompt word to obtain a target presentation document.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the presentation document generation method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for generating a presentation document according to any one of claims 1 to 7 is implemented.

Images