Abstract generation method and device and electronic equipment
By chunking text and using the summary information of the previous text block to generate prompt information, the problems of poor readability and content omission in the prior art are solved, and efficient and relevant text summary generation is achieved.
Patent Information
- Application Number
- CN202311155412.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-07
- Publication Date
- 2025-03-14
AI Technical Summary
In the prior art, the text summary generated by the model is poorly readable or there are content omissions, making it difficult to effectively and quickly understand the core information of long text content quickly and efficiently.
By chunking the text to be processed, multiple text blocks are generated, and prompt information is generated based on the content information of the current text block and the summary information of the previous text block in the order of the text blocks, the summary generation model is input to output the summary information of the current text block, and finally the target summary is generated based on the summary information of the last text block.
Improve the relevance of the summary content with the original text content, avoid information omissions, ensure the consistency and focus of the summary, and enable users to quickly understand the important content of the text.
Smart Images
Figure CN119621957A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present disclosure relate to the field of computer and artificial intelligence technology, and in particular, to a summary generation method, device and electronic device. Background Art
[0002] As the pace of life accelerates, it is difficult for people to have a long time to read long text content. Therefore, generating summaries for text content helps people quickly and efficiently understand the core information of long text content.
[0003] At present, the methods for generating summaries for text content may include: manually generating and generating by models. The inventors have found that the summaries generated by the models are either poorly readable or have missing content. Summary of the invention
[0004] Embodiments of the present disclosure provide a method, device, and electronic device for generating a summary.
[0005] In a first aspect, an embodiment of the present disclosure provides a summary generation method, the method comprising: dividing a text to be processed into blocks to obtain a plurality of text blocks; performing the following summary generation operations on the plurality of text blocks in sequence according to the order of the text blocks in the text to be processed: generating prompt information based on content information of a current text block and summary information of a previous text block; inputting the prompt information into a summary generation model so that the summary generation model outputs summary information of the current text block; the method further comprises: generating a target summary of the text to be processed based on the summary information of the last text block, and outputting the target summary.
[0006] In a second aspect, an embodiment of the present disclosure provides a summary generation device, which includes: a block division unit, which is used to divide a text to be processed into blocks to obtain multiple text blocks; an execution unit, which is used to perform the following summary generation operations on the multiple text blocks in sequence according to the order of each text block in the text to be processed: generate prompt information based on content information of the current text block and summary information of the previous text block; input the prompt information into a summary generation model so that the summary generation model outputs summary information of the current text block; and an output unit, which is used to generate a target summary of the text to be processed based on the summary information of the last text block, and output the target summary.
[0007] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising: a processor and a memory; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the first aspect as well as various methods that may be involved in the first aspect.
[0008] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the first aspect and various possible methods involved in the first aspect are implemented.
[0009] In a fifth aspect, an embodiment of the present disclosure provides a computer program product, including a computer program, which, when executed by a processor, implements the first aspect as well as various methods that may be involved in the first aspect.
[0010] The summary generation method, device and electronic device provided in the present embodiment divide the text to be processed into blocks to obtain multiple text blocks; according to the order of each text block in the text to be processed, the following summary generation operations are performed on the multiple text blocks in sequence: prompt information is generated based on the content information of the current text block and the summary information of the previous text block; the prompt information is input into a summary generation model so that the summary generation model outputs the summary information of the current text block; the method also includes: generating a target summary of the text to be processed based on the summary information of the last text block, and outputting the target summary. Since the summary included in the prompt information corresponding to the last text block carries (or remembers) the summary information of all text blocks before the last text block, the generated target summary can improve the relevance of the summary content with the original text content. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0012] Figure 1 A schematic flowchart of a summary generation method provided by the present disclosure;
[0013] Figure 2 A schematic diagram of the principle flow of the summary generation method provided by the present disclosure;
[0014] Figure 3 for Figure 1 A schematic diagram of a specific implementation of step S102 in the illustrated embodiment;
[0015] Figure 4 A structural block diagram of a summary generation device provided by the present disclosure;
[0016] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION
[0017] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.
[0018] In one implementation scheme for generating a text summary provided by the present disclosure, the text is divided into blocks, and the block results are respectively input into a summary generation model, and the summary generation model generates a summary corresponding to each block result. Then, the summaries corresponding to each block are synthesized to obtain a summary of the text. This implementation scheme can obtain a summary of the text. However, due to the lack of connection between text blocks, this scheme makes the summaries of different blocks generated lack relevance, and even causes the problem of distorted details, which is not conducive to users reading the summary and quickly grasping the content of the text.
[0019] The solution provided by the present disclosure establishes a connection between different text blocks by inputting the information output by the previous text block as a part of the current text block, and outputting a summary of the current text block in combination with the information output by the previous text block and the current text block, so that the generated summary does not omit information and can highlight key points, which is conducive to users quickly grasping the important content of the text based on the summary.
[0020] Please refer to Figure 1 , which shows a schematic flow chart of the summary generation method provided by the present disclosure. Figure 1 As shown, the method includes:
[0021] S101: Divide the text to be processed into blocks to obtain multiple text blocks.
[0022] In this embodiment, the execution subject of the summary generation method can be a terminal device or a server. Specifically, the execution subject can run a summary generation model or be connected to an electronic device running a summary generation model.
[0023] The text to be processed may be the text to be processed input by the user. In one example, the user may input a link address corresponding to the text to be processed, and the execution subject may obtain the text to be processed from the link address. In another example, the user may directly upload a document recording the text to be processed.
[0024] The text to be processed may be a text recording various contents.
[0025] In some application scenarios, the above-mentioned text to be processed may be a meeting text recording the detailed content of the meeting. Generating a text summary of the meeting can help users quickly obtain the content of the meeting. In a scenario where the user does not attend the meeting, the meeting content can be quickly browsed by browsing the meeting summary to understand important information.
[0026] The above-mentioned text to be processed may be a text having a length greater than a preset length threshold, that is, a long text, for example, a text having a word count greater than a preset word count threshold, or a text having a character count greater than a preset character count threshold.
[0027] For the text to be processed, the above execution subject can divide it into blocks according to the block rules.
[0028] In some application scenarios, the text to be processed may be divided into blocks according to length to obtain multiple text blocks.
[0029] As an implementation method, the text to be processed can be divided into chunks according to the maximum length that the summary generation model can support.
[0030] In the field of natural language processing, tokens can be used as input units of language processing models. Tokens can be the smallest unit in a text. Schematically, for English text, a word or a punctuation mark can be a token. For Chinese text, characters or words can be used as tokens. A summary generation model can be a model that processes text recorded in a natural language to generate a summary. The maximum degree that a summary generation model can support can be, for example, the maximum number of tokens. Before inputting the text to be processed into the summary generation model, in order for the summary generation model to smoothly generate the corresponding summary, the text to be processed needs to be divided into blocks according to the maximum number of tokens. Each of the resulting text blocks can be processed by the summary generation model.
[0031] The multiple text blocks obtained by dividing the text to be processed according to the length can be smoothly received and processed by the summary generation model.
[0032] In some embodiments, the above step S101 may include: dividing the text to be processed into blocks according to semantics to obtain multiple text blocks.
[0033] In these embodiments, the execution subject may perform semantic analysis on the text to be processed, and treat the text contents with the same semantics as a text block, so as to obtain multiple text blocks.
[0034] Schematically, when the above text to be processed is a text recording the content of a meeting, the meeting discussed 3 topics. In the meeting, the speeches of different participants belonging to the same topic have semantic relevance. The meeting content belonging to the same topic can be determined based on the semantic relevance between different sentences. Continue to divide the above text into blocks according to semantics, and 3 text blocks can be obtained. Each text block can correspond to a topic.
[0035] By dividing the text to be processed into blocks according to semantics, the contents with semantic association can be divided into a text block. When generating a summary, for each text block, a semantically more complete summary can be generated based on the text block.
[0036] After the text to be processed is divided into blocks, multiple text blocks can be obtained. The multiple text blocks can be sorted according to the order in which they appear in the text to be processed. For example, paragraphs 01 to 10 of the text to be processed A are divided into text block 1, paragraphs 11 to 20 are divided into text block 2, and paragraphs 21 to 30 are divided into text block 3. The sorting results of the above three text blocks are: text block 1, text block 2 and text block 3.
[0037] S102: performing the following summary generation operations on multiple text blocks in sequence according to the order of the text blocks in the text to be processed:
[0038] Generate prompt information based on the content information of the current text block and the summary information of the previous text block;
[0039] The prompt information is input into the summary generation model so that the summary generation model outputs summary information of the current text block.
[0040] S103: Generate a target summary of the text to be processed based on the summary information of the last text block, and output the target summary.
[0041] The summary generation model may include various machine learning models, and the summary generation model may be a model for analyzing and processing natural language. The summary generation model may be various large-scale machine learning models.
[0042] After obtaining multiple text blocks in step S101, the order of the text blocks in the text to be processed can be determined. For example, the order of text blocks 1, 2 and 3 in the text to be processed A is: text block 1, text block 2 and text block 3.
[0043] Then, the summary generation operation is performed on the multiple text blocks in order from first to last.
[0044] For the first text block, prompt information may be generated according to the content information of the first text block.
[0045] For each text block after the first text block, prompt information may be generated according to the content information of the current text block and the summary information of the previous text block.
[0046] The prompt information of multiple text blocks can be sequentially input into the summary generation model starting from the first text block, and the summary generation model generates and outputs summary information of each text block according to the prompt information.
[0047] For each text block after the first text block, the prompt information of the text block may include summary information of the previous text block in addition to the content information of the text block.
[0048] In this way, the prompt information of each text block can contain the summary information of all previous text blocks. Therefore, the summary information of the text block generated by the summary generation model based on the prompt information including the content information of the text block and the summary information of the previous text block can cover all the text blocks before the text block and the content of the text block. The above-mentioned output target summary includes displaying the target summary, or sending the target summary to the terminal device for display.
[0049] In this embodiment, the text to be processed is divided into blocks to obtain multiple text blocks; according to the order of each text block in the text to be processed, the following summary generation operations are performed on the multiple text blocks in sequence: prompt information is generated based on the content information of the current text block and the summary information of the previous text block; the prompt information is input into the summary generation model so that the summary generation model outputs the summary information of the current text block; based on the summary information of the last text block, a target summary of the text to be processed is generated, and the target summary is output. Compared with the method of using the text content of each text block to generate a summary separately, and then splicing the summaries of each text block to generate a summary, in this scheme, since the summary information of the previous text block is referenced when generating the summary information of each text block, the summary information of the last text block output by the summary generation model covers the entire content of the text to be processed, and the generated summary will not miss any information, which can improve the relevance of the summary content with the original text content, and the coherence between different contents of the summary information generated by the above scheme is also good.
[0050] In some embodiments, the step S102 of generating prompt information based on the content information of the current text block and the summary information of the previous text block includes:
[0051] Generates hint information based on the content information of the current text block, the summary information of the previous text block, and one or more of the following:
[0052] The key information is determined from the content information of the previous text block, and the prediction information is determined from the summary information of the previous text block; wherein the prediction information includes information obtained by predicting the summary content of the next text block.
[0053] In some embodiments, the key information determined from the content information of the previous text block includes: text subject, to-do items, topic, key words / sentences in the content information, and information to be forgotten determined from the content information; wherein,
[0054] The key information also includes one or more corresponding inference information.
[0055] For a text block, after generating the summary information of the text block, the summary generation model will no longer obtain the content information of the text block. The summary information of the text block is processed from the text content of the original text block. Therefore, the summary information of the text block is different from the original content of the original text block. In order to establish the connection between the previous and next text blocks, that is, the connection between the contexts, it is necessary to determine the key information of the text block. And the key information of the previous text block is used as part of the prompt information of the current text block.
[0056] Take the case where the text to be processed is a meeting text that records the details of the meeting. The text blocks after the meeting text is divided into blocks can include themes, topics, and to-do items, which can be used as key information.
[0057] In addition, key information, such as keywords or key sentences, can also be extracted from the text content of the current text block. The method for extracting key information from the text content of the current text block can adopt a related method for extracting keywords or key sentences from text content.
[0058] For example, algorithms such as keyword extraction based on statistical features, keyword extraction based on word graph models, and keyword extraction based on topic models can be used to extract keywords from the text content of the current text block.
[0059] For example, an algorithm based on text similarity may be used to determine key sentences from the above text blocks.
[0060] These keywords and key sentences can be used as key information of the current text block.
[0061] In addition, in order to reduce the amount of calculation of the summary generation model, unimportant information in each text block can also be determined as the to-be-forgotten information of the content of the text block.
[0062] For example, text block B2 in the conference text segmentation result corresponds to topic T1. In this text block B2, text content C1 of T2 that is less related to topic T1 and is exchanged by participants may be included. These contents may be included in the current summary corresponding to text block B2. In order to prompt the summary generation model to ignore text content C1 when generating the summary corresponding to text block B3, the above-mentioned information to be forgotten (that is, text content C1) can be recorded in the key information corresponding to text block B2. When generating the summary corresponding to text block B3, the summary generation model will generate the current summary corresponding to text block B3 based on the text content of text block B3, the summary corresponding to B2, and the above-mentioned key information of B2.
[0063] In this way, when the summary generation model generates the summary information of the current text block, in addition to referring to the summary information of the previous text block and the content information of the current text block, it also refers to the key information of the original text of the previous text block, so that the generated summary information of the current text block has a better correlation with the previous text block. In addition, the above key information retains part of the original content of the previous text block. Therefore, when generating the summary information of the current text block, not only the summary information of the previous text block but also the key information of the previous text block are referred to. Therefore, the generated summary information of the current block can also retain part of the original content of the previous text block, which can improve the problem of distortion of the summary generated in the related scheme.
[0064] In some embodiments, the key information of each text block can be generated by the summary generation model according to the content information of the text block input. In these embodiments, the summary generation model also outputs the reasoning information corresponding to each key information. The reasoning information corresponding to the key information refers to the reason why the information successfully represents the key information of the text block.
[0065] In order to prevent the summary generation model from being fanciful and outputting information irrelevant to expectations, the summary generation model may be instructed to provide reasoning information for deriving the one or more key information when outputting the one or more key information.
[0066] In some embodiments, the prediction information includes one or more of the following:
[0067] To-do list guidance, topic guidance, and content summary guidance;
[0068] The prediction information also includes one or more corresponding inference information
[0069] As an implementation, the above prediction information may be generated by a summary generation model. Specifically, when the above summary generation model generates summary information of the current text block according to the input prompt information of the current text block, it may predict the summary content of the next text block according to the summary information of the current text block, thereby generating prediction information. In addition, the above summary generation model may also provide reasoning information corresponding to each prediction information.
[0070] Schematically, assuming that the text of the current text block B2 is a summary of Project 1, usually after the summary of Project 1, the implementation of the project and the conclusion of the implementation can be written. The summary generation model can predict the possible summary content (i.e., prediction information) of the next text block to be written based on the content of the current text block B2, such as the implementation method and conclusion of Project 1.
[0071] In order to prevent the summary generation model from being fanciful and outputting information irrelevant to expectations, the summary generation model may be instructed to provide reasoning information for obtaining the one or more prediction information when outputting the one or more prediction information.
[0072] The summary information of the previous text block, the content information of the current text block and the prediction information obtained from the previous text block are input into the summary generation model as the prompt information of the current text block to generate the summary information of the current text block, so that the generated summary information of the current text block has better coherence.
[0073] Please refer to Figure 2 , which shows a schematic diagram of the principle flow of the summary generation method provided by the present disclosure. After the text to be processed is divided into K text blocks, such as Figure 2 The text blocks 1, 2, ..., and K in the text block are text blocks 1, 2, ..., and 3, where K is an integer greater than or equal to 2. The above K text blocks can be input into the prompt information generation module 201 in sequence. Taking text block N as an example, N is an integer greater than 1 and less than or equal to K. The text content of text block N, the summary information, key information, and prediction information of text block N-1 output by the summary generation model 202 are fed back to the prompt information generation module 201, and the prompt information generation module 201 generates the prompt information N corresponding to the text block N based on the above text block N content, the key information, summary information, and prediction information of text block N-1. The prompt information N is input into the summary generation model 202, and the summary generation model 202 generates the summary information, key information, and prediction information of the text block N based on the summary information of the text block N. This process is repeated until the summary generation model 202 outputs the summary information of the text block K, and the summary information of the text block is used as the target summary corresponding to the text to be processed. By Figure 2The summary of the text to be processed generated by the process shown in the figure, because the summary included in the prompt information corresponding to the last text block K carries the summary information, key information and prediction information of all text blocks before the last text block, the generated target summary can improve the relevance of the summary content with the original text content, and the generated summary will not cause information omission. In addition, the generated summary highlights the key points, which is conducive to users quickly grasping the important content of the text according to the summary.
[0074] In some embodiments, the prompt information includes content information of the current text block, summary information of the previous text block, key information determined from the content information of the previous text block, and prediction information determined from the summary information of the previous text block.
[0075] The summary information of the previous text block, the content information of the current text block, the key information of the previous text block, and the key information of the predicted information obtained from the summary information of the previous text block and the predicted information are input as prompt information into the summary generation model, and the summary generation model generates the summary information of the current text block. In these embodiments, since any two preceding and following text blocks in the multiple text blocks of the text to be processed are connected through the key information and the predicted information, the target summary generated by the summary generation model will not produce detail distortion, and has a clear structure and highlights the key points, so that the user can quickly understand the key content of the text to be processed through the target summary.
[0076] In some embodiments, the prompt information further includes output prompt information corresponding to the current text block, wherein the output prompt information is used to instruct the summary generation model to output a summary, key information, and prediction information.
[0077] In these embodiments, outputting prompt information includes prompt information indicating to output the following contents: a summary of the current text block, key information of the current text block, and prediction information predicted based on the current text block, etc.
[0078] In these embodiments, setting output prompt information for the summary generation model can enable the model to better understand what output data needs to be generated, thereby making the output more consistent with user expectations.
[0079] Please refer to Figure 3 , which shows a schematic diagram of a specific implementation of inputting the prompt information and the current text block into the summary generation model in step S102.
[0080] like Figure 3 As shown, generating prompt information based on the content information of the current text block and the summary information of the previous text block in step S102 includes the following sub-steps:
[0081] For each text block, the content information of the text block and the summary information of the previous text block are filled into the pre-built prompt template to obtain prompt information.
[0082] In order to enable the summary generation model to predict the connection between the above text blocks, and in order to enable the summary model to use the above connection when generating summaries of different text blocks so that the generated summaries cover comprehensive content and have a clear structure and highlight the key points, a prompt template can be pre-built.
[0083] As an illustrative illustration, the prompt template may include the following contents: prompt information indicating input of content information of the current text block, prompt information indicating input of summary information of the previous text block, prompt information prompting input of key information of the previous text block, prompt information prompting input of predicted information of the previous text block; as well as output prompt information indicating generation of a summary of the current text block, output prompt information indicating generation of key information of the current text block, and output prompt information indicating generation of predicted information of the current text block.
[0084] As an illustrative illustration, the above prompt template may include the following content: the first text block is: (); the summary of the previous text block is (); the key information of the previous text block is (); the predicted information of the previous text block is (); the output summary is (), the key information is (), and the predicted information is ().
[0085] The above prompt template is merely an illustrative description and does not constitute a limitation on the prompt template.
[0086] By providing a prompt template, the summary generation model can be provided with the information of the previous text block and what information the user wants to output, so that the summary generation model can better process the summary information, key information, prediction information of the previous text block input above and the content information of the current text block according to the output specified by the prompt template, thereby quickly providing the output result that the user wants.
[0087] Corresponding to the above Figure 1 The summary generation method of the illustrated embodiment, Figure 4 This is a structural block diagram of a summary generation device provided by an embodiment of the present disclosure. For ease of explanation, only the parts related to the embodiment of the present disclosure are shown. Figure 4 The device 40 includes a block division unit 401, an execution unit 402 and an output unit 403.
[0088] in,
[0089] A block division unit 401 is used to divide the text to be processed into blocks to obtain multiple text blocks;
[0090] The execution unit 402 is used to perform the following summary generation operations on the multiple text blocks in sequence according to the order of the text blocks in the text to be processed:
[0091] Obtaining prompt information of the current text block and content information of the current text block, wherein the prompt information includes input prompt information; inputting the prompt information of the current text block and the content information of the current text block into a summary generation model, and the summary generation model outputs the current summary, key information of the current text block, and predicted information; using the summary of the current text block, the key information of the current text block, and the predicted information as input prompt information for the next text block; the key information includes key information related to the current text block; and the predicted information includes content information to be written of the predicted summary of the next text block;
[0092] The output unit 403 is used to generate a target summary of the text to be processed based on the summary information of the last text block, and output the target summary.
[0093] In one embodiment of the present disclosure, the execution unit 402 is further configured to generate prompt information based on content information of the current text block, summary information of the previous text block, and one or more of the following:
[0094] The key information determined from the content information of the previous text block and the predicted information determined from the summary information of the previous text block; wherein,
[0095] The prediction information includes information obtained by predicting the summary content of the next text block.
[0096] In one embodiment of the present disclosure, the block division unit 401 is further configured to:
[0097] Divide the text to be processed into blocks according to semantics to obtain multiple text blocks; or
[0098] The text to be processed is divided into blocks according to the length to obtain multiple text blocks.
[0099] In one embodiment of the present disclosure, the prompt information further includes output prompt information corresponding to the current text block, wherein the output prompt information is used to instruct the summary generation model to output a summary, key information of the current text block, and prediction information.
[0100] In one embodiment of the present disclosure, the execution unit 402 is further configured to: for each text block, fill the content information of the text block and the summary information of the previous text block into a pre-built prompt template to obtain prompt information.
[0101] In one embodiment of the present disclosure, the key information determined from the content information of the previous text block includes one or more of the following:
[0102] Text themes, to-do items, topics, keywords / sentences in content information, and information to be forgotten determined from content information.
[0103] In one embodiment of the present disclosure, the prediction information includes one or more of the following:
[0104] To-do list guidance, topic guidance, and content summary guidance;
[0105] The prediction information also includes one or more corresponding inference information.
[0106] In one embodiment of the present disclosure, the text to be processed includes text recording the content of a meeting.
[0107] In order to implement the above embodiment, the embodiment of the present disclosure also provides an electronic device.
[0108] refer to Figure 5 , which shows a schematic diagram of the structure of an electronic device 500 suitable for implementing the embodiment of the present disclosure, and the electronic device 500 may be a terminal device or a server. The terminal device may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, personal digital assistants (PDAs), tablet computers (Portable Android Devices, PADs), portable multimedia players (PMPs), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 5 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.
[0109] like Figure 5 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 to a random access memory (RAM) 503. Various programs and data required for the operation of the electronic device 500 are also stored in the RAM 503. The processing device 501, the ROM 502, and the RAM 503 are connected to each other via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.
[0110] Typically, the following devices may be connected to the I / O interface 505: input devices 506 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 507 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 508 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 509. The communication devices 509 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 5 The electronic device 500 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.
[0111] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.
[0112] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which (computer-executable instructions) may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.
[0113] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.
[0114] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device executes the method shown in the above embodiment.
[0115] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0116] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0117] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of the unit does not limit the unit itself in some cases. For example, the block unit may also be described as "a unit that blocks the text to be processed to obtain multiple text blocks."
[0118] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.
[0119] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0120] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) by each other to form a technical solution.
[0121] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.
[0122] Although the subject matter has been described in language specific to structural features and / or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims.
Claims
1. A method for generating a summary, comprising: Divide the text to be processed into blocks to obtain multiple text blocks; According to the order of the text blocks in the text to be processed, the following summary generation operations are performed on the multiple text blocks in sequence: Generate prompt information based on the content information of the current text block and the summary information of the previous text block; Inputting the prompt information into a summary generation model so that the summary generation model outputs summary information of the current text block; The method further includes: generating a target summary of the text to be processed based on summary information of the last text block, and outputting the target summary.
2. The method according to claim 1, characterized in that The generating of prompt information based on the content information of the current text block and the summary information of the previous text block includes: Generates hint information based on the content information of the current text block, the summary information of the previous text block, and one or more of the following: Key information determined from the content information of the previous text block, and prediction information determined from the summary information of the previous text block; wherein, The prediction information includes information obtained by predicting the summary content of the next text block.
3. The method according to claim 1, characterized in that: The text to be processed is divided into blocks to obtain multiple text blocks, including: Divide the text to be processed into blocks according to semantics to obtain multiple text blocks; or The text to be processed is divided into blocks according to length to obtain multiple text blocks.
4. The method according to claim 1, characterized in that: The prompt information also includes output prompt information corresponding to the current text block, wherein the output prompt information is used to indicate that the summary generation model outputs a summary, key information of the current text block, and prediction information.
5. The method according to any one of claims 1 to 4, characterized in that: The generating of prompt information based on the content information of the current text block and the summary information of the previous text block includes: For each text block, the content information of the text block and the summary information of the previous text block are filled into the pre-built prompt template to obtain the prompt information.
6. The method according to claim 2, characterized in that The key information determined from the content information of the previous text block includes one or more of the following: Text subject, to-do items, topics, keywords / sentences in the content information, and information to be forgotten determined from the content information.
7. The method according to claim 2, characterized in that The prediction information includes one or more of the following: To-do list guidance, topic guidance, and content summary guidance; The prediction information also includes the one or more corresponding reasoning information.
8. The method according to claim 1, characterized in that The text to be processed includes text recording the content of the meeting.
9. A summary generation device, comprising: A block division unit, used for dividing the text to be processed into blocks to obtain multiple text blocks; The execution unit is used to perform the following summary generation operations on the multiple text blocks in sequence according to the order of the text blocks in the text to be processed: Generate prompt information based on the content information of the current text block and the summary information of the previous text block; Inputting the prompt information into a summary generation model so that the summary generation model outputs summary information of the current text block; The output unit is used to generate a target summary of the text to be processed based on the summary information of the last text block, and output the target summary.
10. An electronic device, characterized in that: include: Processor and memory; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 8.
11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions. When a processor executes the computer-executable instructions, the method according to any one of claims 1 to 8 is implemented.