Article generation method and device, computer program product and electronic equipment

Abstract

The invention provides an article generation method, an article generation device, a computer program product and electronic equipment, and relates to the technical field of artificial intelligence. The article generation method comprises the steps of inputting writing demand information into an information collection agent, so that the information collection agent performs information collection based on the writing demand information to obtain a writing outline and a material matched with the writing demand information; and inputting the writing outline and the material matched with the writing demand information into an article generation agent, so that the article generation agent generates a first article based on the writing outline and the material matched with the writing demand information. The article is generated according to the writing outline and the material determined by the information collection agent, the richness, logicality and accuracy of the article content can be improved, and the automation and intelligence of article generation can be improved through multi-agent cooperation.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to an article generation method, an article generation device, a computer program product, and an electronic device. Background Art

[0002] With the development of artificial intelligence technology, various general big models and industry big models have emerged one after another, and big model-assisted article writing has also become a major research hotspot in industry and academia.

[0003] In the related art, a large model is used to directly generate article content based on prompt word requirements and user input information, such as article topics and article materials. However, articles generated in this way are usually highly random, lacking in content, and may also generate content with confusing logic and erroneous information.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0005] The present disclosure provides an article generation method, an article generation device, a computer program product and an electronic device, so as to at least to some extent solve the problem of poor article content richness, logic and accuracy in the related art.

[0006] According to a first aspect of the present disclosure, there is provided an article generation method, the method comprising: inputting writing requirement information into an information collection agent, so that the information collection agent collects information based on the writing requirement information to obtain a writing outline and materials matching the writing requirement information; inputting the writing outline and materials matching the writing requirement information into an article generation agent, so that the article generation agent generates a first article based on the writing outline and materials matching the writing requirement information.

[0007] According to a second aspect of the present disclosure, there is provided an article generating device, the device comprising: an outline and material generating module, for inputting writing requirement information into an information collecting intelligent agent, so that the information collecting intelligent agent collects information based on the writing requirement information, and obtains a writing outline and material matching the writing requirement information; an article generating module, for inputting a writing outline and material matching the writing requirement information into an article generating intelligent agent, so that the article generating intelligent agent generates a first article based on the writing outline and material matching the writing requirement information.

[0008] According to a third aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the article generation method of the first aspect and possible implementation methods thereof are implemented.

[0009] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the article generation method of the above-mentioned first aspect and its possible implementation method by executing the executable instructions.

[0010] The technical solution disclosed in this disclosure has the following beneficial effects:

[0011] In the above article generation process, the writing demand information is input into the information collection agent, so that the information collection agent collects information based on the writing demand information and obtains the writing outline and materials that match the writing demand information; the writing outline and materials that match the writing demand information are input into the article generation agent, so that the article generation agent generates the first article based on the writing outline and materials that match the writing demand information. On the one hand, generating articles based on the writing outline and materials determined by the information collection agent can improve the richness, logic and accuracy of the article content, and is suitable for application scenarios with high requirements for the structure and content quality of the article; on the other hand, through multi-agent collaboration, it can further reduce manual input and improve the automation and intelligence of article generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flowchart showing a method for generating an article in this exemplary embodiment is shown;

[0013] Figure 2 A schematic diagram showing an article generation system based on multi-agent collaboration in this exemplary embodiment;

[0014] Figure 3 A structural diagram of a multi-agent collaborative article generation system in this exemplary embodiment is shown;

[0015] Figure 4 A schematic diagram showing data flow of a multi-agent system that can resist disturbances in this exemplary embodiment;

[0016] Figure 5 A structural block diagram of an article generating device in this exemplary embodiment is shown;

[0017] Figure 6 An electronic device for implementing the above-mentioned article generating method in this exemplary embodiment is shown. DETAILED DESCRIPTION

[0018] Exemplary embodiments of the present disclosure will be described more fully hereinafter with reference to the accompanying drawings.

[0019] The accompanying drawings are schematic diagrams of the present disclosure and are not necessarily drawn to scale. Some of the block diagrams shown in the accompanying drawings may be functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or in hardware modules or integrated circuits, or in networks, processors or microcontrollers. The embodiments can be implemented in various forms and should not be construed as being limited to the examples set forth herein. The features, structures or characteristics described in the present disclosure may be combined in one or more embodiments in any suitable manner. In the description below, many specific details are provided to provide a full description of the embodiments of the present disclosure. However, those skilled in the art should appreciate that one or more specific details may be omitted when implementing the technical solution of the present disclosure, or one or more specific details may be replaced by other methods, components, devices, steps, etc.

[0020] In the related art, a large model is used to directly generate article content based on prompt word requirements and user input information, such as article topics and article materials. However, articles generated in this way are usually highly random, lacking in content, and may also generate content with confusing logic and erroneous information.

[0021] In view of the above problems, the exemplary embodiments of the present disclosure provide an article generation method, an article generation device, a computer program product and an electronic device, which can be applied to application scenarios with high requirements on the structure and content quality of articles, such as news manuscript generation.

[0022] It should be noted that an intelligent agent is an entity that has the ability to perceive, make decisions and act, can operate autonomously in a certain environment, and interact with other intelligent agents or the environment to achieve specific goals.

[0023] Intelligent agents usually have the following key characteristics:

[0024] 1) Autonomy: The ability to make decisions and perform actions autonomously without direct human intervention.

[0025] 2) Perception capability: the ability to obtain information about the environment through sensors or other means.

[0026] 3) Learning ability: Ability to learn from interactions with the environment and improve one’s behavior and decision-making strategies.

[0027] 4) Adaptability: Ability to adapt to changes in the environment and adjust one's behavior accordingly.

[0028] 5) Goal-oriented: Having clear goals and working hard to achieve them.

[0029] In an optional embodiment, reference Figure 1 As shown, a method for generating an article is provided, which specifically includes the following steps S110 to S120:

[0030] Step S110, inputting the writing requirement information into the information collection agent, so that the information collection agent collects information based on the writing requirement information, and obtains a writing outline and materials matching the writing requirement information;

[0031] Step S120, inputting the writing outline and materials matching the writing requirement information into the article generation intelligent agent, so that the article generation intelligent agent generates the first article based on the writing outline and materials matching the writing requirement information.

[0032] Figure 1 In the method shown, generating articles based on the writing outline and materials determined by the information collection agent can improve the richness, logic and accuracy of the article content, and is suitable for application scenarios with high requirements on the structure and content quality of the article; through multi-agent collaboration, manual input can be further reduced, and the automation and intelligence of article generation can be improved.

[0033] Below Figure 1 Provide detailed instructions for each step.

[0034] In step S110, the writing requirement information is input into the information collection agent, so that the information collection agent collects information based on the writing requirement information to obtain a writing outline and materials that match the writing requirement information.

[0035] The writing requirement information refers to the basic information required to generate an article, such as the type of manuscript, the subject of the manuscript, the original material of the manuscript, etc.

[0036] Optionally, writing requirement information may be obtained through user input operations.

[0037] Among them, the information collection agent can be composed of a large model, which can be responsible for collecting the necessary information required for article generation, such as writing outlines and materials.

[0038] For example, the information collection agent can have the ability of planning reflection, memory, and tool use. Among them, the memory ability can be used to store historical interaction information with the big model, knowledge in related writing fields, etc., to supplement the amount of information to assist the subsequent decision-making and actions of the big model. Among them, the tool use ability can be used to assist the big model to improve its capabilities and call tools, such as searching tools to obtain additional information;

[0039] Optionally, the information collection agent can plan the steps of generating writing outlines and materials, and execute the planned steps of generating writing outlines and materials step by step to obtain writing outlines and materials that match the writing requirement information. During the step-by-step execution process, information can be collected using database, web page search and other capabilities to obtain writing outlines and materials that match the writing requirement information.

[0040] Optionally, during the step-by-step execution process, constraint information corresponding to the writing field that matches the writing requirement information may also be queried, so that the generated writing outline and materials meet the corresponding constraint information.

[0041] In an optional implementation, the following steps can also be performed: inputting the writing outline and materials that match the writing requirement information into the evaluation agent, so that the evaluation agent evaluates the writing outline and materials that match the writing requirement information to obtain outline and material evaluation information; feeding back the outline and material evaluation information to the information collection agent, so that the information collection agent adjusts the outline and material collection strategy based on the outline and material evaluation information.

[0042] The evaluation dimension of the evaluation agent may include: information collection quality. The evaluation agent may evaluate the writing outline and materials generated by the information collection agent, obtain outline and material evaluation information, and provide feedback. The outline and material evaluation information may include but is not limited to evaluation information on the accuracy, completeness, matching degree, etc. of the writing outline and materials generated by the information collection agent.

[0043] Optionally, after the information collection agent obtains feedback information from the evaluation agent, i.e., the outline and material evaluation information, the information collection agent can adjust the outline and material collection strategy based on the feedback information to dynamically improve the information collection quality of the information collection agent. The outline and material collection strategy can, for example, plan the steps for generating a writing outline and materials.

[0044] In step S120, the writing outline and materials matching the writing requirement information are input into the article generation agent, so that the article generation agent generates a first article based on the writing outline and materials matching the writing requirement information.

[0045] Among them, the first article refers to the article obtained by the article generation agent based on the writing outline and materials generated by the information collection agent.

[0046] Among them, the article generation agent can be composed of a large model, which can generate articles according to the writing outline and materials, and can have the capabilities of planning reflection, memory, and tool use.

[0047] Exemplarily, the agent can be generated based on the article initialization of the basic large model. The basic large model can be a general large model that has not been trained in the corresponding field.

[0048] Optionally, the article generation agent may plan the steps of generating the article based on the writing outline and materials matching the writing requirement information, and execute the planned steps of generating the article step by step to obtain the first article. In the step-by-step execution process, intelligent capabilities such as memory ability and tool use ability may be used according to actual needs, and this disclosure does not specifically limit this.

[0049] In an optional embodiment, the following steps can also be performed: inputting the first article into an evaluation agent so that the evaluation agent evaluates the first article and obtains evaluation information corresponding to the first article; and feeding back the evaluation information corresponding to the first article to an article generation agent so that the article generation agent adjusts the article generation strategy based on the evaluation information corresponding to the first article.

[0050] The evaluation dimension of the evaluation agent may include: article quality. The evaluation agent may evaluate the first article generated by the article generation agent, obtain evaluation information corresponding to the first article, and provide feedback. The evaluation information corresponding to the first article may include but is not limited to evaluation information on correctness, semantics, and logic.

[0051] Optionally, after the article generation agent obtains feedback information from the evaluation agent, i.e., evaluation information corresponding to the first article, the article generation agent can adjust the article generation strategy of the article generation agent based on the feedback information to dynamically improve the article generation quality of the article generation agent. The article generation strategy can be, for example, the steps of planning to generate an article.

[0052] In an optional embodiment, the following steps can also be performed: inputting writing requirement information into the attacking intelligent agent so that the attacking intelligent agent generates an interference data set based on the writing requirement information, the interference data set including multiple sets of interference writing outlines and materials; based on the interference data set, adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent; wherein the interference writing outline and materials are used to attack the article generation intelligent agent.

[0053] Among them, the attacking intelligent agent can be composed of a large model, which can generate multiple sets of interference writing outlines and materials based on writing demand information.

[0054] Exemplarily, the groups of interference writing outlines and materials in the interference data set generated by the attacking agent can be used to attack the article generating agent, so as to make the article generating agent more robust and improve the quality of the article in the adversarial game between the attacking agent and the article generating agent.

[0055] Among them, interference writing outlines and materials may be writing outlines and materials with problems such as incorrect information, missing information, and mismatched information, which may specifically cover common problems in actual use.

[0056] For example, based on the writing requirement information and the interference data set, the interference writing outline and material generation strategy corresponding to the attacking intelligent agent can be adjusted to enhance the attacking capability of the attacking intelligent agent.

[0057] Optionally, the attacking agent can optimize its attacking capability by generating more writing outlines and materials that are sufficient to disturb the article generating agent.

[0058] In an optional embodiment, the above-mentioned adjustment of the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the interference data set can be achieved through the following steps: inputting each group of interference writing outlines and materials in the interference data set into the article generation intelligent agent respectively, so that the article generation intelligent agent generates multiple second articles based on each group of interference writing outlines and materials in the interference data set; inputting multiple second articles into the evaluation intelligent agent respectively, so that the evaluation intelligent agent evaluates the multiple second articles and obtains evaluation information corresponding to each second article; based on the evaluation information corresponding to each second article, adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0059] Among them, the second article refers to the article obtained by the article generation agent based on the interference writing outline and materials generated by the attacking agent.

[0060] After the second article is generated by the article generation agent, the second article can be input into the evaluation agent respectively, so that the evaluation agent evaluates the second article to obtain evaluation information corresponding to the second article.

[0061] The evaluation dimension of the evaluation agent may also include: attack capability. The evaluation agent may evaluate the second article generated by the article generation agent to obtain evaluation information corresponding to the second article. The evaluation information corresponding to the second article may include but is not limited to evaluation information on the effectiveness of the interference data set generated by the attack agent in attacking the article generation agent.

[0062] Optionally, after the attacking agent obtains feedback information from the evaluation agent, that is, the evaluation information corresponding to the second article, it can adjust the interference writing outline and material generation strategy corresponding to the attacking agent based on the feedback information, so as to dynamically improve the attack effectiveness of the attacking agent, so that the article generation agent can enhance its robustness through the game between itself and the attacking agent.

[0063] In an optional implementation, the above-mentioned adjustment of the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the evaluation information corresponding to each second article can be achieved through the following steps: based on the evaluation information corresponding to each second article, determine the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score; based on the writing demand information, the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score, adjust the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0064] Exemplarily, based on the evaluation information corresponding to the first article and the evaluation information corresponding to the second article, the writing outline and materials corresponding to the article with the highest evaluation score and the writing outline and materials corresponding to the article with the lowest evaluation score can be determined from the writing outline and materials corresponding to the first article and the second article, and a preference data pair can be generated; and the generated preference data pair can be used to adjust the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0065] Exemplarily, based on the evaluation information corresponding to the second article, the writing outline and materials corresponding to the article with the highest evaluation score and the writing outline and materials corresponding to the article with the lowest evaluation score can be determined from the writing outlines and materials corresponding to the second article to generate preference data pairs; and the generated preference data pairs can be used to adjust the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0066] The evaluation information corresponding to the article can be used as the evaluation criteria for the writing outline and materials, so as to determine the best writing outline and materials, the worst writing outline and materials, and obtain the preference data pair {y1, y2}, y1 can represent the best writing outline and materials, and y2 can represent the worst writing outline and materials. Among them, the best writing outline and materials can be the writing outline and materials corresponding to the article with the highest evaluation score; the worst writing outline and materials can be the writing outline and materials corresponding to the article with the lowest evaluation score. The writing outlines and materials corresponding to articles with different evaluation scores can correspond to different attack intensities.

[0067] For example, a data case in the preference alignment data set D can be formed based on the writing demand information, the writing outline and materials with the highest evaluation score, and the writing outline and materials with the lowest evaluation score, such as {x, y1, y2}, where x represents the writing demand information, y1 represents the writing outline and materials corresponding to the article with the highest evaluation score, and y2 represents the writing outline and materials corresponding to the article with the lowest evaluation score; further, the DPO (Direct Preference Optimization) algorithm can be used to align and optimize the large language model part of the attacking agent to adjust the interference writing outline and material generation strategy corresponding to the attacking agent, and the loss function is as follows:

[0068]

[0069] Among them, π θ and π ref They respectively represent the large model corresponding to the attacking agent and the reference model of the large model in the reinforcement learning process, σ represents the sigmoid activation function, and β is the weighted hyperparameter.

[0070] Optionally, the article generation agent may have the ability to defend against the interference writing outline and materials generated by the attacking agent. The article generation agent may supplement the ability to use related tools required for the defense capability as needed, and this disclosure does not specifically limit this.

[0071] For example, the article generation agent can identify the quality of the writing outline and materials, and take different response measures for writing outlines and materials of different qualities to ensure the quality of the generated articles and the robustness of the system.

[0072] Optionally, the article generation agent may adjust the article generation strategy of the article generation agent according to the evaluation information corresponding to the first article and the evaluation information corresponding to the second article, so that the article generation agent can take different response measures for writing outlines and material plans of different qualities, thereby ensuring the quality of the generated articles and ensuring the robustness of the system.

[0073] In an optional implementation, the evaluation strategy corresponding to the evaluation agent may also be adjusted according to the external environment of the system.

[0074] Among them, the external environment of the system refers to the external conditions and scenarios of the multi-agent collaborative system, which may involve human feedback, social environment, laws and regulations, technical environment and other aspects, and is used to prompt the system to adjust and optimize.

[0075] Specifically, the evaluation strategy corresponding to the evaluation agent can be adjusted according to the environmental data corresponding to the external environment of the system, such as environmental data related to the system's behavior, interaction, task completion status, etc.

[0076] Optionally, the evaluation agent can obtain environmental data of the system's external environment in real time, and transmit the environmental data back to the multi-agent system to promote the evolution of the agent.

[0077] Optionally, the evaluation agent can also evaluate the resource utilization, communication efficiency, strategy stability, etc. of other agents to encourage other agents to make improvements.

[0078] like Figure 2 As shown in the figure, a schematic diagram of generating articles based on a multi-agent collaborative article generation system is provided. Writing requirement information can be input into the multi-agent collaborative article generation system to obtain articles that match the writing requirements. The multi-agent collaborative article generation system can promote the evolution of agents through the system's external environment. Agents in the multi-agent collaborative article generation system can have the capabilities of planning, memory, and tool use.

[0079] Further, such as Figure 3 As shown, a structural diagram of a multi-agent collaborative article generation system is provided, which may include: information collection agent, article generation agent, attack agent, evaluation agent and other agents. Each agent may be composed of a large model, and may have the capabilities of planning reflection, memory and tool use. Among them, the information collection agent may generate a writing outline and materials according to the writing demand information input by the user; the article generation agent may generate articles according to the writing outline and materials; the attack agent may generate interference writing outlines and materials with information errors, missing information, mismatched information and other problems based on the writing demand information input by the user, which are used to attack the article generation agent, so as to prompt the article generation agent to have better robustness and article generation ability in the process of confronting the attack agent; the evaluation agent may generate evaluation information for other agents to prompt other agents to adjust and optimize, and may also optimize its own capabilities based on the external environment of the system to ensure that the feedback information is always effective.

[0080] Further, such as Figure 4As shown, a schematic diagram of data flow of a multi-agent system that can resist disturbance is provided. Writing demand information can be input into an information collection agent, so that the information collection agent collects information based on the writing demand information and obtains a writing outline and materials that match the writing demand information; the writing demand information can be input into an attacking agent, so that the attacking agent generates an interference data set based on the writing demand information, and the interference data set includes multiple groups of interference writing outlines and materials; the writing outline and materials that match the writing demand information are input into an article generation agent to generate a first article; the interference writing outline and materials are input into an article generation agent to generate a second article; the evaluation agent can evaluate the first article and the second article, and feed back the evaluation results to the article generation agent to adjust the article generation strategy of the article generation agent; the evaluation results can be used as the evaluation criteria for the writing outline and materials, and the preferred data pairs are determined from the writing outline and materials that match the writing demand information and the interference data set, and the attacking agent is optimized using the DPO optimization algorithm based on the preferred data pairs.

[0081] The above article generation method can quickly generate a large number of article drafts, reduce the workload of editors, and improve the efficiency of article writing; it can quickly provide users with a large amount of writing content to meet people's needs for rapid acquisition of information in different fields and topics; it can rely on the language processing capabilities and creativity of large models to explore new forms and styles of article writing to attract more readers; it can generate personalized articles based on user interests and preferences, and improve user attention and satisfaction with article content.

[0082] The exemplary embodiment of the present disclosure also provides an article generating device. Figure 5 As shown, the article generating device 500 may include the following program modules:

[0083] An outline and material generation module 510 is used to input writing requirement information into an information collection agent, so that the information collection agent collects information based on the writing requirement information to obtain a writing outline and material matching the writing requirement information;

[0084] The article generation module 520 is used to input the writing outline and materials matching the writing requirement information into the article generation intelligent body, so that the article generation intelligent body generates the first article based on the writing outline and materials matching the writing requirement information.

[0085] In an optional embodiment, based on the aforementioned scheme, the article generation device 500 may include: an outline and material evaluation module, which is used to input the writing outline and materials that match the writing requirement information into the evaluation agent, so that the evaluation agent evaluates the writing outline and materials that match the writing requirement information to obtain outline and material evaluation information; and feeds back the outline and material evaluation information to the information collection agent, so that the information collection agent adjusts the outline and material collection strategy based on the outline and material evaluation information.

[0086] In an optional embodiment, based on the aforementioned scheme, the article generation device 500 may include: a first article evaluation module, used to input the first article into an evaluation agent, so that the evaluation agent evaluates the first article and obtains evaluation information corresponding to the first article; and feeds back the evaluation information corresponding to the first article to the article generation agent, so that the article generation agent adjusts the article generation strategy based on the evaluation information corresponding to the first article.

[0087] In an optional embodiment, based on the aforementioned scheme, the article generation device 500 may include: an interference data set generation module, used to input the writing requirement information into the attacking intelligent agent, so that the attacking intelligent agent generates an interference data set based on the writing requirement information, and the interference data set includes multiple groups of interference writing outlines and materials; an interference writing outline and material generation strategy adjustment module, used to adjust the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the interference data set; wherein the interference writing outline and material are used to attack the article generation intelligent agent.

[0088] In an optional embodiment, based on the aforementioned scheme, the interference writing outline and material generation strategy adjustment module can be configured as follows: input each group of interference writing outlines and materials in the interference data set into the article generation agent respectively, so that the article generation agent generates multiple second articles based on each group of interference writing outlines and materials in the interference data set; input the multiple second articles into the evaluation agent respectively, so that the evaluation agent evaluates the multiple second articles and obtains the evaluation information corresponding to each second article; based on the evaluation information corresponding to each second article, adjust the interference writing outline and material generation strategy corresponding to the attack agent.

[0089] In an optional embodiment, based on the aforementioned scheme, the above-mentioned adjustment of the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the evaluation information corresponding to each second article can be configured as follows: based on the evaluation information corresponding to each second article, determine the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score; based on the writing demand information, the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score, adjust the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0090] In an optional implementation, based on the above solution, the article generation device 500 may include: an evaluation strategy adjustment module, which is used to adjust the evaluation strategy corresponding to the evaluation agent according to the external environment of the system.

[0091] The specific details of each part of the above-mentioned article generation device 500 have been described in detail in the implementation method of the method part. The undisclosed details can be found in the implementation method of the method part, so they will not be repeated here.

[0092] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the exemplary embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0093] The exemplary embodiments of the present disclosure also provide a computer program product, which includes a computer program, and when the computer program is executed by a processor, the above-mentioned article generation method is implemented.

[0094] In one embodiment, the computer program product may be a tangible product containing a computer program, such as a computer-readable storage medium storing a computer program. The readable storage medium may be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory (Flash), mechanical hard disk (HDD), solid-state drive (SSD), and the like. Exemplarily, the computer program product may be implemented as a non-volatile storage medium storing a computer program, such as a read-only memory, a NAND flash memory, and the like.

[0095] In one embodiment, the computer program product may be an intangible product including a computer program. Exemplarily, the computer program product may be implemented as a virtual digital product, such as a digital file storing an executable file, an installation package, etc. of the computer program.

[0096] The code of the computer program can be written in one or more programming languages. Programming languages ​​such as C language, Java, C++, etc. The program code can be executed entirely on the user computing device, or partially on the user computing device, or as a separate software package, or partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, such as a local area network (LAN), a wide area network (WAN), etc., or can be connected to an external computing device (e.g., an Internet connection provided by an operator).

[0097] The computer program may be carried or transmitted via electrical, magnetic, optical, electromagnetic, infrared, or other signals. The electronic device may convert the signal carrying the computer program into a digital signal, thereby running the computer program. When the computer program is run on an electronic device, its code is used to enable the electronic device to execute (more specifically, the processor of the electronic device may execute) the method steps of various exemplary embodiments of the present disclosure, such as the above method, which may include the following steps:

[0098] Inputting the writing requirement information into the information collection agent, so that the information collection agent collects information based on the writing requirement information to obtain a writing outline and materials matching the writing requirement information;

[0099] The writing outline and materials matching the writing requirement information are input into the article generation intelligent agent, so that the article generation intelligent agent generates the first article based on the writing outline and materials matching the writing requirement information.

[0100] In an optional implementation, based on the aforementioned scheme, the following steps may also be performed: inputting the writing outline and materials that match the writing requirement information into the evaluation agent, so that the evaluation agent evaluates the writing outline and materials that match the writing requirement information to obtain outline and material evaluation information; feeding back the outline and material evaluation information to the information collection agent, so that the information collection agent adjusts the outline and material collection strategy based on the outline and material evaluation information.

[0101] In an optional implementation, based on the aforementioned scheme, the following steps may also be performed: inputting the first article into an evaluation agent so that the evaluation agent evaluates the first article and obtains evaluation information corresponding to the first article; and feeding back the evaluation information corresponding to the first article to an article generation agent so that the article generation agent adjusts the article generation strategy based on the evaluation information corresponding to the first article.

[0102] In an optional embodiment, based on the aforementioned scheme, the following steps may also be performed: inputting writing requirement information into the attacking intelligent agent so that the attacking intelligent agent generates an interference data set based on the writing requirement information, the interference data set including multiple sets of interference writing outlines and materials; based on the interference data set, adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent; wherein the interference writing outline and materials are used to attack the article generation intelligent agent.

[0103] In an optional embodiment, based on the aforementioned scheme, the above-mentioned adjustment of the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the interference data set can be achieved through the following steps: inputting each group of interference writing outlines and materials in the interference data set into the article generation intelligent agent respectively, so that the article generation intelligent agent generates multiple second articles based on each group of interference writing outlines and materials in the interference data set; inputting multiple second articles into the evaluation intelligent agent respectively, so that the evaluation intelligent agent evaluates the multiple second articles and obtains evaluation information corresponding to each second article; based on the evaluation information corresponding to each second article, adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0104] In an optional embodiment, based on the aforementioned scheme, the above-mentioned adjustment of the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the evaluation information corresponding to each second article can be achieved through the following steps: based on the evaluation information corresponding to each second article, determine the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score; based on the writing demand information, the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score, adjust the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0105] In an optional implementation, based on the above solution, the following steps may also be performed: adjusting the evaluation strategy corresponding to the evaluation agent according to the external environment of the system.

[0106] In the above article generation process, generating articles according to the writing outline and materials determined by the information collection agent can improve the richness, logic and accuracy of the article content, which is suitable for application scenarios with high requirements on the structure and content quality of the article; through multi-agent collaboration, manual input can be further reduced, and the automation and intelligence of article generation can be improved.

[0107] The exemplary embodiment of the present disclosure also provides an electronic device capable of implementing the above-mentioned article generation method. The electronic device may include a processor and a memory. The memory stores executable instructions of the processor, such as program code. The processor executes the method in the exemplary embodiment by executing the executable instructions. In addition, the electronic device may also include a display for displaying a graphical user interface.

[0108] Reference below Figure 6 , the electronic device is exemplarily described in the form of a general-purpose computing device. It should be understood that Figure 6 The electronic device 600 shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0109] like Figure 6 As shown, the electronic device 600 may include: a processor 610 , a memory 620 , a bus 630 , an I / O (input / output) interface 640 , a network adapter 650 , and a display 660 .

[0110] The memory 620 may include a volatile memory, such as a RAM 621, a cache unit 622, and may also include a non-volatile memory, such as a ROM 623. The memory 620 may also include one or more program modules 624, such program modules 624 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or a combination thereof may include the implementation of a network environment. For example, the program module 624 may include each module in the above-mentioned device.

[0111] The processor 610 may include one or more processing units. For example, the processor 610 may include an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor and / or an NPU (Neural-Network Processing Unit), etc.

[0112] The processor 610 may be configured to execute executable instructions stored in the memory 620 , such as executing any one or more method steps in this exemplary embodiment.

[0113] Exemplarily, the processor 610 may perform the following steps:

[0114] Inputting the writing requirement information into the information collection agent, so that the information collection agent collects information based on the writing requirement information to obtain a writing outline and materials matching the writing requirement information;

[0115] The writing outline and materials matching the writing requirement information are input into the article generation intelligent agent, so that the article generation intelligent agent generates the first article based on the writing outline and materials matching the writing requirement information.

[0116] In an optional implementation, based on the aforementioned scheme, the following steps may also be performed: inputting the writing outline and materials that match the writing requirement information into the evaluation agent, so that the evaluation agent evaluates the writing outline and materials that match the writing requirement information to obtain outline and material evaluation information; feeding back the outline and material evaluation information to the information collection agent, so that the information collection agent adjusts the outline and material collection strategy based on the outline and material evaluation information.

[0117] In an optional implementation, based on the aforementioned scheme, the following steps may also be performed: inputting the first article into an evaluation agent so that the evaluation agent evaluates the first article and obtains evaluation information corresponding to the first article; and feeding back the evaluation information corresponding to the first article to an article generation agent so that the article generation agent adjusts the article generation strategy based on the evaluation information corresponding to the first article.

[0118] In an optional embodiment, based on the aforementioned scheme, the following steps may also be performed: inputting writing requirement information into the attacking intelligent agent so that the attacking intelligent agent generates an interference data set based on the writing requirement information, the interference data set including multiple sets of interference writing outlines and materials; based on the interference data set, adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent; wherein the interference writing outline and materials are used to attack the article generation intelligent agent.

[0119] In an optional embodiment, based on the aforementioned scheme, the above-mentioned adjustment of the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the interference data set can be achieved through the following steps: inputting each group of interference writing outlines and materials in the interference data set into the article generation intelligent agent respectively, so that the article generation intelligent agent generates multiple second articles based on each group of interference writing outlines and materials in the interference data set; inputting multiple second articles into the evaluation intelligent agent respectively, so that the evaluation intelligent agent evaluates the multiple second articles and obtains evaluation information corresponding to each second article; based on the evaluation information corresponding to each second article, adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0120] In an optional embodiment, based on the aforementioned scheme, the above-mentioned adjustment of the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the evaluation information corresponding to each second article can be achieved through the following steps: based on the evaluation information corresponding to each second article, determine the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score; based on the writing demand information, the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score, adjust the interference writing outline and material generation strategy corresponding to the attacking intelligent agent.

[0121] In an optional implementation, based on the above solution, the following steps may also be performed: adjusting the evaluation strategy corresponding to the evaluation agent according to the external environment of the system.

[0122] In the above article generation process, generating articles according to the writing outline and materials determined by the information collection agent can improve the richness, logic and accuracy of the article content, which is suitable for application scenarios with high requirements on the structure and content quality of the article; through multi-agent collaboration, manual input can be further reduced, and the automation and intelligence of article generation can be improved.

[0123] The bus 630 is used to realize the connection between different components of the electronic device 600, and may include a data bus, an address bus, and a control bus.

[0124] The electronic device 600 can communicate with one or more external devices 700 (eg, a keyboard, a mouse, an external controller, etc.) through the I / O interface 640 .

[0125] The electronic device 600 can communicate with one or more networks through the network adapter 650. For example, the network adapter 650 can provide mobile communication solutions such as 3G / 4G / 5G, or provide wireless communication solutions such as wireless LAN, Bluetooth, near field communication, etc. The network adapter 650 can communicate with other modules of the electronic device 600 through the bus 630.

[0126] The electronic device 600 may display a graphical user interface and the like through the display 660 .

[0127] although Figure 6 Not shown, other hardware and / or software modules may also be provided in the electronic device 600, including but not limited to: a display, a microcode, a device driver, a redundant processor, an external disk drive array, a RAID (Redundant Arrays of Independent Disks) system, a tape drive, and a data backup storage system.

[0128] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the exemplary embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0129] It will be appreciated by those skilled in the art that various aspects of the present disclosure may be implemented as a system, method or program product. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which may be collectively referred to herein as a "circuit", "module" or "system". Those skilled in the art will readily think of other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any variations, uses or adaptive changes of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and implementation are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

[0130] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for generating an article, characterized in that: The method comprises: Inputting writing requirement information into an information collection agent, so that the information collection agent collects information based on the writing requirement information to obtain a writing outline and materials matching the writing requirement information; The writing outline and materials matching the writing requirement information are input into the article generation intelligent agent, so that the article generation intelligent agent generates a first article based on the writing outline and materials matching the writing requirement information.

2. The method according to claim 1, characterized in that The method further comprises: Inputting the writing outline and materials matching the writing requirement information into the evaluation intelligent body, so that the evaluation intelligent body evaluates the writing outline and materials matching the writing requirement information to obtain outline and material evaluation information; The outline and material evaluation information are fed back to the information collection agent so that the information collection agent adjusts the outline and material collection strategy based on the outline and material evaluation information.

3. The method according to claim 1, characterized in that The method further comprises: Inputting the first article into an evaluation agent so that the evaluation agent evaluates the first article and obtains evaluation information corresponding to the first article; The evaluation information corresponding to the first article is fed back to the article generation agent, so that the article generation agent adjusts the article generation strategy based on the evaluation information corresponding to the first article.

4. The method according to claim 1, characterized in that The method further comprises: Inputting the writing requirement information into an attacking agent, so that the attacking agent generates an interference data set based on the writing requirement information, wherein the interference data set includes multiple groups of interference writing outlines and materials; Based on the interference data set, adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent; Among them, the interference writing outline and materials are used to attack the article generation agent.

5. The method according to claim 4, characterized in that The step of adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the interference data set includes: Inputting each group of interference writing outlines and materials in the interference data set into the article generation agent respectively, so that the article generation agent generates a plurality of second articles based on each group of interference writing outlines and materials in the interference data set; Inputting the plurality of second articles into the evaluation agent respectively, so that the evaluation agent evaluates the plurality of second articles to obtain evaluation information corresponding to each of the second articles; Based on the evaluation information corresponding to each of the second articles, the interference writing outline and material generation strategy corresponding to the attacking intelligent agent are adjusted.

6. The method according to claim 5, characterized in that The step of adjusting the interference writing outline and material generation strategy corresponding to the attacking intelligent agent based on the evaluation information corresponding to each of the second articles includes: Based on the evaluation information corresponding to each of the second articles, determining a writing outline and materials corresponding to the article with the highest evaluation score and a writing outline and materials corresponding to the article with the lowest evaluation score; Based on the writing demand information, the writing outline and materials corresponding to the article with the highest evaluation score, and the writing outline and materials corresponding to the article with the lowest evaluation score, the interference writing outline and material generation strategy corresponding to the attacking intelligent agent is adjusted.

7. The method according to any one of claims 2, 3, 5 and 6, characterized in that: The method further comprises: According to the external environment of the system, the evaluation strategy corresponding to the evaluation agent is adjusted.

8. An article generating device, characterized in that: The device comprises: An outline and material generation module, used to input writing demand information into an information collection agent, so that the information collection agent collects information based on the writing demand information to obtain a writing outline and materials matching the writing demand information; The article generation module is used to input the writing outline and materials matching the writing requirement information into the article generation intelligent body, so that the article generation intelligent body generates the first article based on the writing outline and materials matching the writing requirement information.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 7 by executing the executable instructions.

Images