An information processing method and device, electronic equipment and storage medium
By co-processing generative models and target plugins, the problem that artificial intelligence systems cannot answer multiple types of questions is solved, and more accurate and extensive information processing capabilities are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING ZITIAO NETWORK TECH CO LTD
- Filing Date
- 2023-08-08
- Publication Date
- 2026-06-09
AI Technical Summary
Existing artificial intelligence systems typically only address a single type of question and cannot effectively answer multiple types of questions, resulting in inaccurate answers.
By employing a collaborative approach between generative models and target plugins, the system receives input information, identifies the target intent category, invokes the matching target plugin, and generates the final answer to meet the needs of different types of questions.
It improves the accuracy and applicability of problem answers, and enhances the overall problem-solving capabilities of artificial intelligence systems.
Smart Images
Figure CN117076629B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of computer technology, and more specifically, to an information processing method, apparatus, electronic device, and storage medium. Background Technology
[0002] With the development of computer technology and artificial intelligence, users can perform tasks such as searching and image drawing. However, currently, artificial intelligence can usually only address single types of problems. How to obtain more accurate results for different types of problems is an urgent problem to be solved. Summary of the Invention
[0003] This disclosure provides at least one information processing method, apparatus, electronic device, and storage medium.
[0004] In a first aspect, embodiments of this disclosure provide an information processing method, including:
[0005] Receive input information to be processed;
[0006] Obtain the target answer result for the information to be processed. The target answer result is generated based on a generative model and a target plugin. The collaborative method between the generative model and the target plugin in generating the target answer result is related to the capability requirements of the target plugin. The target plugin is a plugin that matches the information to be processed and is able to answer the information to be processed.
[0007] Display the target response results.
[0008] Secondly, embodiments of this disclosure also provide an information processing apparatus, comprising:
[0009] The input module is used to receive input information to be processed.
[0010] The acquisition module is used to acquire the target answer result of the information to be processed. The target answer result is generated based on the generative model and the target plugin. The collaborative method of the generative model and the target plugin in generating the target answer result is related to the capability implementation requirements of the target plugin. The target plugin is a plugin that matches the information to be processed and can answer the information to be processed.
[0011] The display module is used to display the target answer results.
[0012] Thirdly, an optional implementation of this disclosure also provides an electronic device, including a processor and a memory, wherein the memory stores machine-readable instructions executable by the processor, and the processor is configured to execute the machine-readable instructions stored in the memory. When the machine-readable instructions are executed by the processor, the processor performs the steps of the first aspect above, or any possible implementation of the first aspect.
[0013] Fourthly, an optional implementation of this disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the first aspect above, or any possible implementation of the first aspect.
[0014] In this embodiment, input information to be processed is received; a target answer result for the information to be processed is obtained, wherein the target answer result is generated based on a generative model and a target plugin, and the collaborative method by which the generative model and the target plugin generate the target answer result is related to the capability requirements of the target plugin, wherein the target plugin is a plugin that matches the information to be processed and can answer the information to be processed; and the target answer result is displayed. In this way, the generative model can call the target plugin that matches the information to be processed and can answer it, thereby collaboratively interacting to generate the final target answer result based on a collaborative method. This can effectively answer questions in different vertical scenarios, improve accuracy, enhance problem-solving capabilities and applicability, and improve performance.
[0015] For a description of the effects of the aforementioned information processing device, electronic device, and computer-readable storage medium, please refer to the description of the aforementioned information processing method; it will not be repeated here.
[0016] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of this disclosure.
[0017] To make the above-mentioned objects, features and advantages of this disclosure more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description
[0018] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.
[0019] Figure 1 A flowchart of an information processing method provided by an embodiment of this disclosure is shown;
[0020] Figure 2 A flowchart of another information processing method provided by an embodiment of this disclosure is shown;
[0021] Figure 3 A flowchart corresponding to the search plugin in the information processing method provided in this embodiment of the disclosure is shown;
[0022] Figure 4 A flowchart of another information processing method provided by an embodiment of this disclosure is shown;
[0023] Figure 5 A schematic diagram of an information processing apparatus provided in an embodiment of this disclosure is shown;
[0024] Figure 6 A schematic diagram of an electronic device provided in an embodiment of the present disclosure is shown. Detailed Implementation
[0025] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.
[0026] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. The components of the embodiments of this disclosure described and shown herein can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.
[0027] Research has found that with the development of computer technology and artificial intelligence, functions such as content creation and image drawing can be realized based on artificial intelligence. However, users' needs are usually multifaceted, while artificial intelligence can usually only answer single types of questions. Without the ability to solve a certain type of question, it cannot provide an answer.
[0028] Based on the above research, this disclosure provides an information processing method that receives input information to be processed, and then, based on a generative model, calls a target plugin that can answer the information to be processed. According to the collaborative method of the generative model and the target plugin, a target answer result for the information to be processed is generated and displayed. In this way, for the input information to be processed, the generative model can call the corresponding matching and answerable target plugin, thereby collaboratively processing to generate the final target answer result. This method can effectively and accurately answer different types of questions, improving problem-solving capabilities and applicable scenarios.
[0029] The shortcomings of the above solutions are the result of the inventor's practical experience and careful research. Therefore, the discovery process of the above problems and the solutions proposed in this disclosure below should be considered as the inventor's contribution to this disclosure.
[0030] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0031] To facilitate understanding of this embodiment, a detailed description of the information processing method disclosed in this disclosure is provided first. The executing entity of the information processing method provided in this disclosure is generally an electronic device with a certain computing capability. This electronic device may include, for example, a terminal device, a server, or other processing devices. The terminal device can be a user equipment (UE), mobile device, cellular phone, cordless phone, personal digital assistant (PDA), handheld device, computing device, in-vehicle device, wearable device, etc. A personal digital assistant is a handheld electronic device that possesses some functions of a computer. It can be used to manage personal information, browse the internet, send and receive emails, etc., and generally does not have a keyboard; it can also be called a handheld computer. In some possible implementations, this information processing method can be implemented by a processor calling computer-readable instructions stored in memory.
[0032] The following describes the information processing method provided in the embodiments of this disclosure, taking the terminal device as the executing entity as an example.
[0033] See Figure 1 The diagram shown is a flowchart of an information processing method provided in an embodiment of this disclosure. The method includes:
[0034] S101: Receive the input information to be processed.
[0035] In this embodiment of the disclosure, based on artificial intelligence, an artificial intelligence (AI) tool can be provided to the user. Using the AI tool, the user can input any information to be processed in the form of a question on the AI tool's page.
[0036] S102: Obtain the target answer result for the information to be processed. The target answer result is generated based on the generative model and the target plugin. The collaborative method between the generative model and the target plugin in generating the target answer result is related to the capability requirements of the target plugin. The target plugin is a plugin that matches the information to be processed and can answer the questions about the information to be processed.
[0037] In this embodiment, the generative model can be understood as a basic large language model, which can solve many problems in vertical scenarios. However, the capabilities of a single generative model may be limited. For example, some real-time search questions, such as "What's the weather like today?", require real-time search to answer. Therefore, in this embodiment, based on the generative model, a corresponding matching and answerable target plugin can be called. Then, the generative model and the target plugin cooperate and process together to generate the final target answer result, which can meet the needs of different questions and improve the accuracy of the answer.
[0038] Furthermore, in this embodiment of the disclosure, a possible implementation is also provided: in response to the fact that the information to be processed does not match the target plugin, and the generative model is able to answer the information to be processed, semantic analysis is performed on the information to be processed based on the generative model to generate the target answer result for the information to be processed.
[0039] In this embodiment of the disclosure, if there is no matching target plugin and the generative model itself can provide the answer, the corresponding target answer result can be obtained directly based on the generative model.
[0040] S103: Display the target answer results.
[0041] In this embodiment, input information to be processed is received. Based on a generative model and a target plugin, the generative model and the target plugin work together to generate a target answer result, which is then displayed. This allows for accurate identification and invocation of the corresponding matching target plugin for any user-input information to be processed. Furthermore, based on a collaboration method related to the capability requirements of the target plugin, the generative model and the target plugin work together to generate the target answer result. This can address the question requirements of different vertical scenarios. Different collaboration methods can be used for different target plugins, or the same collaboration method can be used that meets the capability requirements of the target plugin. This improves the cooperation effect between the target generative model and the target plugin, further enhancing the accuracy of the target answer result.
[0042] In this embodiment of the disclosure, when calling the corresponding target plugin based on the generative model, it is necessary to accurately identify the matching target plugin, and then generate the target answer result based on the generative model and the target plugin. Specifically, this disclosure also provides several possible implementation methods for generating the target answer result:
[0043] In one possible implementation, the target response is determined in the following way:
[0044] S1. Perform intent recognition on the information to be processed to determine the target intent category of the information to be processed; whereby the target intent category represents the capability requirement to answer the information to be processed.
[0045] In this embodiment of the disclosure, intent recognition can be implemented based on a generative model. Specifically, this disclosure provides an embodiment that includes: based on a generative model, taking the information to be processed and intent judgment prompt statements as input, performing semantic analysis on the words included in the information to be processed according to the intent judgment prompt statements, and determining the target intent category matched by the information to be processed; wherein, the intent judgment prompt statements are used to indicate the capability requirement judgment requirements of each intent category and the word examples representing the corresponding intent category.
[0046] In this embodiment of the disclosure, the intent judgment prompt statement can be understood as an instruction for the learning and operation of the generative model. It can make multiple requirements for the generative model and can also input some reference examples. By utilizing the imitation and learning capabilities of the generative model, the generative model can determine what kind of information to be processed and what kind of intent category it belongs to.
[0047] For example, the intent-based prompts could include: "You can use different plugins to answer questions." When there's a time-sensitive requirement to answer a question, it's identified as a search intent, and a search plugin is used. Examples of search intent terms could include "today's weather," "news," "stocks," or "today's hot topics." When the input question contains image descriptions, it's identified as a drawing intent, and a drawing plugin is used. Examples of drawing intent terms could include "draw one" or "draw a picture." Additionally, when you can answer the question without plugins, the answer can be generated directly. Examples of terms could include "write an essay."
[0048] In addition, the intent-based prompt statement may include other requirements. For example, when generating the answer, the context of the input information to be processed may be considered. Or, language requirements may be provided. In this embodiment, no specific limitations are imposed.
[0049] S2. Based on the target intent category and the association between the intent category and the plugin, determine the target plugin that matches the target intent category. Different plugins have different capabilities.
[0050] For example, the drawing intent is associated with a drawing plugin, which can be a text-based image model. Furthermore, the drawing vertical scenarios can be further subdivided, such as including photographic drawing, cartoon drawing, etc., thereby matching drawing plugins for more specific vertical scenarios and improving the accuracy of the drawing results.
[0051] For example, the search intent may be associated with a search plugin, such as a search engine, which can be invoked to perform a real-time search.
[0052] S3. Based on the generative model and the target plugin, generate the target answer result in a collaborative manner.
[0053] Thus, in this embodiment of the disclosure, for example in the main dialogue process or in the case of AI tools in other comprehensive vertical scenarios, after the user inputs the information to be processed, the intent of the information to be processed can be identified, the target plugin to be matched can be determined through intent identification, and then the target plugin and the generative model can be combined to generate the target answer result. Through intent identification, it is possible to accurately determine which target plugin to call, which improves efficiency and accuracy, and thus can effectively answer the information to be processed.
[0054] In another possible implementation, the target response result is determined in the following way:
[0055] S1. When the information to be processed is input for the selected target functional module, determine the target plugin associated with the target functional module. Different target functional modules are used to answer questions in different vertical scenarios.
[0056] In this embodiment of the disclosure, based on artificial intelligence technology, different vertical scenarios can be further subdivided, and AI tools under different vertical scenarios can be implemented to provide users with such tools. For example, drawing AI tools, content creation AI tools, and today's stock market AI tools. Here, different target functional modules can be understood as corresponding to an AI tool. For a specific AI tool, the matching target plugin can be pre-set. For example, today's weather AI tools and today's stock market AI tools have real-time timeliness requirements, so the target plugin to be used can be directly specified as a search plugin.
[0057] Then, after a user selects an AI tool for a specific vertical scenario, they can enter the information to be processed in the chat page of that AI tool. At this point, it is not necessary to determine the target plugin through intent recognition. Instead, the plugin can be directly invoked based on the plugins pre-set for that AI tool.
[0058] S2. Based on the generative model and the target plugin, generate the target answer result in a collaborative manner.
[0059] Thus, in this embodiment of the disclosure, users can be provided with AI tools for different vertical scenarios. When a user selects a specific vertical scenario and inputs information to be processed, the target plugin can be determined directly based on the pre-set association relationship. This is simple to implement and highly efficient. Then, based on the collaboration mode adapted to the target plugin, the target plugin and the generative model work together to generate the target answer result.
[0060] In the two embodiments described above, the target generation model and the target plugin can interact and cooperate in a collaborative manner related to the capability requirements of the target plugin, so as to generate target answer results more accurately and effectively. This disclosure also provides several possible collaboration methods. Specifically, regarding the above-mentioned collaboration method between the generative model and the target plugin, generating target answer results based on the generative model and the target plugin can include the following possible embodiments:
[0061] 1) In one possible embodiment, the target plugin is invoked through a generative model to obtain the initial answer result of the target plugin, and the target answer result of the information to be processed is obtained based on the generative model and the initial answer result.
[0062] In this embodiment, the collaboration between the generative model and the target plugin can be achieved by having the target plugin pre-installed in the generative model. For example, some search plugins or some plugins with time-sensitive requirements can be used. This embodiment does not impose any restrictions. The search plugin can be used to perform a search first to obtain real-time search results, and then the generative model can be used to summarize the real-time search results to obtain the final target answer.
[0063] 2) In another possible embodiment, the target plugin is invoked through a generative model to obtain the target answer result of the information to be processed generated by the target plugin.
[0064] In this embodiment, the collaboration between the generative model and the target plugin can be achieved by placing the target plugin after the generative model. For example, a drawing plugin can be used by the generative model to call the drawing plugin and generate image drawing results based on the drawing plugin.
[0065] Of course, in this embodiment of the disclosure, there are no restrictions on the collaboration method between the target plugin and the generative model, nor is it limited to the two collaboration methods mentioned above. It can be set according to the capability requirements of different target plugins. Furthermore, the target plugin and the generative model may not only collaborate once, but may also collaborate multiple times. For example, the collaboration method is to process based on the target plugin, then give it to the generative model for processing, and then generate the target answer result based on the target plugin.
[0066] Furthermore, in this embodiment of the disclosure, taking the search plugin and the drawing plugin as examples, a more specific collaborative processing procedure is provided, which will be described below.
[0067] In one possible scenario, when the target plugin is a search plugin, the collaboration between the search plugin and the generative model is as follows: the generative model calls the target plugin to obtain the initial answer result of the target plugin, and based on the generative model and the initial answer result, the target answer result of the information to be processed is obtained.
[0068] In particular, this disclosure also provides possible implementation methods for obtaining the initial answer result of the target plugin by calling the target plugin through a generative model, including:
[0069] 1) When the target plugin is a search plugin, based on the generative model, the information to be processed and the inductive prompt statement are used as input. According to the inductive prompt statement, the information to be processed is inductively processed to generate a search statement with the same semantics as the information to be processed. The inductive prompt statement is used to indicate the inductive requirements for inductive processing.
[0070] In this embodiment of the disclosure, the search capability of the search plugin typically requires inputting search terms or search statements into the search plugin before it performs a real-time search, which can retrieve multiple search results. However, in this embodiment of the disclosure, the AI tool does not need to return many search results to the user, nor is it conducive to the user quickly obtaining the answer. Therefore, it can also send multiple search results obtained by the search plugin to the generative model, and then the generative model performs semantic analysis on the multiple search results, and finally generates one or a preset number of target answer results to return to the user.
[0071] The purpose of the inductive suggestion statement is to instruct the generative model to generate a search statement that meets the search requirements. The information to be processed by the user may be extensive and arbitrary, and cannot be directly searched as a search statement. Therefore, in this embodiment, the information to be processed can be inductively processed to generate a search statement with the same semantics as the information to be processed.
[0072] Specifically, regarding the process of summarizing information based on inductive prompts to generate search statements with the same semantic meaning as the information to be processed, this disclosure also provides possible implementation methods:
[0073] In one possible implementation, semantic analysis is performed on the information to be processed based on the inductive prompt statement to determine the first semantic topic of the information to be processed, and search statements that match the first semantic topic are extracted from the information to be processed based on the first semantic topic.
[0074] In this implementation, by summarizing prompts, the generative model can be instructed to summarize the search statement that represents the first semantic topic of the information to be processed. In order to ensure accuracy and avoid large deviations, the generative model can be instructed to use the text that has appeared in the information to be processed to extract the search statement from the information to be processed.
[0075] In another possible implementation, semantic analysis is performed on the information to be processed and the multi-turn dialogue information associated with the information to be processed to determine the second semantic theme corresponding to the information to be processed and the multi-turn dialogue information, and based on the second semantic theme, search statements that conform to the second semantic theme are extracted from the information to be processed and the multi-turn dialogue information.
[0076] In this implementation, which is mainly aimed at multi-turn dialogue scenarios, the generative model can be instructed to summarize the search statement that can represent the second semantic topic of the multi-turn dialogue information based on the inductive prompt statement. It can also be instructed to use the text that appears in the multi-turn dialogue information to extract the search statement.
[0077] Furthermore, the summary prompt statement may also include other summary requirements. For example, if the multi-turn dialogue information includes multiple semantically unrelated texts, then a summary is made based on the last round of dialogue. Another example is that the form of the search statement may be limited, such as the search statement not containing punctuation marks, or the number of characters in the search statement not exceeding a certain set character threshold. However, no specific restrictions are imposed in the embodiments of this disclosure.
[0078] 2) The search plugin is invoked through the generative model, and the initial search results matching the search statement are obtained based on the search plugin.
[0079] The obtained initial search results can then be sent to the generative model, which will further summarize and generalize the initial search results to generate the final target answer. For example, the generative model can perform semantic analysis on multiple initial search results to generate a target answer, or it can generate multiple target answer results. This disclosure does not impose any limitations.
[0080] In one possible scenario, where the target plugin is a drawing plugin, the collaboration between the search plugin and the generative model involves the generative model calling the target plugin to obtain the target answer result of the information to be processed generated by the target plugin.
[0081] Specifically, regarding the target answer result obtained by calling the target plugin through a generative model to obtain the information to be processed generated by the target plugin, this disclosure also provides possible implementation methods, including:
[0082] 1) When the target plugin is a drawing plugin, based on the generative model, the information to be processed is used as input to perform semantic analysis on the information to be processed and extract the key image description information corresponding to the information to be processed.
[0083] In this embodiment of the disclosure, for the drawing plugin, in order to improve the accuracy of the drawing plugin, the key descriptive information of the image to be processed can be extracted first based on the generative model, such as image style information, image content main information, etc.
[0084] 2) Call the drawing plugin and input the key image description information into the drawing plugin. Based on the drawing plugin, generate the image corresponding to the key image description information.
[0085] For example, the drawing plugin uses a text-to-image model. Based on this model, semantic analysis is performed on the key descriptive information of the image to obtain the text feature information of the key descriptive information of the image. Based on the text feature information, the image feature information associated with the text feature information is determined. Based on the image feature information, the image generation result is obtained.
[0086] Thus, in this embodiment of the disclosure, for different plugins such as drawing plugins or search plugins, a collaborative interaction between generative models and plugins can be achieved based on a collaborative method that better meets the plugin capability implementation requirements, ultimately generating the target answer result. This can improve the performance of plugin capability implementation and also improve the accuracy of the target answer result.
[0087] The information processing method in this embodiment of the present disclosure will be described below using a specific application scenario. One possible application scenario is to take the input of information to be processed in the main dialogue flow of the application as an example, which includes a drawing plugin and a search plugin.
[0088] See Figure 2 The diagram shown is a flowchart of another information processing method according to an embodiment of this disclosure. Figure 2As shown, for example, when a user opens the application, the homepage displays target function modules for various vertical scenarios. The user does not select a target function module, but instead enters the information to be processed in the chat page on the homepage. Alternatively, the homepage may also display a comprehensive function module, such as an AI assistant, which does not correspond to a specific vertical scenario. The user can select the AI assistant to enter the information to be processed.
[0089] like Figure 2 As shown, the system receives user input information to be processed, performs intent recognition on the information, and calls the matching target plugin based on the identified target intent category of the information. For example, if the intent is drawing, the target plugin is the drawing plugin; if the intent is searching, the target plugin is the search plugin. This can also include other plugins and their corresponding intents. When the system determines that the user has another intent, it calls that other plugin.
[0090] The output of the target plugin is then sent to the generative model, which summarizes and obtains the target answer, and then replies to the user to complete the dialogue.
[0091] In addition, if no target plugin is matched and the generative model can answer the information to be processed, the target answer result is generated directly based on the generative model.
[0092] Taking the target plugin as an example, the initial answer output by the search plugin is the initial search result. (See [link / reference]). Figure 3 The diagram shown is a flowchart of the search plugin in the information processing method of this disclosure embodiment, including:
[0093] S301: Receive information to be processed.
[0094] S302: Determine if it is a search intent. If yes, proceed to step S305; otherwise, proceed to step S303.
[0095] S303: Based on a generative model, generate the target answer result for the information to be processed.
[0096] In this embodiment, it is mainly for the purpose of explanation that when it is neither a search intent nor a match with other target plugins, and the generative model can answer the information to be processed, the semantic analysis of the information to be processed can be performed directly based on the generative model to generate the target answer result of the information to be processed.
[0097] S304: Return the target answer result.
[0098] S305: Start the search.
[0099] S306: Based on a generative model, the information to be processed is summarized and processed to generate search statements.
[0100] S307: Invoke the search plugin and obtain initial search results that match the search query based on the search plugin.
[0101] S308: Send the initial search results to the generative model.
[0102] S309: Generate the target answer based on the generative model and the initial search results.
[0103] For example, based on generative models, the initial search results are summarized to obtain the final target answer.
[0104] For example, after the search begins, the search plugin will perform a search to obtain initial search results in real time. At the same time, the generative model can also answer the questions about the information to be processed, and obtain the first answer result of the generative model. Then, the generative model will summarize the initial search results and the first answer result to generate the final target answer result.
[0105] S310: Return the target answer result.
[0106] Thus, in this embodiment of the disclosure, the target plugin to be invoked can be determined by intent recognition, enabling the generative model to have the ability to invoke which plugin, thereby answering more questions in different vertical scenarios, improving the applicability, and also improving the efficiency and accuracy of answering questions.
[0107] Another possible application scenario is illustrated by taking the input of information to be processed in a selected target functional module within an application. In this embodiment of the disclosure, for a specific target functional module, associated target plugins can be pre-defined. Different target functional modules can be associated with the same target plugin, or they can be different target plugins. Specifically, this can be based on capability requirements and vertical scenario classifications, etc. If a user selects a target functional module and inputs information to be processed in the chat page of that target functional module, the target plugin that matches the information to be processed is the pre-defined plugin.
[0108] For example, see Figure 4 The diagram shown is a flowchart of another information processing method in an embodiment of this disclosure. Figure 4 As shown, taking target functional modules such as news reviews, latest film reviews, today's stock market, latest industry research, and travel guides as examples, these target functional modules all have timeliness requirements. For these types of target functional modules, the target plugin can be directly designated as the search plugin, and the search plugin can be called every time a user asks a question to obtain the latest initial search results, which are then sent to the generative model for summarization to obtain the final target answer result.
[0109] Thus, in this embodiment of the disclosure, for specific vertical scenarios, it is possible to directly pre-set the associated target plugin and the collaboration mode between the target plugin and the generative model without performing intent recognition. Based on the generative model and the target plugin, the target answer result can be generated, achieving higher efficiency, meeting the needs of different vertical scenarios, and improving performance.
[0110] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.
[0111] Based on the same inventive concept, this disclosure also provides an information processing device corresponding to the information processing method. Since the principle of the device in this disclosure for solving the problem is similar to that of the information processing method described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.
[0112] Reference Figure 5 The diagram shown is a schematic representation of an information processing apparatus provided in an embodiment of this disclosure. The apparatus includes:
[0113] Input module 51 is used to receive input information to be processed;
[0114] The acquisition module 52 is used to acquire the target answer result of the information to be processed. The target answer result is generated based on the generative model and the target plugin. The collaborative method of the generative model and the target plugin in generating the target answer result is related to the capability implementation requirements of the target plugin. The target plugin is a plugin that matches the information to be processed and can answer the information to be processed.
[0115] Display module 53 is used to display the target answer result.
[0116] In an optional implementation, the system further includes a generation module 54, which determines the target answer result in the following manner:
[0117] The information to be processed is subjected to intent recognition to determine the target intent category of the information to be processed; wherein, the target intent category represents the capability requirement to answer the information to be processed;
[0118] Based on the target intent category and the association between the intent category and the plugin, the target plugin that matches the target intent category is determined. Different plugins have different capabilities.
[0119] Based on the generative model and the target plugin, the target answer result is generated in a collaborative manner.
[0120] In an optional implementation, the target answer result is further determined by the generation module 54 in the following manner:
[0121] When the information to be processed is input for the selected target functional module, the target plugin associated with the target functional module is determined, wherein different target functional modules are used to answer questions in different vertical scenarios;
[0122] Based on the generative model and the target plugin, the target answer result is generated in a collaborative manner.
[0123] In an optional implementation, when generating the target answer result based on the generative model and the target plugin in a collaborative manner, the generation module 54 is used to:
[0124] The generative model is used to call the target plugin to obtain the initial answer result of the target plugin, and based on the generative model and the initial answer result, the target answer result of the information to be processed is obtained.
[0125] Alternatively, the target plugin can be invoked through the generative model to obtain the target answer result of the information to be processed generated by the target plugin.
[0126] In an optional implementation, when performing intent recognition on the information to be processed and determining the target intent category of the information to be processed, the generation module 54 is used to:
[0127] Based on the generative model, the information to be processed and the intent judgment prompt statement are used as input. According to the intent judgment prompt statement, semantic analysis is performed on the words included in the information to be processed to determine the target intent category matched by the information to be processed.
[0128] The intent judgment prompt statement is used to indicate the capability requirement judgment requirements for each intent category and to represent the word examples of the corresponding intent category.
[0129] In one optional implementation, when the generative model calls the target plugin to obtain the initial answer result of the target plugin, the generation module 54 is used to:
[0130] When the target plugin is a search plugin, based on the generative model, with the information to be processed and the inductive prompt statement as input, the information to be processed is inductively processed according to the inductive prompt statement to generate a search statement with the same semantics as the information to be processed. The inductive prompt statement is used to indicate the inductive requirements for inductive processing.
[0131] The generative model invokes the search plugin, and the initial search results matching the search statement are obtained based on the search plugin.
[0132] In an optional implementation, when the step of summarizing the information to be processed based on the inductive prompt statement to generate a search statement with the same semantics as the information to be processed, the generation module 54 is used to:
[0133] Based on the inductive prompts, semantic analysis is performed on the information to be processed to determine the first semantic theme of the information to be processed, and based on the first semantic theme, search statements that match the first semantic theme are extracted from the information to be processed.
[0134] Alternatively, semantic analysis can be performed on the information to be processed and the multi-turn dialogue information associated with the information to be processed to determine the second semantic theme corresponding to the information to be processed and the multi-turn dialogue information, and based on the second semantic theme, search statements that conform to the second semantic theme can be extracted from the information to be processed and the multi-turn dialogue information.
[0135] In an optional implementation, when the generative model calls the target plugin to obtain the target answer result of the information to be processed generated by the target plugin, the generation module 54 is used to:
[0136] When the target plugin is a drawing plugin, based on the generative model, the information to be processed is used as input to perform semantic analysis on the information to be processed, and key image description information corresponding to the information to be processed is extracted.
[0137] The drawing plugin is invoked, and the key image description information is input into the drawing plugin. Based on the drawing plugin, the image generation result corresponding to the key image description information is generated.
[0138] In an optional implementation, the generation module 54 is further configured to: in response to the fact that the information to be processed does not match the target plugin and the generative model is able to answer the information to be processed, perform semantic analysis on the information to be processed based on the generative model, and generate a target answer result for the information to be processed.
[0139] The processing flow of each module in the device and the interaction flow between each module can be referred to the relevant descriptions in the above method embodiments, and will not be detailed here.
[0140] This disclosure also provides an electronic device, such as... Figure 6 The diagram shown is a schematic representation of the structure of an electronic device provided in an embodiment of this disclosure, including:
[0141] A processor 61 and a memory 62; the memory 62 stores machine-readable instructions executable by the processor 61, and the processor 61 executes the machine-readable instructions stored in the memory 62. When the machine-readable instructions are executed by the processor 61, the processor 61 performs the following steps:
[0142] Receive input information to be processed;
[0143] Obtain the target answer result for the information to be processed. The target answer result is generated based on a generative model and a target plugin. The collaborative method between the generative model and the target plugin in generating the target answer result is related to the capability requirements of the target plugin. The target plugin is a plugin that matches the information to be processed and is able to answer the information to be processed.
[0144] Display the target response results.
[0145] The aforementioned memory 62 includes a main memory 621 and an external memory 622. The main memory 621, also known as internal memory, is used to temporarily store the computational data in the processor 61, as well as the data exchanged with external memory such as a hard disk. The processor 61 exchanges data with the external memory 622 through the main memory 621.
[0146] The specific execution process of the above instructions can be referred to the steps of the information processing method described in the embodiments of this disclosure, and will not be repeated here.
[0147] This disclosure also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the information processing method described in the above-described method embodiments. The storage medium may be a volatile or non-volatile computer-readable storage medium.
[0148] This disclosure also provides a computer program product carrying program code. The program code includes instructions that can be used to execute the steps of the information processing method described in the above method embodiments. For details, please refer to the above method embodiments, which will not be repeated here.
[0149] The aforementioned computer program product can be implemented through hardware, software, or a combination thereof. In one optional embodiment, the computer program product is specifically embodied in a computer storage medium; in another optional embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.
[0150] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and devices described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. In the several embodiments provided in this disclosure, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division; in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection may be through some communication interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.
[0151] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0152] In addition, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0153] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0154] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit it. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this disclosure. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be determined by the protection scope of the claims.
Claims
1. An information processing method, characterized in that, include: Receive input information to be processed; Obtain the target answer result for the information to be processed. The target answer result is generated based on a generative model and a target plugin. The collaborative method between the generative model and the target plugin in generating the target answer result is related to the capability requirements of the target plugin. The target plugin is a plugin that matches the information to be processed and is able to answer the information to be processed. Display the target response results; The target answer result is determined in the following way: the target answer result is generated based on the generative model and the target plugin in a collaborative manner. The step of generating the target answer result based on the generative model and the target plugin in a collaborative manner includes: calling the target plugin through the generative model to obtain the initial answer result of the target plugin, and obtaining the target answer result of the information to be processed based on the generative model and the initial answer result; or, calling the target plugin through the generative model to obtain the target answer result of the information to be processed generated by the target plugin.
2. The method according to claim 1, characterized in that, The target plugin was determined in the following way: The information to be processed is subjected to intent recognition to determine the target intent category of the information to be processed; wherein, the target intent category represents the capability requirement to answer the information to be processed; Based on the target intent category and the association between the intent category and the plugin, the target plugin that matches the target intent category is determined. Different plugins implement different capabilities.
3. The method according to claim 1, characterized in that, The target plugin was determined in the following way: When the information to be processed is input for a selected target functional module, the target plugin associated with the target functional module is determined, wherein different target functional modules are used to answer questions in different vertical scenarios.
4. The method according to claim 2, characterized in that, The step of performing intent recognition on the information to be processed and determining the target intent category of the information to be processed includes: Based on the generative model, the information to be processed and the intent judgment prompt statement are used as input. According to the intent judgment prompt statement, semantic analysis is performed on the words included in the information to be processed to determine the target intent category matched by the information to be processed. The intent judgment prompt statement is used to indicate the capability requirement judgment requirements for each intent category and to represent the word examples of the corresponding intent category.
5. The method according to claim 1, characterized in that, The generative model invokes the target plugin to obtain the initial response result of the target plugin, including: When the target plugin is a search plugin, based on the generative model, with the information to be processed and the inductive prompt statement as input, the information to be processed is inductively processed according to the inductive prompt statement to generate a search statement with the same semantics as the information to be processed. The inductive prompt statement is used to indicate the inductive requirements for inductive processing. The generative model invokes the search plugin, and the initial search results matching the search statement are obtained based on the search plugin.
6. The method according to claim 5, characterized in that, The step of summarizing the information to be processed based on the inductive prompt statement to generate a search statement with the same semantics as the information to be processed includes: Based on the inductive prompts, semantic analysis is performed on the information to be processed to determine the first semantic theme of the information to be processed, and based on the first semantic theme, search statements that match the first semantic theme are extracted from the information to be processed. Alternatively, semantic analysis can be performed on the information to be processed and the multi-turn dialogue information associated with the information to be processed to determine the second semantic theme corresponding to the information to be processed and the multi-turn dialogue information, and based on the second semantic theme, search statements that conform to the second semantic theme can be extracted from the information to be processed and the multi-turn dialogue information.
7. The method according to claim 1, characterized in that, The step of calling the target plugin through the generative model to obtain the target answer result of the information to be processed generated by the target plugin includes: When the target plugin is a drawing plugin, based on the generative model, the information to be processed is used as input to perform semantic analysis on the information to be processed, and key image description information corresponding to the information to be processed is extracted. The drawing plugin is invoked, and the key image description information is input into the drawing plugin. Based on the drawing plugin, the image generation result corresponding to the key image description information is generated.
8. The method according to claim 1, characterized in that, The method further includes: In response to the fact that the information to be processed does not match the target plugin, and the generative model is able to answer the information to be processed, semantic analysis is performed on the information to be processed based on the generative model to generate the target answer result for the information to be processed.
9. An information processing device, characterized in that, include: The input module is used to receive input information to be processed. The acquisition module is used to acquire the target answer result of the information to be processed. The target answer result is generated based on the generative model and the target plugin. The collaborative method of the generative model and the target plugin in generating the target answer result is related to the capability implementation requirements of the target plugin. The target plugin is a plugin that matches the information to be processed and can answer the information to be processed. The display module is used to display the target answer results; The target answer result is determined in the following way: the target answer result is generated based on the generative model and the target plugin in a collaborative manner. The step of generating the target answer result based on the generative model and the target plugin in a collaborative manner includes: calling the target plugin through the generative model to obtain the initial answer result of the target plugin, and obtaining the target answer result of the information to be processed based on the generative model and the initial answer result; or, calling the target plugin through the generative model to obtain the target answer result of the information to be processed generated by the target plugin.
10. An electronic device, characterized in that, include: A processor and a memory, the memory storing machine-readable instructions executable by the processor, the processor executing the machine-readable instructions stored in the memory, wherein when the machine-readable instructions are executed by the processor, the processor performs the steps of the method as described in any one of claims 1 to 8.
11. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1-8.