Information processing method and device, electronic equipment and medium

Abstract

The invention provides an information processing method and device, electronic equipment and a medium, and relates to the technical field of artificial intelligence, and the specific implementation scheme comprises the steps: responding to the obtained input information, using a large language model to process the input information, and obtaining a task execution strategy, the task execution strategy comprises a plurality of demand tasks, the plurality of demand tasks are matched with demand intentions represented by the input information; determining an operator matched with each demand task in the task execution strategy; based on the operator matched with each demand task, constructing a directed acyclic graph, the directed acyclic graph comprising a plurality of operators and a plurality of directed edges, the plurality of directed edges being used for indicating an execution dependency relationship among the plurality of operators; according to the directed acyclic graph, determining an execution result related to the task execution strategy; and generating feedback information according to the execution result, and displaying the feedback information.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, specifically to the fields of natural language processing and deep learning, and particularly to an information processing method, apparatus, electronic device, storage medium, and computer program product. Background Technology

[0002] With the rapid growth of data in the Geographic Information System (GIS) field, the diversity and complexity of this data pose new challenges to geographic information processing. As the level of intelligence in remote sensing continues to improve, intelligent geographic information processing becomes particularly important. Summary of the Invention

[0003] This invention provides an information processing method, apparatus, electronic device, storage medium, and computer program product.

[0004] According to one aspect of the present invention, an information processing method is provided, comprising: in response to obtaining input information, processing the input information using a large language model to obtain a task execution strategy, the task execution strategy including multiple requirement tasks, the multiple requirement tasks being matched with requirement intentions represented by the input information; determining operators matching each requirement task in the task execution strategy; constructing a directed acyclic graph based on the operators matching each requirement task, the directed acyclic graph including multiple operators and multiple directed edges, the multiple directed edges being used to indicate execution dependencies between the multiple operators; determining execution results related to the task execution strategy according to the directed acyclic graph; generating feedback information based on the execution results, and displaying the feedback information.

[0005] According to another aspect of the present invention, an information processing apparatus is provided, comprising: a processing module, configured to process the input information using a large language model in response to receiving input information to obtain a task execution strategy, the task execution strategy including multiple demand tasks, the multiple demand tasks being matched with demand intentions represented by the input information; a first determining module, configured to determine an operator matching each demand task in the task execution strategy; a construction module, configured to construct a directed acyclic graph based on the operator matching each demand task, the directed acyclic graph including multiple operators and multiple directed edges, the multiple directed edges indicating execution dependencies between the multiple operators; a second determining module, configured to determine an execution result related to the task execution strategy based on the directed acyclic graph; and a display module, configured to generate feedback information based on the execution result and display the feedback information.

[0006] According to another aspect of the present invention, an electronic device is provided, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors perform an information processing method as described above.

[0007] According to another aspect of the present invention, a computer-readable storage medium is provided having executable instructions stored thereon, which, when executed by a processor, cause the processor to perform the information processing method described above.

[0008] According to another aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the information processing method as described above. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0010] Figure 1 This is a flowchart of an information processing method according to an embodiment of the present invention;

[0011] Figure 2 This is a block diagram of an information processing apparatus according to an embodiment of the present invention;

[0012] Figure 3 This is a block diagram of an electronic device suitable for implementing an information processing method according to an embodiment of the present invention. Detailed Implementation

[0013] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0014] It should be noted that the sequence numbers of each operation in the following methods are merely for descriptive purposes and should not be construed as indicating the execution order of the operations. Unless explicitly stated otherwise, the method does not need to be executed in the exact order shown. Furthermore, in the description of this invention, terms such as "first," "second," etc. (if present) are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0015] In the technical solution of this invention, the collection, storage, use, processing, transmission, provision, disclosure, and application of data (including but not limited to user personal information) comply with the relevant laws and regulations of the relevant countries and regions, and do not violate public order and good morals. Furthermore, in the technical solution of this invention, authorization or consent from the data owner is obtained before acquiring or collecting any relevant data.

[0016] Figure 1 This is a flowchart of an information processing method according to an embodiment of the present invention.

[0017] like Figure 1 As shown, the information processing method 100 includes operations S110 to S150.

[0018] In operation S110, in response to the acquisition of input information, the input information is processed using a large language model to obtain a task execution strategy. The task execution strategy includes multiple demand tasks, and the multiple demand tasks are matched with the demand intent represented by the input information.

[0019] In operation S120, the operator that matches each required task in the task execution strategy is determined.

[0020] In operation S130, a directed acyclic graph is constructed based on the operators that match each requirement task. The directed acyclic graph includes multiple operators and multiple directed edges, which are used to indicate the execution dependencies between multiple operators.

[0021] In operation S140, the execution results related to the task execution strategy are determined based on the directed acyclic graph.

[0022] During operation S150, feedback information is generated and displayed based on the execution results.

[0023] According to embodiments of the present invention, the input information can be natural language information that characterizes a user's information retrieval needs. For example, the input information could be "Please provide the rice distribution information for region A in 2023," which characterizes the user's intention to obtain information about the rice distribution in region A in 2023. The input information can be obtained, for example, by inputting text or by converting input speech information; embodiments of the present invention do not limit the method of obtaining the input information.

[0024] According to embodiments of the present invention, a Large Language Model (LLM) is a deep learning model trained using a large amount of text data, such as, but not limited to, language models based on the Transformer architecture. Large Language Models can understand and generate natural language text. They typically contain a large number of parameters, enabling them to capture complex patterns and structures in language, thus performing well in various natural language processing tasks.

[0025] According to an embodiment of the present invention, a task execution strategy can characterize one or more requirement tasks to be performed in order to obtain requirement-related information, wherein the requirement-related information is matched with the requirement intent represented by the input information.

[0026] According to embodiments of the present invention, a large language model can be pre-trained using business application data related to the remote sensing field, enabling the large language model to output a task execution strategy that matches the demand intent represented by the input information based on the input information expressed in natural language. The task execution strategy may include, for example, one or more demand tasks.

[0027] For example, in the input information "Please provide the rice distribution information for region A in 2023", processing the input information using a large language model yields the following task execution strategy: Task 1 "Find images of region A in 2023"; Task 2 "Find an image interpretation model"; Task 3 "Use the image interpretation model to interpret the images of region A in 2023"; Task 4 "Post-process the image interpretation results to obtain the rice distribution information for region A in 2023". Tasks 1 through 4 in the task execution strategy match the intent represented by the input information. By executing Tasks 1 through 4, relevant demand information related to the input information can be obtained.

[0028] According to embodiments of the present invention, operators matching each requirement task in the task execution strategy can be obtained, and operator dependency chains can be constructed based on these operators. The operator dependency chains include multiple operators and execution dependencies between them. Then, each requirement task in the task execution strategy is executed based on the operator dependency chains, thereby obtaining requirement-related information associated with the input information.

[0029] According to embodiments of the present invention, a requirement task may include task attribute information such as task type and task configuration parameters. For example, the task type corresponding to the above-mentioned requirement task 1 "find image of region A in 2023" is, for example, an image search task type, and the task configuration parameters it includes may include, for example, "2023" and "region A". For each requirement task in the task execution strategy, an operator matching the requirement task can be determined based on the task attribute information contained in the requirement task.

[0030] The process of determining the operator that matches each requirement task is described below.

[0031] First, select template operators that match the required task from the preset operator library.

[0032] According to embodiments of the present invention, the operator library stores template operators for various task types, including but not limited to template operators for tasks such as data input, data output, data transformation, data cleaning, data repair, spatial analysis, image search, image interpretation, and post-processing. Each template operator includes operator attribute information and operator parameters. Operator attribute information may include, but is not limited to, the template operator's identification information, version information, category information, status information, supported application scenarios, and other descriptive information related to the template operator. Operator parameters may include, but are not limited to, operator configuration parameters, at least one input data resource parameter, at least one output data resource parameter, and a data resource service identifier. The operator configuration parameters provide the template operator with the configuration information required for execution and typically do not participate in data flow transmission. Each input data resource parameter includes an input data identifier and an input data type. The input data identifier indicates the input data that the template operator needs to receive during execution, and the input data type indicates the data type of the input data that the template operator needs to receive during execution. Each output data resource parameter includes an output data identifier and an output data type. The output data identifier represents the output data generated after the template operator is executed, and the output data type indicates the data type of the output data. The data resource service identifier represents the association between the template operator and the data resource service, which is associated with the input data of the template operator. The input data of the template operator can be obtained by calling the data resource service.

[0033] According to embodiments of the present invention, template operators matching the required task can be queried in an operator library based on the task type corresponding to the required task. For example, taking the aforementioned required task 1, "Find images of region A in 2023," as an example, the task type corresponding to required task 1, "Find images of region A in 2023," is an image search task. Template operators for image search can be searched in the operator library to obtain template operators matching required task 1, "Find images of region A in 2023." In other embodiments, related information can also be searched from a pre-built knowledge base based on the required task. Related information may include contextual information, professional knowledge, etc., related to the required task. Then, based on the required task and related information, template operators matching the required task are queried in a preset operator library.

[0034] Next, key information is extracted from the task requirements, and the operator parameters of the corresponding template operators are configured based on the extracted key information to obtain operators that match the task requirements.

[0035] According to an embodiment of the present invention, after obtaining a template operator that matches the required task, key information can be extracted from the required task to obtain the task configuration parameters contained in the required task. Then, the extracted key information is provided to the template operator that matches the required task to update some or all of the operator configuration parameters in the template operator, so that the updated operator configuration parameters are adapted to the user's required intent.

[0036] For example, taking the aforementioned task 1, "Find imagery of region A in 2023," as an example, after obtaining the template operator matching task 1, key information can be extracted from task 1 to obtain the task configuration parameters contained in task 1, such as "2023" and "region A." Then, this key information is provided to the template operator matching task 1 to update some or all of the operator configuration parameters in the template operator, thereby obtaining the operator matching task 1. This operator will be used to find imagery of region A in 2023.

[0037] It should be noted that in some embodiments, if no key information is extracted from the requirement task, there is no need to adjust the operator parameters of the template operator that matches the requirement task, and the template operator can be directly used as the operator that matches the requirement task.

[0038] According to embodiments of the present invention, the above-mentioned operator dependency chain can be represented based on a directed acyclic graph. Nodes in the directed acyclic graph can represent operators, and directed edges between nodes can represent execution dependencies between operators. Here, the execution dependency between operators refers to the data input-output relationship between operators.

[0039] According to an embodiment of the present invention, constructing a directed acyclic graph based on an operator that matches each required task includes the following steps.

[0040] First, for any target operator among multiple operators, obtain the operator parameters of the target operator, which include the output data identifier and output data type related to the target operator.

[0041] According to an embodiment of the present invention, at least one output data resource parameter of the target operator can be obtained from the operator parameters of the target operator, each output data resource parameter including an output data identifier and an output data type. This at least one output data resource parameter will be used to match downstream operators of the target operator.

[0042] Next, a candidate list is obtained, which includes at least one candidate operator and an input data identifier and input data type associated with each of the at least one candidate operator, wherein the at least one candidate operator is an operator other than the target operator among a plurality of operators.

[0043] According to embodiments of the present invention, each candidate operator and at least one input data resource parameter of each candidate operator can be stored in a candidate list for subsequent matching of downstream operators of the target operator in the candidate list. The at least one input data resource parameter of each candidate operator is obtained from the operator parameters of that candidate operator, and each input data resource parameter includes an input data identifier and an output data type.

[0044] Next, based on the output data identifier and output data type associated with the target operator, downstream operators corresponding to the target operator are selected from the candidate list. The input data identifier of the downstream operator corresponds to the output data identifier, and the input data type of the downstream operator corresponds to the output data type.

[0045] According to embodiments of the present invention, downstream operators of a target operator refer to candidate operators that require at least a portion of the output data of the target operator as input data. Embodiments of the present invention can utilize at least one output data resource parameter of the target operator to determine the downstream operators of the target operator.

[0046] For example, for each output data resource parameter (including output data identifier and output data type) of the target operator, a candidate operator whose input data identifier is consistent with the output data identifier in the output data resource parameter and whose input data type is consistent with the output data type in the output data resource parameter is queried from the candidate list, and the queried candidate operator is used as the downstream operator of the target operator.

[0047] Next, repeat the above operation until you obtain the downstream operators corresponding to each of the multiple operators.

[0048] According to embodiments of the present invention, the downstream operators corresponding to each operator can be obtained through the above method. Subsequently, execution dependencies between operators can be constructed based on the relationship between each operator and its corresponding downstream operator.

[0049] Next, for each operator, the operator and its downstream operators are connected by connecting lines to obtain a directed acyclic graph.

[0050] According to an embodiment of the present invention, each of the plurality of operators is treated as a node in a directed acyclic graph. Then, for each of the plurality of operators, based on the aforementioned downstream operator selection results, directed edges are constructed from the operator to its downstream operators via directed connecting lines. These directed edges are used to characterize the execution dependency relationship between the operator and its downstream operators. Based on each node and each directed edge, a directed acyclic graph is obtained.

[0051] According to an embodiment of the present invention, determining the execution result related to the task execution strategy based on the directed acyclic graph includes: calling the data resource service associated with each operator in the directed acyclic graph to perform task processing based on the directed acyclic graph, and obtaining the execution result.

[0052] According to an embodiment of the present invention, the operator parameters of each operator include a data resource service identifier, which is used to characterize the association between the operator and the data resource service, and the data resource service is associated with the input data of the operator. During task processing based on a directed acyclic graph (DAG), the data resource services associated with each operator can be sequentially invoked according to the execution dependencies between multiple operators indicated by the DAG to obtain the corresponding input data, and task processing is performed based on the input data to finally obtain the execution result. This execution result contains information that matches the intended demand represented by the input information.

[0053] According to an embodiment of the present invention, before determining the execution result related to the task execution strategy based on the directed acyclic graph, the directed acyclic graph can be displayed through an interactive interface for user confirmation. This helps to improve the matching degree between the directed acyclic graph and the user's true intention, thereby ensuring that the execution result generated based on the directed acyclic graph is more in line with the user's actual needs.

[0054] According to embodiments of the present invention, users can also input adjustment information for directed acyclic graphs (DAGs) through an interactive interface. The adjustment information may be, for example, natural language information instructing the execution of any type of operation, such as deletion, modification, or replacement, on the operator configuration parameters of operators in the DAG. In some embodiments, the adjustment information may also be natural language information instructing the correction of execution dependencies between multiple operators in the DAG.

[0055] After obtaining the adjustment information input by the user, the directed acyclic graph (DAG) can be adjusted based on this information to obtain an adjusted DAG that matches the user's current needs. Then, data resource services associated with each operator in the adjusted DAG can be invoked to process tasks based on the adjusted DAG and obtain the execution results.

[0056] According to an embodiment of the present invention, after obtaining the execution result, feedback information can be generated based on the execution result and displayed through an interactive interface. Exemplarily, embodiments of the present invention can directly use the execution result as feedback information, or can obtain feedback information by filling a preset configuration template based on the execution result; this is not limited here.

[0057] According to an embodiment of the present invention, input information is processed using a large language model to obtain a task execution strategy. Multiple requirement tasks in the task execution strategy are matched with the requirement intent represented by the input information. Then, operators matching each requirement task are determined, and a directed acyclic graph (DAG) is constructed based on these operators. The DAG includes multiple operators and multiple directed edges, with the directed edges indicating the execution dependencies between operators. Subsequently, based on the DAG, the execution results related to the task execution strategy are determined, and feedback information is further generated based on the execution results. Through the solution of the present invention, it is possible to automatically and intelligently orchestrate operator dependency chains that conform to user requirement intents based on an understanding of the input information, execute information processing tasks, and obtain corresponding feedback results. This not only improves the efficiency and accuracy of information processing but also has good scalability, meeting the information processing needs of different scenarios.

[0058] According to an embodiment of the present invention, the process of using a large language model to process input information and obtain a task execution strategy includes the following steps.

[0059] First, a large language model is used to process the input information to determine whether the demand intent represented by the input information is ambiguous.

[0060] Understandably, the intent represented by the input information may be clear or vague. However, the ambiguity of the input information can affect subsequent understanding and execution, leading to a deviation between the final execution result and the user's intent. This not only affects the efficiency and accuracy of information processing, resulting in wasted computing resources, but also impacts the user experience. Therefore, after acquiring the input information, a large language model can be used to process it to determine whether the intent represented by the input information is vague.

[0061] According to embodiments of the present invention, dialogue sample data can be constructed using pre-defined annotation information. This dialogue sample data can be dialogues between users and a large language model. Then, the large language model is pre-trained based on the dialogue sample data, enabling it to determine the ambiguity of the demand intent represented by the input information based on natural language expressions. If the demand intent represented by the input information is ambiguous, it identifies missing target attribute items in the input information and uses at least one information interaction between the large language model and the user to supplement the attribute values ​​of the target attribute items, thereby obtaining task demand information to represent a clear demand intent.

[0062] According to embodiments of the present invention, the preset annotation information includes, for example, annotation information corresponding to multiple task instructions. Each task instruction is natural language information used to represent the user's information acquisition needs. Multiple task instructions may include, for example, task instructions representing clear needs and task instructions representing ambiguous needs. The annotation information corresponding to each task instruction includes, for example, but is not limited to, the needs represented by the task instruction, a judgment on whether the needs represented by the task instruction are ambiguous, known attribute information, and missing attribute information in the task instruction. The known attribute information includes known attribute items contained in the task instruction, attribute identifiers of the known attribute items, and attribute values ​​of the known attribute items. The missing attribute information includes the missing target attribute item in the task instruction, the attribute identifier of the target attribute item, and at least one potential attribute value associated with the target attribute item. The attribute identifier is used to represent the importance of the corresponding attribute item. This importance reflects the information value or criticality of the corresponding attribute item in the task instruction. If the information value or criticality of an attribute item in the task instruction is higher, then the importance of the attribute item in the task instruction is greater, and vice versa. For example, the attribute identifiers "required parameter" and "optional parameter" can be used to represent the importance of the attribute item. For example, the attribute identifier "required parameter" can be used to identify key attribute items, and the attribute identifier "optional parameter" can be used to identify general attribute items, and so on. It should be understood that attribute identifiers are not limited to the above examples, and can be set according to actual needs; this invention does not limit this.

[0063] According to embodiments of the present invention, the known and missing attribute information in each task instruction can be annotated based on attribute items adapted to the task instruction. The attribute items adapted to each task instruction can be set according to the business objective to which the task instruction belongs, wherein the business objective describes the business scenario matching the task instruction. For example, when the business objective of a task instruction is "crop identification," the attribute items adapted to the task instruction may include key attribute items such as "crop category," "spatial range," and "time," as well as general attribute items such as "spatial resolution" and "star source." Based on the attribute items adapted to the task instruction, the known and missing attribute information in the task instruction can be annotated.

[0064] According to an embodiment of the present invention, in the preset annotation information, the determination of whether the intended demand represented by each task instruction is ambiguous can be made, for example, based on whether a key attribute item is missing from the task instruction. If a task instruction is missing a key attribute item, then the intended demand represented by that task instruction is ambiguous; otherwise, the intended demand represented by that task instruction is clear.

[0065] According to an embodiment of the present invention, processing input information using a large language model to determine whether the demand intent represented by the input information is ambiguous includes: using the large language model to identify attributes of the input information to determine the missing target attribute item and the attribute identifier of the target attribute item, wherein the attribute identifier is used to characterize the importance of the target attribute item. Then, in response to determining that the input information is missing a target attribute item, and that the attribute identifier of the target attribute item indicates that the target attribute item is a critical attribute item, it is determined that the demand intent represented by the input information is ambiguous.

[0066] According to embodiments of the present invention, a large language model can be used to perform attribute recognition on input information to determine known attribute items, attribute values ​​of known attribute items, attribute identifiers of known attribute items, and missing target attribute items and their attribute identifiers, thereby obtaining attribute recognition results. Then, based on the attribute recognition results, it is determined whether the intended demand represented by the input information is ambiguous. For example, if the attribute recognition result indicates that the input information does not lack a target attribute item, or if the input information lacks a target attribute item, but the attribute identifier of the target attribute item indicates that the target attribute item is not a key attribute item, then the intended demand represented by the input information is determined to be clear. If the attribute recognition result indicates that the input information lacks a target attribute item, and the attribute identifier of the target attribute item indicates that the target attribute item is a key attribute item, then the intended demand represented by the input information is determined to be ambiguous.

[0067] Next, in response to the ambiguity of the demand intent represented by the input information, task requirement information is generated based on the input information.

[0068] According to embodiments of the present invention, if the input information is processed using a large language model and the desired intent represented by the input information is clear (i.e., the input information does not lack key attribute items), then the large language model can directly derive the task execution strategy based on the input information. If the input information is processed using a large language model and the desired intent represented by the input information is ambiguous (i.e., the missing target attribute items in the input information include key attribute items), then at least one information interaction between the large language model and the user can be used to supplement the attribute values ​​of the missing target attribute items in the input information, thereby obtaining task requirement information. The desired intent represented by the task requirement information is clear. Subsequently, the large language model can be further used to process the task requirement information to obtain the task execution strategy.

[0069] The following describes the process of generating task requirement information based on input information.

[0070] First, generate interactive information based on the input information, and then display the interactive information.

[0071] For example, preset annotation information can be queried based on the missing target attribute item in the input information to obtain prompt information. The prompt information may include the target attribute item and at least one latent attribute value corresponding to the target attribute item. Then, a large language model is used to generate interactive information based on the prompt information. The interactive information includes query information for the missing target attribute item in the input information and at least one latent attribute value corresponding to the target attribute item. Subsequently, the interactive information can be displayed through an interactive interface so that the user can promptly browse and input response information related to the interactive information.

[0072] In this embodiment of the invention, the prompt information helps the large language model understand the meaning expressed by the missing target attribute item in the input information, enabling the large language model to generate interactive information associated with the target attribute item, rather than information unrelated to the target attribute item. The latent attribute values ​​corresponding to the target attribute item in the interactive information can provide corresponding prompts to the user, guiding the user to provide response information that matches the interactive information, thereby helping to reduce the number of dialogue interactions and improve the efficiency of obtaining missing information.

[0073] Next, obtain the response information in response to the interaction.

[0074] In this embodiment of the invention, the response information is feedback information provided by the user related to the interaction information. The response information may include target attribute values ​​for each target attribute item in the input information. In some embodiments, the response information may also include additional attribute values ​​for known attribute items in the input information. The additional attribute values ​​for known attribute items can be used to correct or supplement the attribute values ​​of the known attribute items, so that the corrected or supplemented attribute values ​​of the known attribute items better match the user's intent.

[0075] It should be noted that queries targeting all target attributes in the input information can be implemented through a single-turn dialogue or through multiple-turn dialogues. When implementing through multiple-turn dialogues, each turn can focus on querying and providing suggestions for only one target attribute.

[0076] Next, the large language model is used to generate task requirement information based on input information, interaction information, and response information.

[0077] In this embodiment of the invention, a large language model is used to extract key information (including feedback information for target attribute items and known attribute items) based on input information and contextual information (including interaction information and response information) during the dialogue interaction process. It is then used to determine whether this key information is sufficient to clarify the user's needs and intentions. If it is determined that this key information can clarify the user's needs and intentions, task requirement information is generated based on this key information.

[0078] Next, the task requirement information is processed using a large language model to obtain the task execution strategy.

[0079] In embodiments of the present invention, by utilizing a large language model to determine the ambiguity of the input information, and when it is determined that the demand intention represented by the input information is ambiguous, the ambiguous intention represented by the input information is transformed into a clear intention, so that the task execution strategy can be determined based on the clear intention, and subsequent processing can be carried out accordingly, thereby providing a basis for obtaining accurate feedback information.

[0080] Figure 2 This is a block diagram of an information processing apparatus according to an embodiment of the present invention.

[0081] like Figure 2 As shown, the information processing device 200 includes: a processing module 210, a first determining module 220, a construction module 230, a second determining module 240, and a display module 250.

[0082] Processing module 210 is used to process the input information using a large language model in response to the acquisition of input information to obtain a task execution strategy. The task execution strategy includes multiple requirement tasks, and the multiple requirement tasks are matched with the requirement intent represented by the input information.

[0083] The first determining module 220 is used to determine the operator that matches each required task in the task execution strategy.

[0084] The construction module 230 is used to construct a directed acyclic graph based on operators that match each of the said requirements tasks. The directed acyclic graph includes multiple operators and multiple directed edges, which are used to indicate the execution dependencies between the multiple operators.

[0085] The second determining module 240 is used to determine the execution result related to the task execution strategy based on the directed acyclic graph.

[0086] The display module 250 is used to generate feedback information based on the execution result and display the feedback information.

[0087] It should be noted that the implementation methods, technical problems solved, functions achieved, and technical effects of each module in the device embodiment are the same as or similar to the implementation methods, technical problems solved, functions achieved, and technical effects of each corresponding step in the method embodiment, and will not be repeated here.

[0088] Figure 3 A block diagram of an electronic device suitable for implementing an information processing method according to an embodiment of the present invention is shown schematically.

[0089] like Figure 3 As shown, an electronic device 300 according to an embodiment of the present invention includes a processor 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage portion 308 into a random access memory (RAM) 303. The processor 301 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 301 may also include onboard memory for caching purposes. The processor 301 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present invention.

[0090] RAM 303 stores various programs and data required for the operation of electronic device 300. Processor 301, ROM 302, and RAM 303 are interconnected via bus 304. Processor 301 executes various operations of the method flow according to embodiments of the present invention by executing programs in ROM 302 and / or RAM 303. It should be noted that the programs may also be stored in one or more memories other than ROM 302 and RAM 303. Processor 301 may also execute various operations of the method flow according to embodiments of the present invention by executing programs stored in said one or more memories.

[0091] According to an embodiment of the present invention, the electronic device 300 may further include an input / output (I / O) interface 305, which is also connected to a bus 304. The electronic device 300 may also include one or more of the following components connected to the I / O interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 310 as needed so that computer programs read from it can be installed into the storage section 308 as needed.

[0092] The present invention also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the information processing method according to embodiments of the present invention.

[0093] According to embodiments of the present invention, a computer-readable storage medium may be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of the present invention, a computer-readable storage medium may include ROM 302 and / or RAM 303 and / or one or more memories other than ROM 302 and RAM 303 described above.

[0094] Embodiments of the present invention also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to enable the computer system to implement the information processing method provided in the embodiments of the present invention.

[0095] When the computer program is executed by the processor 301, it performs the functions defined in the system / apparatus of this invention. According to embodiments of the invention, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0096] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and may be downloaded and installed via communication section 309, and / or installed from removable medium 311. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0097] In such an embodiment, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by processor 301, it performs the functions defined in the system of this embodiment of the invention. According to embodiments of the invention, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0098] According to embodiments of the present invention, program code for executing the computer programs provided in the embodiments of the present invention can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0099] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0100] Those skilled in the art will understand that the features described in the various embodiments of the present invention can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in the present invention. In particular, the features described in the various embodiments of the present invention can be combined and / or combined in various ways without departing from the spirit and teachings of the present invention. All such combinations and / or combinations fall within the scope of the present invention.

[0101] The embodiments of the present invention have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of the invention. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of the invention, and all such substitutions and modifications should fall within the scope of the invention.

Claims

1. An information processing method, characterized in that, include: In response to the acquisition of input information, the input information is processed using a large language model to obtain a task execution strategy. The task execution strategy includes multiple demand tasks, and the multiple demand tasks are matched with the demand intent represented by the input information. Identify the operator that matches each required task in the task execution strategy; Based on the operators that match each of the aforementioned requirements, a directed acyclic graph is constructed, which includes multiple operators and multiple directed edges, which are used to indicate the execution dependencies between the multiple operators. Based on the directed acyclic graph, determine the execution results related to the task execution strategy; Feedback information is generated based on the execution result and then displayed.

2. The method according to claim 1, characterized in that, The operator for determining the match between each required task in the task execution strategy includes: for each required task, Select template operators from the operator library that match the required task; Key information is extracted from the required task, and the operator parameters of the corresponding template operator are configured based on the extracted key information to obtain an operator that matches the required task.

3. The method according to claim 2, characterized in that, The construction of the directed acyclic graph based on operators matching each of the stated requirements includes: For any one of the multiple operators, obtain the operator parameters of the target operator, wherein the operator parameters include the output data identifier and output data type related to the target operator; Obtain a candidate list, the candidate list including at least one candidate operator and an input data identifier and input data type associated with each of the at least one candidate operator, wherein the at least one candidate operator is an operator other than the target operator among the plurality of operators; Based on the output data identifier and the output data type, downstream operators corresponding to the target operator are selected from the candidate list, wherein the input data identifier of the downstream operator corresponds to the output data identifier, and the input data type of the downstream operator corresponds to the output data type; Repeat the above operations until you obtain the downstream operators corresponding to each of the multiple operators; For each operator, the operator and its downstream operators are connected by directed lines to obtain the directed acyclic graph.

4. The method according to claim 3, characterized in that, The step of determining the execution result related to the task execution strategy based on the directed acyclic graph includes: The data resource services associated with each operator in the directed acyclic graph are invoked to perform task processing based on the directed acyclic graph, and the execution result is obtained. The operator parameters of each operator include a data resource service identifier, which is used to characterize the association between the operator and the data resource service.

5. The method according to claim 4, characterized in that, Before determining the execution result related to the task execution strategy based on the directed acyclic graph, the method further includes: The directed acyclic graph is shown. In response to obtaining adjustment information for the directed acyclic graph, the directed acyclic graph is adjusted based on the adjustment information to obtain an adjusted directed acyclic graph; The step of determining the execution result related to the task execution strategy based on the directed acyclic graph includes: The data resource service associated with each operator in the adjusted directed acyclic graph is invoked to perform task processing based on the adjusted directed acyclic graph, and the execution result is obtained.

6. The method according to any one of claims 1 to 5, characterized in that, The process of using a large language model to process the input information to obtain the task execution strategy includes: The input information is processed using the large language model to determine whether the demand intent represented by the input information is ambiguous; In response to the determination that the demand intent represented by the input information is ambiguous, task demand information is generated based on the input information, wherein the demand intent represented by the task demand information is clear; The task requirement information is processed using the large language model to obtain the task execution strategy.

7. The method according to claim 6, characterized in that, The step of processing the input information using the large language model to determine whether the demand intent represented by the input information is ambiguous includes: The large language model is used to perform attribute recognition on the input information to determine the missing target attribute items in the input information and the attribute identifiers of the target attribute items, wherein the attribute identifiers are used to characterize the importance of the target attribute items; In response to determining that a target attribute item is missing in the input information, and that the attribute identifier of the target attribute item indicates that the target attribute item is a critical attribute item, it is determined that the demand intent represented by the input information is ambiguous.

8. The method according to claim 6, characterized in that, The step of generating task requirement information based on the input information includes: Interactive information is generated based on the input information and displayed. The interactive information includes query information for a target attribute item missing in the input information and at least one potential attribute value corresponding to the target attribute item. Obtain response information in response to the interaction information, the response information including the target attribute value for the target attribute item; The task requirement information is generated using the large language model based on the input information, the interaction information, and the response information.

9. The method according to claim 8, characterized in that, The step of generating interactive information based on the input information includes: Based on the missing target attribute item in the input information, query the preset annotation information to obtain prompt information. The prompt information includes the target attribute item and at least one potential attribute value corresponding to the target attribute item. The interactive information is generated based on the prompt information using the large language model.

10. An information processing device, characterized in that, include: The processing module is used to process the input information using a large language model in response to the acquisition of input information to obtain a task execution strategy. The task execution strategy includes multiple demand tasks, and the multiple demand tasks are matched with the demand intent represented by the input information. The first determining module is used to determine the operator that matches each required task in the task execution strategy; A construction module is used to construct a directed acyclic graph based on operators that match each of the said requirements tasks. The directed acyclic graph includes multiple operators and multiple directed edges, which are used to indicate the execution dependencies between the multiple operators. The second determining module is used to determine the execution result related to the task execution strategy based on the directed acyclic graph; The display module is used to generate feedback information based on the execution result and to display the feedback information.

11. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors perform the information processing method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, It stores executable instructions that, when executed by a processor, cause the processor to perform the information processing method according to any one of claims 1 to 9.

13. A computer program product, characterized in that, It includes a computer program that, when executed by a processor, implements the information processing method according to any one of claims 1 to 9.

Images