Task processing method and device

Abstract

The embodiment of the invention provides a task processing method and device.The task processing method comprises the steps that task data of a to-be-processed task and structural information corresponding to the to-be-processed task are obtained, the structural information comprises structural data and multiple pieces of knowledge data, the structural data comprises multiple nodes and the hierarchical relation between the nodes, and the multiple pieces of knowledge data correspond to the structural data and the multiple pieces of knowledge data; the nodes represent key points of the to-be-processed task; determining a target node from the plurality of nodes based on the task data; the node data and the structure data of the target node are input into a task generation model, a guide interaction task is generated, and the task generation model is a model based on deep learning; and based on an interaction result obtained by executing the guide interaction task, determining target knowledge data from the multiple pieces of knowledge data, and obtaining a task processing result of the to-be-processed task. The structure data and the knowledge data are decoupled, the structure data guides the task generation model to generate the guide interaction task, the task processing result is determined in combination with the knowledge data, and the task processing capacity is improved.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of artificial intelligence technology, and particularly to a task processing method and apparatus. Background Art

[0002] With the rapid development of the Internet, information has shown the characteristics of exponential growth and fragmentation. It has become increasingly difficult for various professionals to search for professional information on the Internet. Professionals can hardly quickly search for the required professional information, which seriously affects the efficiency of professional research and learning.

[0003] Currently, the common information acquisition methods are search engines and knowledge-based question answering systems. The knowledge-based question answering system can be a question answering system based on a knowledge base, a question answering system based on common question pairs, and a question answering system based on information search. However, these question answering systems mainly use methods such as semantic matching and keyword matching to obtain answers and provide feedback. However, the above methods not only have low processing efficiency but also low answer accuracy. Therefore, there is an urgent need for a more effective task processing method to solve the above problems. Summary of the Invention

[0004] In view of this, the embodiments of this specification provide a task processing method. One or more embodiments of this specification also relate to a task processing apparatus, a question answering processing method, a question answering processing apparatus, a computing device, a computer-readable storage medium, and a computer program to solve the technical defects existing in the prior art.

[0005] According to the first aspect of the embodiments of this specification, a task processing method is provided, including:

[0006] Obtain the task data of the task to be processed and the structured information corresponding to the task to be processed, where the structured information includes structure data and multiple knowledge data, the structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the task to be processed;

[0007] Based on the task data, determine a target node from the multiple nodes;

[0008] Input the node data of the target node and the structure data into a task generation model to generate a guided interaction task, where the task generation model is a model based on deep learning;

[0009] Based on the interaction result obtained by executing the guided interaction task, determine target knowledge data from the multiple knowledge data to obtain the task processing result of the task to be processed.

[0010] According to the second aspect of the embodiments of this specification, a task processing apparatus is provided, including:

[0011] An acquisition module, configured to acquire task data of a task to be processed and structured information corresponding to the task to be processed, wherein the structured information includes structure data and a plurality of knowledge data, the structure data includes a plurality of nodes and hierarchical relationships between the nodes, and the nodes represent key points of the task to be processed;

[0012] A determination module, configured to determine a target node from the plurality of nodes based on the task data;

[0013] A generation module, configured to input the node data of the target node and the structure data into a task generation model to generate a guiding interaction task, wherein the task generation model is a model based on deep learning;

[0014] An execution module, configured to determine target knowledge data from the plurality of knowledge data based on an interaction result obtained by executing the guiding interaction task, to obtain a task processing result of the task to be processed.

[0015] According to a third aspect of the embodiments of the present specification, there is provided a question and answer processing method, including:

[0016] In response to a question processing request submitted by a front-end user, acquiring question data of a question to be processed and structured information corresponding to the question to be processed, wherein the structured information includes structure data and a plurality of knowledge data, the structure data includes a plurality of nodes and hierarchical relationships between the nodes, and the nodes represent key points of the question to be processed;

[0017] Based on the question data, determining a target node from the plurality of nodes;

[0018] Inputting the node data of the target node and the structure data into a question and answer task generation model to generate a question and answer guiding task, wherein the question and answer task generation model is a model based on deep learning;

[0019] Based on an interaction result obtained by executing the question and answer guiding task, determining target knowledge data from the plurality of knowledge data, to obtain a question processing result of the question to be processed;

[0020] Sending the question processing result as a feedback of the question processing request to the front-end user.

[0021] According to a fourth aspect of the embodiments of the present specification, there is provided a question and answer processing device, including:

[0022] An acquisition module, configured to acquire problem data of a problem to be processed and structured information corresponding to the problem to be processed in response to a problem processing request submitted by a front-end user, where the structured information includes structure data and a plurality of knowledge data, the structure data includes a plurality of nodes and hierarchical relationships between the nodes, and the nodes represent key points of the problem to be processed;

[0023] A first determination module, configured to determine a target node from the plurality of nodes based on the problem data;

[0024] A generation module, configured to input node data of the target node and the structure data into a question-and-answer task generation model to generate a question-and-answer guidance task, where the question-and-answer task generation model is a model based on deep learning;

[0025] A second determination module, configured to determine target knowledge data from the plurality of knowledge data based on an interaction result obtained by executing the question-and-answer guidance task, and obtain a problem processing result of the problem to be processed;

[0026] A sending module, configured to send the problem processing result as a feedback of the problem processing request to the front-end user.

[0027] According to a fifth aspect of the embodiments of the present specification, there is provided a computing device, including:

[0028] A memory and a processor;

[0029] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the above task processing method are implemented.

[0030] According to a sixth aspect of the embodiments of the present specification, there is provided a computer-readable storage medium, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above task processing method are implemented.

[0031] According to a seventh aspect of the embodiments of the present specification, there is provided a computer program, where when the computer program is executed in a computer, the computer is made to execute the steps of the above task processing method.

[0032] In one embodiment of this specification, task data of a task to be processed and structured information corresponding to the task to be processed are obtained. The structured information includes structure data and multiple knowledge data. The structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent key points of the task to be processed. Based on the task data, a target node is determined from the multiple nodes. The node data of the target node and the structure data are input into a task generation model to generate a guided interaction task, where the task generation model is a model based on deep learning. Based on the interaction result obtained by executing the guided interaction task, target knowledge data is determined from the multiple knowledge data to obtain the task processing result of the task to be processed. By decoupling the structure data and the knowledge data in the structured information, the structure data guides the task generation model, and a guided interaction task is generated based on the structure data. Furthermore, the task processing result of the task to be processed is determined by combining the interaction result obtained by executing the guided interaction task and the knowledge data, which improves the task processing ability and the result accuracy at the same time. Description of the Drawings

[0033] Figure 1 is a schematic diagram of the processing process of a task processing method provided by an embodiment of this specification;

[0034] Figure 2 is a flowchart of a task processing method provided by an embodiment of this specification;

[0035] Figure 3 is a schematic diagram of the structured information of a task processing method provided by an embodiment of this specification;

[0036] Figure 4 is a flowchart of the processing process of a task processing method provided by an embodiment of this specification;

[0037] Figure 5 is a schematic diagram of the knowledge question and answer structure of a task processing method provided by an embodiment of this specification;

[0038] Figure 6 is a schematic diagram of the structure of a task processing device provided by an embodiment of this specification;

[0039] Figure 7 is a flowchart of a question and answer processing method provided by an embodiment of this specification;

[0040] Figure 8 is a schematic diagram of the structure of a question and answer processing device provided by an embodiment of this specification;

[0041] Figure 9 is a block diagram of the structure of a computing device provided by an embodiment of this specification. Detailed Embodiments

[0042] Numerous specific details are set forth in the following description to provide a thorough understanding of this specification. However, this specification can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the connotation of this specification. Therefore, this specification is not limited by the specific implementations disclosed below.

[0043] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a", "the", and "said" used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly dictates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0044] It should be understood that although the terms first, second, etc. may be used in one or more embodiments of this specification to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein can be interpreted as "when" or "while" or "in response to determining".

[0045] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant laws, regulations, and standards of relevant countries and regions, and corresponding operation entrances are provided for users to choose to authorize or reject.

[0046] In one or more embodiments of this specification, a large model refers to a deep learning model with a large number of model parameters, usually including hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than one quadrillion model parameters. A large model can also be referred to as a foundation model. Through pre-training of a large model with a large amount of unlabeled corpus, a pre-trained model with more than one billion parameters is produced. This model can adapt to a wide range of downstream tasks and has good generalization ability, such as large language models (LLMs), multi-modal pre-training models, etc.

[0047] When large models are applied in practice, only a small number of samples are needed to fine-tune the pre-trained models for application in different tasks. Large models can be widely applied in fields such as Natural Language Processing (NLP) and computer vision. Specifically, they can be applied to tasks in the field of computer vision such as Visual Question Answering (VQA), Image Caption (IC), and image generation, as well as tasks in the field of natural language processing such as text-based sentiment classification, text summary generation, and machine translation. The main application scenarios of large models include digital assistants, intelligent robots, search, online education, office software, e-commerce, intelligent design, etc.

[0048] First, the noun terms involved in one or more embodiments of this specification are explained.

[0049] Enterprise knowledge dialogue: A dialogue system based on enterprise knowledge, not limited to enterprise document knowledge

[0050] DocGraph: A form of document organization that can organize documents in a form similar to a knowledge graph after understanding them.

[0051] Agent: Refers to an entity that performs specific tasks in a computer system or network. It can be a software program, a hardware device, or a person. The role of an Agent is to complete tasks on behalf of users or other systems, such as data collection, data processing, and decision support. An Agent can be autonomous, with a certain degree of intelligence and adaptability to perform tasks in different situations.

[0052] Chain of Thought (CoT): A method of problem-solving and decision-making, also known as chained thinking or on-topic method. It emphasizes breaking down problems step by step, thinking about multiple aspects and causal relationships of problems, finding the essence of problems, and proposing reasonable solutions.

[0053] Figure 1 It is a schematic diagram of the processing process of a task processing method provided by an embodiment of this specification; as Figure 1As shown, after receiving a task to be processed, obtain the task data of the task to be processed and the DocGraph corresponding to the task to be processed, that is, structured information. The structured information includes structural data and multiple knowledge data. The structural data includes multiple nodes and the hierarchical relationships between the nodes. The nodes represent the key points of the task to be processed. The task to be processed can be a knowledge Q&A task or an enterprise knowledge dialogue task. Based on the task data, determine the target node from the multiple nodes according to the node data of each node. Input the node data of the target node and the structural data into a task generation model, and the task generation model generates a guided interaction task. The task generation model can be a model based on deep learning. Execute the guided interaction task, and based on the interaction result obtained by executing the guided interaction task, determine the target knowledge data from the multiple knowledge data, and determine the task processing result of the task to be processed based on the target knowledge data. The task generation model can be a large model with data processing capabilities. In the process of the task generation model generating the guided interaction task and subsequent execution of the guided interaction task, the agent (proxy) method is adopted to complete environment perception, planning, and action.

[0054] In summary, decouple the structural data and knowledge data in the structured information, use the structural data to guide the task generation model, generate a guided interaction task based on the structural data, and then combine the interaction result obtained by executing the guided interaction task and the knowledge data to determine the task processing result of the task to be processed, improving the task processing ability and the result accuracy at the same time.

[0055] In this specification, a task processing method is provided. This specification also relates to a task processing device, a Q&A processing method, a Q&A processing device, a computing device, and a computer-readable storage medium, which will be described in detail one by one in the following embodiments.

[0056] See Figure 2 , Figure 2 shows a flowchart of a task processing method according to an embodiment of this specification, specifically including the following steps.

[0057] Step 202: Obtain the task data of the task to be processed and the corresponding structured information of the task to be processed, where the structured information includes structural data and multiple knowledge data, the structural data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the task to be processed.

[0058] Specifically, the task to be processed can be a knowledge dialogue task, a question-and-answer task, a human-computer interaction task, or other information acquisition tasks for users to obtain information. Correspondingly, the task data is the data submitted by the user, such as question data and dialogue data, or the data triggered and submitted by the user's operation, and can also be the scenario data in the human-computer interaction scenario. The structure data is data containing multiple nodes, and there is a hierarchical relationship between the multiple nodes of the structure data, arranged and connected according to the hierarchical structure. The knowledge data is the data obtained by integrating the node data corresponding to the multiple nodes in the structure data. The structured information can be generated based on documents, historical records, and logs. The key points of the task to be processed represent the task keywords or task nodes of the task to be processed.

[0059] Based on this, receive the task to be processed, obtain the task data of the task to be processed, and the structured information matching the task to be processed. The structured information includes structure data and multiple knowledge data. The structure data includes multiple nodes and the hierarchical relationship between each node. The nodes represent the key points of the task to be processed. At least two nodes with a hierarchical relationship in the structure data form a node link. The knowledge data is the data generated based on the node link, that is, the data obtained by integrating the node data of the nodes in the node link.

[0060] In practical applications, the structure data and knowledge data included in the structured information are stored separately in a decoupled manner.

[0061] Furthermore, considering that the tasks to be processed have different types and involve different fields, when determining the structured information, it is necessary to refer to the type and field of the task to be processed. The specific implementation is as follows:

[0062] Obtain the task data of the task to be processed and the task information of the task to be processed; determine the structured information corresponding to the task to be processed according to the task information.

[0063] Specifically, the task information includes, but is not limited to, the task type information, field information, and geographical information of the task to be processed.

[0064] Based on this, obtain the task data of the task to be processed and determine the task information of the task to be processed. Determine the structured information corresponding to the task to be processed according to the task information.

[0065] Determine the structured information corresponding to the task to be processed based on the task information of the task to be processed, so that the structured information is highly matched with the task to be processed and improve the accuracy of determining the structured information.

[0066] Furthermore, considering the large variety of types of tasks to be processed and the limited quantity of structured information, when determining the structured information of a task to be processed, not only can structured information be selected from a pre-stored set of structured information, but also after determining the task to be processed, structured information matching the task to be processed can be generated. The specific implementation is as follows:

[0067] Obtain target data from a database according to the task information, and generate structured information based on the target data; or, select structured information from a pre-stored set of structured information according to the task information.

[0068] Specifically, the database can be a knowledge base constructed based on information such as publicly available information on the Internet in multiple fields and self-built information; the target data can be target documents in the form of papers, articles, etc., or data segments or Q&A data related to the task information.

[0069] Based on this, obtain target data that matches the keywords in the task information from the database according to the task information, and generate structured information based on the target data. Or, from a pre-stored set of structured information, select structured information that matches the task information according to the task information.

[0070] For example, in a knowledge Q&A scenario, the task to be processed is a knowledge Q&A task, and the task data can be the question data raised by the questioner. In the case where the task data is "how to conduct birth registration", relevant text content or documents can be searched in the database based on the task data. After determining the document that matches the task data, construct structured information with a hierarchical structure as shown in Figure 3 The structure data includes nodes such as birth registration (first level), Region A and non-Region A (second level), legitimate children and illegitimate children (third level), etc.; and knowledge data: birth registration policies for legitimate children in Region A; birth registration policies for illegitimate children in Region A; birth registration policies for legitimate children in non-Region A, and birth registration policies for illegitimate children in non-Region A.

[0071] To sum up, the structured information corresponding to the task information is determined through at least two methods, so as to obtain structured information with a relatively high degree of matching with the task information, and improve the accuracy of subsequent task processing results.

[0072] Step 204: Based on the task data, determine a target node from the multiple nodes.

[0073] Specifically, after obtaining the task data of the task to be processed and the structured information corresponding to the task to be processed, where the structured information includes structural data and multiple knowledge data, the structural data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the task to be processed, the target node can be determined from the multiple nodes based on the task data. Among them, the target node is selected from the multiple nodes in the structured information, and the node data of the target node has a high degree of matching with the task data.

[0074] Based on this, after obtaining the task data of the task to be processed and the structured information corresponding to the task to be processed, since the structured information includes structural data and multiple knowledge data, the structural data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the task to be processed, therefore, the task data can be respectively matched with the node data of the multiple nodes, and the target node with a high degree of matching with the task data can be determined among the multiple nodes.

[0075] Furthermore, considering that the structural data contains multiple nodes, and each node has corresponding node data respectively, therefore, when selecting the target node, the matching can be performed based on the node data corresponding to each node and the task data. The specific implementation is as follows:

[0076] Determine the node data of each node in the structural data; match the task data with the node data of each node, and determine the target node from the multiple nodes according to the matching result.

[0077] Specifically, the node data is the data represented by the nodes in the structural data. There are hierarchical relationships between the nodes, and there are also hierarchical relationships between the node data; the matching can be semantic matching or feature vector matching.

[0078] Based on this, determine the node data of each node in the structural data. Match the task data with the node data of each node respectively, obtain the matching degree between the task data and each node data, and determine the target node among the multiple nodes according to the matching degree.

[0079] Continuing with the above example, match the task data "how to conduct birth registration" with the node data of each node in the structural data respectively. It is determined that the node "birth registration" in the structural data has a high degree of matching with the task data, then the node "birth registration" is used as the target node.

[0080] In summary, by matching the task data with the node data of each node respectively, the target node is determined, thereby improving the determination efficiency of the target node.

[0081] Step 206: Input the node data of the target node and the structure data into a task generation model to generate a guided interaction task, where the task generation model is a deep learning-based model.

[0082] Specifically, after determining the target node from multiple nodes based on the task data, the node data and structure data of the target node can be input into the task generation model to generate a guided interaction task. The task generation model is a deep learning-based model, which can be a large model based on deep learning. The task generation model combines an agent and has the capabilities of perceiving the environment, planning, and acting. The guided interaction task is generated by the task generation model based on the node data, structure data, and task data. After perceiving the environmental information, it plans the task execution path and generates based on the task execution path. The guided interaction task is used to guide the questioner to answer questions in a knowledge Q&A scenario and determine the answer corresponding to the task data based on the answer information submitted by the questioner.

[0083] Based on this, after determining the target node from multiple nodes based on the task data, input the node data and structure data of the target node into the task generation model to generate a guided interaction task. The task data can also be input into the task generation model. After generating the initial guided interaction task, the initial guided interaction task is updated based on the task data to obtain the guided interaction task.

[0084] Furthermore, considering that there is a hierarchical relationship among the nodes in the structure data, the nodes in the structure data do not exist in isolation. When generating the guided interaction task, the hierarchical relationship among the nodes and the data association relationship under the hierarchical relationship also need to be referred to. The specific implementation is as follows:

[0085] Input the node data of the target node and the structure data into the task generation model, and use the task generation model to determine the structure link data matching the node data in the structure data; generate a guided interaction task based on the structure link data.

[0086] Specifically, the structure link data refers to the node link containing the target node determined in the structure data, and the arrangement order of the nodes in the node link matches the hierarchical relationship of the nodes in the structure data.

[0087] Based on this, input the node data and structure data of the target node into the task generation model, and use the task generation model to determine the structure link data matching the node data in the structure data. Generate a guided interaction task based on the node data of the nodes included in the structure link data and the association relationship between the nodes.

[0088] Continuing with the above example, when the task data is "How to conduct birth registration in Area A", the target node corresponding to the task data is the node "Area A" in the structure data. Then, determine the structure data link containing the node "Area A": "Birth registration" - "Area A" - "Legitimate children", and "Birth registration" - "Area A" - "Illegitimate children". Furthermore, generate a guided interaction task based on the data of these two structure links.

[0089] In summary, generate a guided interaction task based on the node data of the nodes contained in the structure link data and the association relationships between the nodes. The process of subsequent execution of the guided interaction task is the process of gradually determining the task processing result of the task to be processed.

[0090] Furthermore, considering that there are numerous nodes in the structure data and there are hierarchical relationships between the nodes, in order to simplify the guided interaction task, the generation of the guided interaction task can be implemented based on the neighboring nodes of the target node. The specific implementation is as follows:

[0091] Determine the node position information of the target node data in the structure link data; according to the node position information, determine the neighboring node data of the target node data in the structure link data; generate a guided interaction task based on the neighboring node data, where the neighboring node data is the previous node data and / or the subsequent node data of the target node data.

[0092] Specifically, the node position information includes but is not limited to the hierarchical information of the target node in the structure data; the neighboring node data is the node data corresponding to the node adjacent to the target node in the hierarchical dimension; the previous node data is the node data of the node adjacent to the target node in the front; the subsequent node data is the node data of the node adjacent to the target node in the back.

[0093] Based on this, determine the node position information of the target node data in the structure link data. According to the node position information, determine at least one neighboring node data of the target node data in the structure link data. Generate a guided interaction task based on the neighboring node data. The neighboring node data can be the previous node data of the target node data, can be the subsequent node data of the target node data, or can be the previous node data and the subsequent node data of the target node data.

[0094] Continuing with the above example, when the task data is "How to conduct birth registration in Area A", the target node can be the node "Area A" in the structure data. Then, combined with the node position information, the neighboring nodes are "Birth registration", "Legitimate children", and "Illegitimate children". Thus, a guided interaction task can be generated based on the neighboring node data.

[0095] In summary, a guiding interaction task is generated based on the data of neighboring nodes, thereby simplifying the guiding interaction task and improving the processing efficiency of the task to be processed.

[0096] Furthermore, considering that task data can exist in the historical data processed by the task processing model, therefore, after inputting the node data and structure data of the target node into the task generation model, the task generation model combines the historical data related to the node data to generate a guiding interaction task, and the specific implementation is as follows:

[0097] Input the node data and the structure data of the target node into the task generation model, and use the task generation model to determine the historical data associated with the node data; based on the historical data, generate a text processing task with the node data and the structure data, and use the text processing task as the guiding interaction task.

[0098] Specifically, the historical data can be the task data of the pre-processed task, the determined structured information, the target node, and the task processing result; the text generation task is a to-be-executed task corresponding to the task data and used to determine the result matching the task data.

[0099] Based on this, input the node data and structure data of the target node into the task generation model, use the task generation model to determine the historical data associated with the node data; generate a text processing task based on the historical data on the basis of the node data and the structure data, and use the text processing task as the guiding interaction task. This makes the influence of historical data added during the generation process of the guiding interaction task.

[0100] Continuing with the above example, in the case where the task data is "How to conduct birth registration in Area A", the target node can be the node "Area A" in the structure data. Then, use "Area A" as the node data and input it into the task generation model together with the structure data. The task generation model combines the historical task data "How to conduct birth registration for legitimate children in Area A" and the target node and task processing result corresponding to the historical task data to generate a text generation task. When the text generation task is completed, the answer data corresponding to the task data is obtained.

[0101] In summary, based on historical data, node data, and structure data, a text processing task is generated, and by referring to historical data, the credibility of the guiding interaction task is improved.

[0102] Step 208: Based on the interaction result obtained by executing the guiding interaction task, determine the target knowledge data from the multiple knowledge data to obtain the task processing result of the to-be-processed task.

[0103] Specifically, after inputting the node data and structure data of the target node into the task generation model to generate a guided interaction task, where the task generation model is a deep learning-based model, the target knowledge data can be determined from multiple knowledge data based on the interaction result obtained from executing the guided interaction task, and the task processing result of the task to be processed can be obtained. Here, the interaction result is the task execution result of the guided interaction task; the target knowledge data is the data with a relatively high matching degree with the task data in the knowledge data; correspondingly, the task processing result can be the target knowledge data.

[0104] Based on this, after inputting the node data and structure data of the target node into the task generation model to generate a guided interaction task, executing the guided interaction task, determining the target knowledge data from multiple knowledge data based on the interaction result obtained from executing the guided interaction task, using the target knowledge data as the task processing result of the task to be processed, or generating the task processing result of the task to be processed based on the target knowledge data, the purpose of simplifying the target knowledge data can be achieved.

[0105] Furthermore, considering that the purpose of executing the guided interaction task is to obtain the task processing result of the task to be processed, and the number of interaction subtasks in the guided interaction task will affect the accuracy of the task processing result, the specific implementation is as follows:

[0106] In the case where the guided interaction task includes one interaction subtask, the target knowledge data is determined from the multiple knowledge data based on the sub-interaction result obtained from executing the interaction subtask, and the task processing result of the task to be processed is obtained; in the case where the guided interaction task includes at least two interaction subtasks, based on the first interaction result obtained from executing the first interaction subtask among the at least two interaction subtasks, the second interaction subtask among the at least two interaction subtasks is updated and executed until an interaction result is obtained, and the target knowledge data is determined from the multiple knowledge data based on the interaction result, and the task processing result of the task to be processed is obtained.

[0107] Specifically, the interaction subtask is a subtask in the guided interaction task, and the execution result obtained after all subtasks in the guided interaction task are executed is the execution result of the guided interaction task; correspondingly, the sub-interaction result is the execution result of the interaction subtask.

[0108] Based on this, in the case where the guided interaction task includes an interaction subtask, the execution result of the interaction subtask is the interaction result of the guided interaction task. The target knowledge data is determined from multiple knowledge data based on the sub-interaction result obtained by executing the interaction subtask, and the task processing result of the task to be processed is obtained. In the case where the guided interaction task includes at least two interaction subtasks, the execution result obtained after the at least two interaction subtasks are executed in sequence is the interaction result of the guided interaction task. The first interaction subtask among the at least two interaction subtasks is executed to obtain the first interaction result, and the second interaction subtask among the at least two interaction subtasks is updated based on the first interaction result, and then the second interaction subtask is executed, and so on, until the interaction subtask at the end position among the at least two interaction subtasks is executed, and the interaction result can be obtained. The target knowledge data is determined from multiple knowledge data based on the interaction result, and the task processing result of the task to be processed is obtained.

[0109] Continuing with the above example, when the task data is "How to conduct birth registration in Area A", the generated guided interaction task includes the question "Whether born in wedlock" and feedback this question to the questioner. When the questioner answers "Yes", the guided interaction task is completed, and the interaction result of "Birth registration policy for legitimate children in Area A" is obtained. When the task data is "How to conduct birth registration", the generated guided interaction task includes two questions: "Whether born in wedlock" and "Area of residence". First, the question "Area of residence" is feedback to the questioner. When the questioner answers "Area A", then the question "Whether born in wedlock" is feedback to the questioner. When the questioner answers "Yes", the guided interaction task is completed, and the interaction result of "Birth registration policy for legitimate children in Area A" is obtained.

[0110] In summary, the interaction subtasks included in the guided interaction task are executed, and the target knowledge data is determined from multiple knowledge data based on the interaction result, thereby improving the matching degree between the task processing result and the task to be processed.

[0111] Furthermore, when there are two tasks in the guided interaction task, the execution of the second task depends on the execution result of the first task. Therefore, after the first task is completed, the second task can be updated based on the execution result and then executed. The specific implementation is as follows:

[0112] Determine a first guiding task and a second guiding task included in the guiding interaction task, where there is an execution sequence relationship between the first guiding task and the second guiding task; execute the first guiding task to obtain a first execution result; based on the first execution result, determine a first node from the multiple nodes, and update the second guiding task to a target guiding task based on the first node; execute the target guiding task to obtain a target execution result, and use the target execution result as the interaction result of the guiding interaction task.

[0113] Specifically, the first guiding task and the second guiding task are two adjacent subtasks, and the node level of the node corresponding to the first guiding task is higher than the node level of the node corresponding to the second guiding task; the first node is the node determined in the structure data based on the first execution result after the first guiding task is completed, and the node level of the first node is lower than the node level of the node corresponding to the first guiding task.

[0114] Based on this, determine a first guiding task and a second guiding task with an execution sequence relationship included in the guiding interaction task. Execute the first guiding task to obtain a first execution result. Based on the first execution result, determine a first node from the multiple nodes in the structure data, and the node level of the first node is lower than the node level of the node corresponding to the first guiding task. Update the second guiding task to a target guiding task based on the first node. Execute the target guiding task to obtain a target execution result. Use the target execution result as the interaction result of the guiding interaction task.

[0115] Continuing with the above example, when the task data is "how to conduct birth registration", the generated guiding interaction task includes two tasks, namely "legitimate birth" and "region of residence". Execute the first guiding task "region of residence" and feedback the question "region of residence" to the questioner. When the questioner answers "Region A", determine the next-level nodes "legitimate children" and "illegitimate children" corresponding to the node "Region A" in the result data, and update "legitimate birth" to the second guiding task "whether legitimate birth". Then feedback the question "whether legitimate birth" to the questioner. When the questioner answers "yes", the guiding interaction task is completed, and the interaction result "birth registration policy for legitimate children in Region A" is obtained.

[0116] To sum up, after updating the second guiding task and then executing it, the influence of the execution result of the first guiding task is incorporated into the execution process of the second guiding task, which plays a role in correcting the second guiding task.

[0117] Furthermore, after the guiding interaction task is completed, it is also necessary to determine target knowledge data based on the knowledge data, and the determination of the target knowledge data depends on the node data of each node in the structure data. The specific implementation is as follows:

[0118] Execute the guided interaction task to obtain an interaction result; determine a target link corresponding to the interaction result in the structure data, and determine target knowledge data from the multiple knowledge data based on the target link.

[0119] Specifically, the link nodes in the target link are node links with connection relationships in the structure data, and the hierarchical structures between the link nodes included in the target link are different.

[0120] Based on this, execute the guided interaction task to obtain an interaction result; determine a target link corresponding to the interaction result in the structure data, and determine target knowledge data from the multiple knowledge data based on the target link.

[0121] Continuing with the above example, in the case of obtaining the interaction result "Birth registration of legitimate children in Area A", determine the target link in the structure data based on the interaction result: "Birth registration" - "Area A" - "Legitimate children". And determine the target knowledge data "Birth registration policy of legitimate children in Area A" from the knowledge data based on this target link.

[0122] In summary, determine the target knowledge data from the multiple knowledge data based on the target link, thereby improving the accuracy of determining the target knowledge data.

[0123] In one embodiment of this specification, by obtaining task data of a task to be processed and structured information corresponding to the task to be processed, where the structured information includes structure data and multiple knowledge data, the structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent key points of the task to be processed; based on the task data, determine a target node from the multiple nodes; input the node data of the target node and the structure data into a task generation model to generate a guided interaction task, where the task generation model is a model based on deep learning; based on the interaction result obtained by executing the guided interaction task, determine target knowledge data from the multiple knowledge data to obtain the task processing result of the task to be processed. Decouple the structure data and knowledge data in the structured information, use the structure data to guide the task generation model, generate a guided interaction task based on the structure data, and then combine the interaction result obtained by executing the guided interaction task and the knowledge data to determine the task processing result of the task to be processed, improving the task processing ability and the result accuracy at the same time.

[0124] The following combines the attached Figure 4 , taking the application of the task processing method provided in this specification in knowledge Q&A as an example, to further illustrate the task processing method. Among them, Figure 4 Fig. shows the processing procedure flowchart of a task processing method provided in one embodiment of this specification, which specifically includes the following steps.

[0125] Step 402: Construct a structured information set. The structured information set contains multiple pieces of structured information, and each piece of structured information includes structural data and knowledge data. The structural data contains multiple nodes and the hierarchical relationships between the nodes.

[0126] Reference Figure 5 , through the knowledge construction module, complete the construction of the structured information - DocGraph. The structured information includes structural data and knowledge data, and the structural data and knowledge data are decoupled. Among them, the structural data can be a graph structure, which is responsible for guiding the large model to make plans, and the knowledge data is responsible for providing the knowledge source for generating end - answers. The structural data has clear nodes and levels. The level refers to the hierarchical information. For example, the level of the "birth registration" node in the structural data is 1.

[0127] Step 404: Obtain the task to be processed, and determine the target link data in the link data corresponding to the structural data based on the task data of the task to be processed.

[0128] The task to be processed can be a knowledge dialogue task, or an information acquisition task such as a question - answering task or a human - machine interaction task for the user to obtain information. Correspondingly, the task data is the data submitted by the user such as question data or dialogue data, or the data triggered and submitted by the user's operation, or it can also be the data in the human - machine interaction scenario.

[0129] In the case where the task data of the task to be processed is the question "How to conduct birth registration", use the agent corresponding to the large model to search in the structured information set based on this question, and determine the target link data that matches the task data.

[0130] Step 406: Determine the target structural data corresponding to the target link data.

[0131] Take the graph structure of the target link data as the target structural data.

[0132] Step 408: Determine the target node among the multiple nodes included in the target structural data based on the task data.

[0133] Step 410: Determine the environmental data based on the knowledge dialogue model.

[0134] The knowledge dialogue model is the large model. Use the Agent corresponding to the large model to perceive the environmental data (such as the current time, project - level parameters), and form memory data (multi - round context short - term memory and long - term memory of structural data and knowledge data). The environmental data can be the user's multi - round dialogue information, the system time, project - level parameter configurations, etc.

[0135] Step 412: Invoke the knowledge dialogue model to generate a guided interaction task based on the environmental data, structural data, target node, and task data.

[0136] Guiding an interaction task is a task generated based on planning. Planning includes, but is not limited to, clarification, rhetorical questioning, recommendation, guidance, search, and direct output. The Agent corresponding to the large model will use the powerful planning ability of the large model to plan according to the acquired environmental data, structural data, target nodes, and task data to determine what task will be executed. In practical applications, the next step or task to be executed can be judged through CoT decision (chain of thought).

[0137] Step 414: Determine whether the interaction result obtained based on the execution of the guiding interaction task matches the task to be processed; if so, execute Step 416; if not, execute Step 418.

[0138] Step 416: Use the interaction result as the task execution result of the task to be processed.

[0139] Step 418: Generate a target interaction task based on the interaction result and execute it until a task execution result that matches the task to be processed is obtained.

[0140] Execute the target interaction task, use the Agent corresponding to the large model to complete knowledge search, guidance, or request the large model again based on the determined rhetorical question information to obtain the task execution result. The task execution result is the answer to the question corresponding to the task data.

[0141] In summary, in one embodiment of the present specification, by obtaining the task data of the task to be processed and the structured information corresponding to the task to be processed, where the structured information includes structural data and multiple knowledge data, the structural data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the task to be processed; based on the task data, determine the target node from multiple nodes; input the node data and structural data of the target node into the task generation model to generate a guiding interaction task, where the task generation model is a model based on deep learning; based on the interaction result obtained from the execution of the guiding interaction task, determine the target knowledge data from multiple knowledge data to obtain the task processing result of the task to be processed. Decouple the structural data and knowledge data in the structured information, use the structural data to guide the task generation model, generate a guiding interaction task based on the structural data, and then combine the interaction result obtained from the execution of the guiding interaction task and the knowledge data to determine the task processing result of the task to be processed, improving the task processing ability and the result accuracy at the same time.

[0142] Corresponding to the above method embodiment, the present specification also provides an embodiment of a task processing device Figure 6 shows a schematic structural diagram of a task processing device provided by an embodiment of the present specification. As Figure 6 shown, the device includes:

[0143] An acquisition module 602, configured to acquire task data of a task to be processed and structured information corresponding to the task to be processed, where the structured information includes structure data and a plurality of knowledge data, the structure data includes a plurality of nodes and hierarchical relationships between the nodes, and the nodes represent key points of the task to be processed;

[0144] A determination module 604, configured to determine a target node from the plurality of nodes based on the task data;

[0145] A generation module 606, configured to input the node data of the target node and the structure data into a task generation model to generate a guided interaction task, where the task generation model is a model based on deep learning;

[0146] An execution module 608, configured to determine target knowledge data from the plurality of knowledge data based on an interaction result obtained by executing the guided interaction task, and obtain a task processing result of the task to be processed.

[0147] In an optional embodiment, the acquisition module 602 is further configured to:

[0148] Acquire task data of a task to be processed, and task information of the task to be processed;

[0149] Determine the structured information corresponding to the task to be processed according to the task information.

[0150] In an optional embodiment, the acquisition module 602 is further configured to:

[0151] Acquire target data in a database according to the task information, and generate structured information based on the target data; or,

[0152] Select structured information from a pre-stored set of structured information according to the task information.

[0153] In an optional embodiment, the determination module 604 is further configured to: determine node data of each node in the structure data;

[0154] Match the task data with the node data of each node, and determine a target node from the plurality of nodes according to the matching result.

[0155] In an optional embodiment, the generation module 606 is further configured to:

[0156] Input the node data of the target node and the structure data into a task generation model, and use the task generation model to determine structure link data matching the node data in the structure data;

[0157] Generate a guided interaction task based on the structural link data.

[0158] In an optional embodiment, the generating module 606 is further configured to:

[0159] Determine the node position information of the target node data in the structural link data;

[0160] Determine the neighboring node data of the target node data in the structural link data according to the node position information;

[0161] Generate a guided interaction task based on the neighboring node data, where the neighboring node data is the preceding node data and / or the succeeding node data of the target node data.

[0162] In an optional embodiment, the generating module 606 is further configured to:

[0163] Input the node data of the target node and the structure data into a task generation model, and use the task generation model to determine the historical data associated with the node data;

[0164] Generate a text processing task based on the historical data, the node data, and the structure data, and use the text processing task as the guided interaction task.

[0165] In an optional embodiment, the execution module 608 is further configured to:

[0166] In the case where the guided interaction task includes an interaction subtask, determine the target knowledge data from the multiple knowledge data based on the sub-interaction result obtained by executing the interaction subtask, and obtain the task processing result of the task to be processed;

[0167] In the case where the guided interaction task includes at least two interaction subtasks, update and execute the second interaction subtask among the at least two interaction subtasks based on the first interaction result obtained by executing the first interaction subtask among the at least two interaction subtasks until an interaction result is obtained, and determine the target knowledge data from the multiple knowledge data based on the interaction result, and obtain the task processing result of the task to be processed.

[0168] In an optional embodiment, the execution module 608 is further configured to:

[0169] Determine a first guiding task and a second guiding task included in the guided interaction task, where there is an execution sequence relationship between the first guiding task and the second guiding task;

[0170] Execute the first guiding task to obtain a first execution result;

[0171] Based on the first execution result, determine a first node from the multiple nodes, and update the second guiding task to a target guiding task based on the first node;

[0172] Execute the target guiding task to obtain a target execution result, and use the target execution result as the interaction result of the guiding interaction task.

[0173] In an optional embodiment, the execution module 608 is further configured to:

[0174] Execute the guiding interaction task to obtain an interaction result;

[0175] Determine a target link corresponding to the interaction result in the structure data, and determine target knowledge data from the multiple knowledge data based on the target link.

[0176] In summary, an embodiment of this specification obtains task data of a task to be processed and structured information corresponding to the task to be processed, where the structured information includes structure data and multiple knowledge data, the structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent key points of the task to be processed; based on the task data, determine a target node from the multiple nodes; input the node data of the target node and the structure data into a task generation model to generate a guiding interaction task, where the task generation model is a model based on deep learning; based on the interaction result obtained by executing the guiding interaction task, determine target knowledge data from the multiple knowledge data to obtain the task processing result of the task to be processed. Decouple the structure data and knowledge data in the structured information, use the structure data to guide the task generation model, generate a guiding interaction task based on the structure data, and then combine the interaction result obtained by executing the guiding interaction task and the knowledge data to determine the task processing result of the task to be processed, improving the task processing ability and the result accuracy at the same time.

[0177] The above is a schematic solution of a task processing device according to an embodiment of this specification. It should be noted that the technical solution of this task processing device and the technical solution of the above task processing method belong to the same concept. For the details not described in the technical solution of this task processing device, reference can be made to the description of the technical solution of the above task processing method.

[0178] See Figure 7 , Figure 7 shows a flowchart of a question and answer processing method according to an embodiment of this specification, which specifically includes the following steps.

[0179] Step 702: Obtain the problem data of the problem to be processed and the structured information corresponding to the problem to be processed in response to a problem processing request submitted by a front-end user. Among them, the structured information includes structure data and multiple knowledge data. The structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the problem to be processed.

[0180] Step 704: Determine a target node from the multiple nodes based on the problem data.

[0181] Step 706: Input the node data of the target node and the structure data into a question-and-answer task generation model to generate a question-and-answer guidance task, where the question-and-answer task generation model is a model based on deep learning.

[0182] Step 708: Determine target knowledge data from the multiple knowledge data based on the interaction result obtained by executing the question-and-answer guidance task, and obtain the problem processing result of the problem to be processed.

[0183] Step 710: Send the problem processing result as the feedback of the problem processing request to the front-end user.

[0184] In practical applications, after receiving a problem processing request submitted by a front-end user, determine the problem to be processed based on the problem processing request. Obtain the problem data of the problem to be processed and the structured information corresponding to the problem to be processed. The structured information includes structure data and multiple knowledge data. The structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the problem to be processed. Determine a target node from the multiple nodes based on the problem data, input the node data of the target node and the structure data into a question-and-answer task generation model to generate a question-and-answer guidance task. Determine target knowledge data from the multiple knowledge data based on the interaction result obtained by executing the question-and-answer guidance task, and obtain the problem processing result of the problem to be processed. Send the problem processing result as the feedback of the problem processing request to the front-end user. Thus, when the problem submitted by the front-end user is an enterprise knowledge problem, enterprise knowledge question and answer can be realized.

[0185] In summary, in one embodiment of this specification, task data of a task to be processed and structured information corresponding to the task to be processed are obtained. The structured information includes structure data and multiple knowledge data. The structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent key points of the task to be processed. Based on the task data, a target node is determined from the multiple nodes. The node data of the target node and the structure data are input into a task generation model to generate a guided interaction task, where the task generation model is a model based on deep learning. Based on the interaction result obtained by executing the guided interaction task, target knowledge data is determined from the multiple knowledge data to obtain the task processing result of the task to be processed. By decoupling the structure data and the knowledge data in the structured information, the structure data is used to guide the task generation model, and a guided interaction task is generated based on the structure data. Then, the task processing result of the task to be processed is determined by combining the interaction result obtained by executing the guided interaction task and the knowledge data, which improves the task processing ability and the result accuracy at the same time.

[0186] Corresponding to the above method embodiment, this specification also provides an embodiment of a question and answer processing device. Figure 8 FIG. shows a schematic structural diagram of a question and answer processing device provided in one embodiment of this specification. As Figure 8 shown, the device includes:

[0187] An obtaining module 802, configured to obtain question data of a question to be processed and the structured information corresponding to the question to be processed in response to a question processing request submitted by a front-end user, where the structured information includes structure data and multiple knowledge data, the structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent key points of the question to be processed;

[0188] A first determination module 804, configured to determine a target node from the multiple nodes based on the question data;

[0189] A generation module 806, configured to input the node data of the target node and the structure data into a question and answer task generation model to generate a question and answer guidance task, where the question and answer task generation model is a model based on deep learning;

[0190] A second determination module 808, configured to determine target knowledge data from the multiple knowledge data based on the interaction result obtained by executing the question and answer guidance task to obtain the question processing result of the question to be processed;

[0191] A sending module 810, configured to send the question processing result as a feedback of the question processing request to the front-end user.

[0192] In summary, in one embodiment of this specification, task data of a task to be processed and structured information corresponding to the task to be processed are obtained. The structured information includes structure data and multiple knowledge data. The structure data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent key points of the task to be processed. Based on the task data, a target node is determined from the multiple nodes. The node data and structure data of the target node are input into a task generation model to generate a guided interaction task, where the task generation model is a model based on deep learning. Based on the interaction result obtained by executing the guided interaction task, target knowledge data is determined from the multiple knowledge data to obtain the task processing result of the task to be processed. By decoupling the structure data and knowledge data in the structured information, the structure data guides the task generation model, and a guided interaction task is generated based on the structure data. Furthermore, the task processing result of the task to be processed is determined by combining the interaction result obtained by executing the guided interaction task and the knowledge data, improving the task processing ability and the result accuracy at the same time.

[0193] The above is a schematic solution of a question-answering processing device according to this embodiment. It should be noted that the technical solution of this question-answering processing device and the technical solution of the above question-answering processing method belong to the same concept. For the details not described in the technical solution of the question-answering processing device, reference can be made to the description of the technical solution of the above question-answering processing method.

[0194] Figure 9 The structural block diagram of a computing device 900 provided according to an embodiment of this specification is shown. The components of the computing device 900 include, but are not limited to, a memory 910 and a processor 920. The processor 920 is connected to the memory 910 through a bus 930, and a database 950 is used to store data.

[0195] The computing device 900 also includes an access device 940, which enables the computing device 900 to communicate via one or more networks 960. Examples of such networks include the Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or a combination of communication networks such as the Internet. The access device 940 may include one or more of any type of wired or wireless network interface (e.g., network interface card (NIC)), such as an IEEE 902.11 Wireless Local Area Network (WLAN) wireless interface, Worldwide Interoperability for Microwave Access (Wi-MAX) interface, Ethernet interface, Universal Serial Bus (USB) interface, cellular network interface, Bluetooth interface, Near Field Communication (NFC).

[0196] In one embodiment of the present specification, the above components of the computing device 900 and Figure 9 other components not shown therein may also be connected to each other, for example, via a bus. It should be understood that Figure 9 the block diagram of the computing device shown is for illustrative purposes only and is not a limitation on the scope of the present specification. Those skilled in the art can add or replace other components as needed.

[0197] The computing device 900 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 900 can also be a mobile or stationary server.

[0198] Among them, the processor 920 is used to execute the following computer-executable instructions, which implement the steps of the above method when executed by the processor.

[0199] The above is a schematic solution of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the above method belong to the same concept. For the details not described in detail in the technical solution of the computing device, reference can be made to the description of the technical solution of the above method.

[0200] An embodiment of this specification also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by a processor, the steps of the above method are implemented.

[0201] The above is a schematic solution of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium and the technical solution of the above method belong to the same concept. For the details not described in detail in the technical solution of the storage medium, reference can be made to the description of the technical solution of the above method.

[0202] An embodiment of this specification also provides a computer program. When the computer program is executed on a computer, the computer is made to execute the steps of the above method.

[0203] The above is a schematic solution of a computer program according to this embodiment. It should be noted that the technical solution of this computer program and the technical solution of the above method belong to the same concept. For the details not described in detail in the technical solution of the computer program, reference can be made to the description of the technical solution of the above method.

[0204] The above describes specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the figures do not necessarily require the particular order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0205] The computer instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium may be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0206] It should be noted that for the foregoing method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the embodiments of this specification are not limited by the described action sequence, because according to the embodiments of this specification, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential for the embodiments of this specification.

[0207] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0208] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The alternative embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can well understand and utilize this specification. This specification is only limited by the claims and their full scope and equivalents.

Claims

1. A task processing method, comprising: Obtaining task data of a task to be processed and structured information corresponding to the task to be processed, wherein the structured information includes structure data and a plurality of knowledge data, the structure data includes a plurality of nodes and the hierarchical relationship between each node, and the node represents a key point of the task to be processed; Determining a target node from the plurality of nodes based on the task data; Inputting the node data of the target node and the structure data into a task generation model to generate a guided interaction task, wherein the task generation model is a model based on deep learning; Determining target knowledge data from the plurality of knowledge data based on the interaction result obtained by executing the guided interaction task to obtain a task processing result of the task to be processed.

2. The task processing method according to claim 1, wherein the obtaining task data of the task to be processed and the structured information corresponding to the task to be processed comprises: Obtaining task data of the task to be processed, and task information of the task to be processed; Determining the structured information corresponding to the task to be processed according to the task information.

3. The task processing method according to claim 2, wherein the determining the structured information corresponding to the task to be processed according to the task information comprises: Obtaining target data in a database according to the task information, and generating structured information based on the target data; Or, Selecting structured information from a pre-stored set of structured information according to the task information.

4. The task processing method according to claim 1, wherein the determining a target node from the plurality of nodes based on the task data comprises: Determining the node data of each node in the structure data; Matching the task data with the node data of each node, and determining a target node from the plurality of nodes according to the matching result.

5. The task processing method according to claim 1, wherein the inputting the node data of the target node and the structure data into a task generation model to generate a guided interaction task comprises: Inputting the node data of the target node and the structure data into a task generation model, and using the task generation model to determine structure link data matching the node data in the structure data; Generating a guided interaction task based on the structure link data.

6. The task processing method according to claim 5, wherein the generating a guided interaction task based on the structure link data comprises: Determining node position information of the target node data in the structure link data; Determining neighbor node data of the target node data in the structure link data according to the node position information; Generating a guided interaction task based on the neighbor node data, wherein the neighbor node data is pre-order node data and / or post-order node data of the target node data.

7. The task processing method according to claim 6, wherein the inputting the node data of the target node and the structure data into a task generation model to generate a guided interaction task comprises: Input the node data of the target node and the structure data into the task generation model, and use the task generation model to determine the historical data associated with the node data; Generate a text processing task based on the historical data, the node data, and the structure data, and use the text processing task as the guided interaction task.

8. The task processing method according to claim 1, wherein determining target knowledge data from the multiple knowledge data based on the interaction result obtained by executing the guided interaction task to obtain the task processing result of the to-be-processed task includes: In the case where the guided interaction task includes one interaction subtask, determining target knowledge data from the multiple knowledge data based on the sub-interaction result obtained by executing the interaction subtask to obtain the task processing result of the to-be-processed task; In the case where the guided interaction task includes at least two interaction subtasks, updating and executing the second interaction subtask among the at least two interaction subtasks based on the first interaction result obtained by executing the first interaction subtask among the at least two interaction subtasks until an interaction result is obtained, and determining target knowledge data from the multiple knowledge data based on the interaction result to obtain the task processing result of the to-be-processed task.

9. The task processing method according to claim 1, wherein in the case where the guided interaction task includes a first guided task and a second guided task, the execution of the guided interaction task includes: Determine the first guided task and the second guided task included in the guided interaction task, wherein there is an execution sequence relationship between the first guided task and the second guided task; Execute the first guided task to obtain a first execution result; Based on the first execution result, determine a first node from the multiple nodes, and update the second guided task to a target guided task based on the first node; Execute the target guided task to obtain a target execution result, and use the target execution result as the interaction result of the guided interaction task.

10. The task processing method according to claim 1, wherein determining target knowledge data from the multiple knowledge data based on the interaction result obtained by executing the guided interaction task includes: Execute the guided interaction task to obtain an interaction result; Determine a target link corresponding to the interaction result in the structure data, and determine target knowledge data from the multiple knowledge data based on the target link.

11. A question and answer processing method, comprising: Obtain the question data of the to-be-processed question and the structured information corresponding to the to-be-processed question in response to a question processing request submitted by a front-end user, wherein the structured information includes structure data and multiple knowledge data, the structure data includes multiple nodes and the hierarchical relationship between each node, and the node represents the key point of the to-be-processed question; Determine a target node from the multiple nodes based on the question data; Input the node data of the target node and the structure data into a question and answer task generation model to generate a question and answer guidance task, wherein the question and answer task generation model is a model based on deep learning; Determine target knowledge data from the multiple pieces of knowledge data based on the interaction result obtained by performing the Q&A guidance task, to obtain the problem processing result of the problem to be processed; Send the problem processing result as the feedback of the problem processing request to the front-end user.

12. A task processing device, comprising: An acquisition module, configured to acquire the task data of the task to be processed and the structured information corresponding to the task to be processed, wherein the structured information includes structural data and multiple pieces of knowledge data, the structural data includes multiple nodes and the hierarchical relationships between the nodes, and the nodes represent the key points of the task to be processed; A determination module, configured to determine a target node from the multiple nodes based on the task data; A generation module, configured to input the node data of the target node and the structural data into a task generation model to generate a guided interaction task, wherein the task generation model is a model based on deep learning; An execution module, configured to determine target knowledge data from the multiple pieces of knowledge data based on the interaction result obtained by performing the guided interaction task, to obtain the task processing result of the task to be processed.

13. A computing device, comprising: A memory and a processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 11 are implemented.

14. A computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by a processor, the steps of the method according to any one of claims 1 to 11 are implemented.

Images