Interaction method, apparatus, electronic device, storage medium, and program product

Abstract

This disclosure provides interaction methods, devices, electronic devices, storage media, program products, and intelligent agents, relating to the field of artificial intelligence technology, particularly distributed technology, deep learning technology, and large model technology. The specific implementation scheme is as follows: In response to receiving interaction information from a target object, historical reference information is determined, including content belonging to the same session as the interaction information; based on the historical reference information, the interaction information is updated to obtain target interaction information; a target service node is determined from multiple service nodes in a distributed service cluster to execute the target interaction task indicated by the target interaction information; and the historical reference information and the target interaction information are sent to the target service node to utilize the target service node to execute the target interaction task and obtain the interaction result.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to distributed technology, deep learning technology and large model technology, specifically to interaction methods, devices, electronic devices, storage media, program products and intelligent agents. Background Technology

[0002] Human-Computer Interaction (HCI) refers to the process of information exchange between humans and computers. Humans can interact with computers through various channels such as voice, touchscreen handwriting, and keyboard typing. Different computer users use different expressions, styles, and languages, which increases the difficulty for computers to understand and execute interactive tasks. Summary of the Invention

[0003] This disclosure provides an interaction method, apparatus, electronic device, storage medium, program product, and intelligent agent.

[0004] According to one aspect of this disclosure, an interaction method is provided, comprising: in response to receiving interaction information of a target object, determining historical reference information, wherein the historical reference information includes content belonging to the same session as the interaction information; updating the interaction information based on the historical reference information to obtain target interaction information; determining a target service node from multiple service nodes of a distributed service cluster for executing a target interaction task indicated by the target interaction information; and sending the historical reference information and the target interaction information to the target service node to execute the target interaction task using the target service node to obtain an interaction result.

[0005] According to another aspect of this disclosure, an interaction apparatus is provided, comprising: a reference determination module, configured to determine historical reference information in response to receiving interaction information from a target object, wherein the historical reference information includes content belonging to the same session as the interaction information; an update module, configured to update the interaction information based on the historical reference information to obtain target interaction information; a node determination module, configured to determine a target service node from multiple service nodes in a distributed service cluster for executing a target interaction task indicated by the target interaction information; and a sending module, configured to send the historical reference information and the target interaction information to the target service node to execute the target interaction task using the target service node to obtain an interaction result.

[0006] According to another aspect of this disclosure, an intelligent agent is provided, comprising: an input module for receiving input information; a processing module for determining a target task based on the input information received by the input module, determining a large model based on the target task, and obtaining output information by calling the large model to execute the method described above; and an output module for outputting the output information obtained by the processing module.

[0007] According to another aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.

[0008] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause the computer to perform the method described above.

[0009] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method described above.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0011] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0012] Figure 1 This illustration schematically shows an exemplary system architecture to which interactive methods and apparatus can be applied according to embodiments of the present disclosure;

[0013] Figure 2 A flowchart illustrating an interaction method according to an embodiment of the present disclosure is shown schematically;

[0014] Figure 3A The illustration shows a schematic diagram of a process for determining a target document according to an embodiment of the present disclosure;

[0015] Figure 3B This illustration schematically shows a flowchart of determining target context information according to an embodiment of the present disclosure;

[0016] Figure 4 A timing diagram illustrating an interaction method according to an embodiment of the present disclosure is shown schematically;

[0017] Figure 5A block diagram of an interactive device according to an embodiment of the present disclosure is shown schematically;

[0018] Figure 6 A schematic diagram illustrating the structure of an intelligent agent according to embodiments of the present disclosure is shown; and

[0019] Figure 7 A block diagram of an electronic device suitable for implementing an interaction method according to an embodiment of the present disclosure is shown schematically. Detailed Implementation

[0020] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0021] According to embodiments of this disclosure, an interaction method is provided, comprising: in response to receiving interaction information of a target object, determining historical reference information, wherein the historical reference information includes content belonging to the same session as the interaction information; updating the interaction information based on the historical reference information to obtain target interaction information; determining a target service node from multiple service nodes of a distributed service cluster for executing a target interaction task indicated by the target interaction information; and sending the historical reference information and the target interaction information to the target service node to utilize the target service node to execute the target interaction task and obtain an interaction result.

[0022] By utilizing the distributed service cluster provided in this embodiment, different interactive tasks can be executed through multiple service nodes, thereby expanding the application scope of the interactive method. Furthermore, by using a routing server to determine historical reference information matching the interactive information, and updating the interactive information with clear intent and semantics based on this historical reference information, the target interactive task determined based on the updated target interactive information becomes accurate and effective. This improves the effectiveness and accuracy of routing the target interactive information to the target service node, enhancing the user's interactive experience.

[0023] Figure 1 The illustration schematically depicts an exemplary system architecture to which interactive methods and apparatus can be applied according to embodiments of the present disclosure.

[0024] It is important to note that Figure 1The examples shown are merely examples of system architectures that can be applied to embodiments of this disclosure, intended to help those skilled in the art understand the technical content of this disclosure. However, they do not imply that embodiments of this disclosure cannot be used in other devices, systems, environments, or scenarios. For example, in another embodiment, an exemplary system architecture to which interactive methods and devices can be applied may include a terminal device, but the terminal device may implement the interactive methods and devices provided by embodiments of this disclosure without interacting with a server.

[0025] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a routing server 104, and a distributed service cluster 105. The network serves as a medium for providing communication links between the terminal devices 101, 102, 103, the routing server 104, and the distributed service cluster 105. The network may include various connection types, such as wired and / or wireless communication links, etc.

[0026] Users can use terminal devices 101, 102, and 103 to interact with the routing server 104 via the network to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (for example only).

[0027] Terminal devices 101, 102, and 103 can be various electronic devices with displays and web browsing capabilities, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0028] Routing server 104 can be a server that provides various services, such as a backend management server that supports the interactive information input by users using terminal devices 101, 102, and 103 (for example only). The backend management server can analyze and update the received interactive information, and based on the updated target interactive information, determine the target service node from the distributed service cluster 105, and send historical reference information and target interactive information to the target service node so that the target service node can execute the target interactive task indicated by the target interactive information.

[0029] The Distributed Service Cluster 105 can be a cloud server, also known as a cloud computing server or cloud host. It is a host product in the cloud computing service system, which solves the defects of traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short) in terms of high management difficulty and weak business scalability.

[0030] The distributed service cluster 105 can include multiple service nodes, each of which is used to perform different interactive tasks.

[0031] It should be noted that the interaction method provided in this embodiment can generally be executed by the routing server 104. Accordingly, the interaction device provided in this embodiment can also be set in the routing server 104.

[0032] It should be understood that Figure 1 The number of terminal devices, networks, routing servers, and distributed service clusters shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, routing servers, and distributed service clusters can be included.

[0033] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision, disclosure, and application of any type of information, such as user personal information, comply with the provisions of relevant laws and regulations, necessary confidentiality measures have been taken, and they do not violate public order and good morals.

[0034] In the technical solution disclosed herein, the user's authorization or consent is obtained before acquiring or collecting the user's personal information.

[0035] It should be noted that the sequence numbers of the operations in the following methods are for descriptive purposes only and should not be considered 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.

[0036] Figure 2 A flowchart illustrating an interaction method according to an embodiment of this disclosure is shown schematically.

[0037] like Figure 2 As shown, the method includes operations S210~S240.

[0038] In operation S210, in response to receiving interaction information from the target object, historical reference information is determined.

[0039] During operation S220, the interaction information is updated based on historical reference information to obtain the target interaction information.

[0040] In operation S230, the target service node for executing the target interaction task indicated by the target interaction information is determined from multiple service nodes in the distributed service cluster.

[0041] In operation S240, historical reference information and target interaction information are sent to the target service node so that the target service node can be used to execute the target interaction task and obtain the interaction result.

[0042] In human-computer interaction, the target object can include the user. The target object can input interactive information through a terminal device. There are no restrictions on the type of interactive information; for example, it can include text, images, video, and voice.

[0043] The routing server, acting as the execution entity, serves as a communication bridge between terminal devices and the distributed service cluster. Terminal devices can send interaction information from a target object to the routing server. Upon receiving the interaction information, the routing server determines historical reference information. Historical reference information can include content belonging to the same session as the interaction information. For example, it can include documents, images, videos, and other data uploaded by the target object during the same session, but it is not limited to this; it can also include historical context information entered by the target object in historical time periods. Any information belonging to the same session as the interaction information is acceptable.

[0044] Because the semantics of dialogue are coherent in a multi-turn conversation, historical reference information belonging to the same session can be used to update the interaction information, resulting in target interaction information with clear semantics and intent.

[0045] For example, during the first round of interaction, the target user uploaded the document "2026 **Regulations.pdf". In the third round of interaction, the interaction information included "What changes have been made?". Based on the document as historical reference information, the interaction information can be supplemented to obtain the target interaction information: "What are the changes in the 2026 **Regulations compared to the previous version of the **Regulations?".

[0046] Therefore, when the target object's input content is incomplete or the interactive information is ambiguous, historical reference information can be used to obtain target interactive information with clear intent and semantics. This allows us to determine that the target interactive task indicated by the target interactive information is the content in the analysis document. Based on the target interactive information, the target service node used to execute the target interactive task is then determined.

[0047] According to embodiments of this disclosure, a distributed service cluster can be used to execute different interactive tasks through multiple service nodes, thereby expanding the application scope of the interactive method. Furthermore, by utilizing a routing server to determine historical reference information matching the interactive information, and updating the interactive information with clear intent and semantics based on this historical reference information, the target interactive task determined based on the updated target interactive information becomes accurate and effective. This improves the effectiveness and accuracy of routing the target interactive information to the target service node, enhancing the user's interactive experience.

[0048] According to embodiments of this disclosure, historical reference information may include at least one of the following: target context information of the target document and interaction information entered by the target object during the session.

[0049] There are no restrictions on the type of target document. For example, it can be a webpage, image, spreadsheet, document, etc., as long as it is a file uploaded by the target object through the interactive interface during the session.

[0050] The target context information may include historical interaction information input by the target object during the session.

[0051] Different methods can be used to determine different types of historical reference information. For example, data uploaded by the target during the session can be collected and summarized into a document set. Historical interaction information entered by the target during the session can be collected and summarized into a session context information set.

[0052] Based on the interaction information in the current interaction round, target documents matching the interaction information are determined from the document set. Target context information matching the interaction information is also determined from the conversation context set. However, this is not limited to this. Alternatively, if the conversation context set is empty, target documents matching the interaction information can be determined only from the document set, or if the document set is empty, target context information matching the interaction information can be determined only from the conversation context set.

[0053] According to embodiments of this disclosure, with the continuous development of artificial intelligence, and based on the actual intelligence of the interaction, different types of content can be used as historical reference information, thereby improving the comprehensiveness and richness of historical reference information. Furthermore, collecting different conversational interaction information into different sets according to information type can improve the efficiency and effectiveness of historical reference information filtering, thereby improving the efficiency and effectiveness of target interaction information.

[0054] The following sections will explain how to determine the target document and the information above it.

[0055] According to embodiments of this disclosure, determining a target document that matches interactive information from a document set may include: using keyword matching to determine a first document from the document set that matches keywords of the interactive information; using vector matching to determine a second document from the document set that matches the semantic vector of the interactive information; and determining the target document based on the first and second documents.

[0056] Figure 3A The illustration shows a schematic diagram of a process for determining a target document according to an embodiment of the present disclosure.

[0057] like Figure 3A As shown, keywords can be extracted from interactive information to obtain at least one keyword. Furthermore, vector transformation can be performed on the interactive information to obtain a semantic vector.

[0058] You can perform vector transformations on the documents in a document set to obtain a document vector set.

[0059] like Figure 3A As shown, keyword matching methods, such as the KMP algorithm (an improved string matching algorithm), can be used to match each document or document fragment in the document set with the keywords in the interactive information to obtain the keyword matching results for each document. The top K1 (Top-K1) documents with the highest keyword matching degree indicated by the keyword matching results are taken as the first document.

[0060] Vector matching methods, such as Euclidean distance, can be used to match the semantic vectors of each document or document fragment in the document set with the semantic vectors of the interaction information to obtain vector matching results. The top K2 (Top-K2) documents with the highest vector matching degree indicated by the vector matching results are used as the second document.

[0061] like Figure 3A As shown, the first and second documents are deduplicated, and then sorted in descending order of matching degree. The top N (Top-N) documents or document fragments are used as the target documents.

[0062] According to embodiments of this disclosure, a resource-aware hybrid retrieval mechanism is implemented to locate information within a full range of related resources through keyword matching and vector matching. Keyword matching accurately matches first documents related to technical terms in the interactive information, while vector matching captures second documents semantically related to the interactive information. This improves the search capability of the document set and avoids information omissions. Furthermore, documents are sorted in descending order of relevance, effectively supporting comprehensive cross-document queries and comparisons, and improving filtering effectiveness.

[0063] According to embodiments of this disclosure, determining target context information that matches the interaction information from a set of contextual information may include: if the amount of information in the set of contextual information is less than an information amount threshold, using the set of contextual information as the target context information; if the amount of information in the set of contextual information is greater than or equal to the information amount threshold, determining the target context information based on the summary information of the set of contextual information and the historical context information that semantically matches the interaction information.

[0064] The summary information is determined by semantic extraction of the information set in the context of the conversation.

[0065] Figure 3B The schematic diagram illustrates a process for determining target context information according to an embodiment of the present disclosure.

[0066] like Figure 3B As shown, it can be determined whether the information content of the preceding information set in the session is less than the information content threshold.

[0067] Information volume can refer to the number of texts in the preceding information set of a conversation, but it is not limited to this; it can also refer to the length of the information sequence. A corresponding information volume threshold can be set, and the processing mode can be automatically switched based on the preset information volume threshold.

[0068] like Figure 3B As shown, when the amount of information in the conversation context is less than the information threshold, a full-content mode can be used, treating the entire conversation context as the target context information. This maintains the highest degree of dialogue coherence and semantic consistency.

[0069] like Figure 3B As shown, when the information content of the conversation context is greater than or equal to the information content threshold, a summary plus retrieval mode can be adopted. A state reduction mechanism can be used to semantically extract the conversation context information, thereby obtaining summary information representing the global semantics. This preserves reached consensus, user preferences, and key facts. Furthermore, vector matching can be used to retrieve the most relevant local historical interaction fragments from the conversation context information, obtaining historical context information.

[0070] like Figure 3B As shown, historical context and summary information can be used together as target context information.

[0071] According to embodiments of this disclosure, different processing modes are employed to determine the target context information based on the amount of information in the conversation context set, thereby improving the flexibility and intelligence of information extraction. Furthermore, by combining global summaries with local details, the problem of information dispersion caused by long texts is solved while ensuring that key information is not lost, thereby improving the effectiveness of the target context information as reference information.

[0072] The above text explained how to determine historical reference information. The following text will provide a detailed explanation of how to use historical reference information to update interactive information.

[0073] According to embodiments of this disclosure, for example, Figure 2 The operation S220 shown updates the interaction information based on historical reference information to obtain target interaction information, and may include at least one of the following: updating predetermined entity information in the interaction information based on reference entity information in the historical reference information; or rewriting predetermined interaction sub-information of the interaction information based on historical reference information.

[0074] In one embodiment, the predetermined entity information is determined by entity recognition of the interaction information.

[0075] For example, the interactive information is segmented into multiple fields. Entity recognition is performed on each field to obtain the entity recognition result for each field. If the entity recognition result indicates that the field belongs to a predetermined type of entity, then that field is used as the predetermined entity information.

[0076] Predefined type entities can include entities with ambiguous referents, such as "it" or "that part".

[0077] The reference entity information in the historical reference information can be used to update the predetermined entity information to obtain a clearly defined entity and complete the referencing resolution.

[0078] For example, replace "it" or "that part" with the reference entity information "2026*** regulations" from the historical reference information.

[0079] According to embodiments of this disclosure, the interaction task can be updated from a "regular dialogue task" indicated by "interaction information" to a "document-specific answer task" indicated by "target interaction information." This improves the accuracy of target interaction task identification and enhances the effectiveness and accuracy of subsequent target service node selection.

[0080] According to embodiments of this disclosure, before updating predetermined entity information in the interactive information, the interaction method may further include: if it is determined that predetermined entity information exists in the interactive information, determining a reference text matching the interactive information from historical reference information. The reference entity information is then determined from the reference text.

[0081] Optionally, reference entity information may include entity or document names that have actual meaning.

[0082] Historical reference information may include, but is not limited to, target context information determined from the target object's session context information set; it may also include target documents uploaded by the target object.

[0083] Based on the interactive information, historical reference information can be further semantically filtered to obtain reference text that matches the interactive information.

[0084] Taking historical reference information, including the target context, as an example, the target context information can include the questions and answers from the entire dialogue round. Reference text related to the entity information can be extracted from this context, and then the reference entity information can be determined from it.

[0085] Taking historical reference information, including the target document, as an example, we can extract the target document's summary information through semantic extraction, which serves as reference text. From the reference text, we can identify reference entity information such as the document name or document section.

[0086] According to embodiments of this disclosure, when it is determined that predetermined entity information exists in the interactive information, a preliminary screening can be performed on historical reference information based on the entity type to obtain reference text, thereby narrowing the scope, and then further screening reference entity information from the reference text, thereby improving the accuracy and effectiveness of screening reference entity information.

[0087] In another embodiment, the predetermined interaction sub-information can be determined by semantic recognition of the interaction information.

[0088] For example, semantic recognition is performed on the interactive information to determine whether the semantics of the interaction are complete. If the semantic recognition result indicates that the semantics of the interactive information are incomplete, predetermined interactive sub-information is determined from the interactive information and rewritten to clarify the interaction intent of the interactive information and avoid incorrect allocation of target service nodes.

[0089] For example, pre-defined interactive sub-information can refer to semantically incomplete content, such as "what changes have occurred".

[0090] If the predetermined interaction sub-information is determined to exist within the interaction information, a target document matching the interaction information is identified from historical reference information. The predetermined interaction sub-information is then rewritten using the document summary of the target document.

[0091] The target document is input by the target object during the session. The document summary can be determined by semantic extraction of the target document.

[0092] Based on historical reference information, the target document that the target object has uploaded can be determined. Based on the document summary of the target document, the semantically incomplete pre-defined interactive sub-information "What changes have occurred?" can be rewritten as "What changes have occurred to the ** clause mentioned in the Regulations X?".

[0093] According to embodiments of this disclosure, semantically incomplete content in interactive information can be rewritten using the document summary of the target document, thereby correcting "ordinary chat" to "document query" and avoiding problems such as misjudgment of intent or inaccurate routing of target service nodes.

[0094] The above section explained how to update the interactive information. The following section will explain how to route the target service node based on the updated target interactive information.

[0095] Figure 4 A timing diagram illustrating an interaction method according to an embodiment of the present disclosure is shown schematically.

[0096] like Figure 4As shown, in response to receiving interactive information about the target object input sent by the terminal device, the routing server asynchronously performs document retrieval and information retrieval from the document set and the session context set respectively to determine the target document and target context information as historical reference information.

[0097] Based on the terminal device's identification information, the system can asynchronously determine the set of documents and the set of conversation context information that match the identification information from the document database and the session database, respectively. This achieves data isolation between different sessions in the database.

[0098] like Figure 4 As shown, the routing server updates the interaction information based on historical reference information to obtain the target interaction information.

[0099] like Figure 4 As shown, the target service node for executing the target interaction task indicated by the target interaction information is determined from multiple service nodes in the distributed service cluster.

[0100] like Figure 4 As shown, historical reference information and target interaction information are sent to the target service node.

[0101] In one embodiment, the target service node executes the target interaction task to obtain the interaction result. Historical reference information and target interaction information can be sent to the target service node together, thereby solving the problem of dialogue gaps caused by the inability of different service nodes to share different rounds of dialogue in a distributed deployment environment.

[0102] like Figure 4 As shown, the routing server receives the content of the interaction results in a streaming manner and feeds it back to the terminal device in a streaming manner until it has fully received the information.

[0103] In one embodiment, the target service node executes the target interaction task and streams feedback information. The feedback information is a portion of the interaction result.

[0104] Compared to interaction methods that wait for the entire target interaction task to be completed before returning the interaction result, the streaming output method can respond quickly and provide timely feedback to the target object on the terminal device, thereby improving the interaction response efficiency of the terminal device.

[0105] However, if the transparent streaming output method cannot monitor whether the execution is completed and archive it in a timely manner when it is confirmed to be completed, there is still a conflict between the feedback information streaming output and the timely recording of the conversation context information.

[0106] To solve this problem, such as Figure 4As shown, once it is confirmed that the interaction result has been fully received, the target interaction information and the interaction result can be updated to the session context information set in the session database.

[0107] The routing server can monitor the task execution status of multiple service nodes. Once it is determined that the interaction result of the target interaction task of the target service node has been fully received, the target interaction information and interaction result can be updated to the session context information set in a timely manner.

[0108] Compared to the method where the target service node establishes a communication connection with the terminal device and directly sends feedback information to the terminal device in a streaming manner, the method where the routing server sends feedback information to the terminal device can promptly determine whether the target interaction task has been completed and promptly archive the session information, thereby solving the problem of not being able to simultaneously achieve "streaming output" and "timely and automatic archiving of intermediate round session information".

[0109] Optionally, the target service node can directly stream feedback information to the terminal device, and after the interaction result has been sent, the target service node can send an end signal to the routing server. This allows the routing server to accurately determine that the interaction result has been sent.

[0110] Optionally, updating the target interaction information and results to the conversation context set in the conversation database can refer to storing the target interaction information and results in the conversation context set. The storage method can include taking a binary snapshot of the target interaction information and results to perform incremental merging, rather than a simple full overwrite. This enables reliable information traceability, allowing users to backtrack information from any round of dialogue in the conversation database, facilitating online troubleshooting.

[0111] According to embodiments of this disclosure, for example, Figure 2 The operation S230 shown, which determines the target service node from multiple service nodes in the distributed service cluster to perform the target interaction task indicated by the target interaction information, may include: determining the target service node from multiple service nodes based on the task type of the target interaction task and the tool attribute information of the tools deployed by each of the multiple service nodes.

[0112] Tool attribute information can include the task types of interactive tasks that the tool can perform.

[0113] The routing server can store the tool attribute information of the tools deployed on each service node. This tool attribute information is then matched with the task type of the target interaction task to determine the target service node from multiple service nodes. This improves the routing efficiency and accuracy of the target service node.

[0114] Optionally, when a service node updates its tools, it can send new tool attribute information to the routing server for timely updates, thereby improving the effectiveness and accuracy of the service node's tool attribute information.

[0115] Optionally, the tool may include one or more of the following: a large model, a search engine, a database, an agent, etc. Any tool capable of performing interactive tasks is acceptable.

[0116] Large models can include large language models, large visual models, or large multimodal models, without any limitation.

[0117] According to a preferred embodiment of this disclosure, multiple service nodes are used to perform the same target interaction task. In this case, a target service node can be determined from the multiple service nodes based on the task type of the target interaction task and the tool attribute information of the tools deployed by each of the multiple service nodes. This includes: determining multiple candidate service nodes from the multiple service nodes based on the task type of the target interaction task and the tool attribute information of the tools deployed by each of the multiple service nodes; and determining the target service node from the multiple candidate service nodes based on the load information of each candidate service node.

[0118] Based on the task type and tool attribute information of the target interaction task, a preliminary screening of service nodes in the distributed service cluster can be performed to obtain multiple candidate service nodes capable of executing the target interaction task. Then, based on the load information of each candidate service node, the candidate service node with the lowest load is selected from the multiple candidate service nodes as the target service node.

[0119] Optionally, the load information may include current load information, determined by the target interactive task being processed or pending, and may also include the maximum load information of each service node, which may be determined based on the hardware resources of the service node.

[0120] The remaining load information can be obtained by subtracting the maximum load information from the current load information. The candidate service node with the largest remaining load information is then selected as the target service node.

[0121] The routing server can count the number of interactive tasks sent to each service node, thereby determining the current load information of each service node. Based on this load information, the target service node with the highest processing efficiency and response speed can be selected from the candidate service nodes, thus improving interaction efficiency and user experience.

[0122] Optionally, a single service node can be configured with different agents as tools, each agent performing different interaction tasks. When a service node performs interaction tasks using multiple threads, the load information of a single agent can be statistically analyzed. Based on this load information, the candidate service node with the lowest load from multiple candidate service nodes can be selected as the target service node. This achieves fine-grained load balancing.

[0123] Figure 5 A block diagram of an interactive device according to an embodiment of the present disclosure is shown schematically.

[0124] like Figure 5 As shown, the interactive device 500 includes: a reference determination module 510, an update module 520, a node determination module 530, and a sending module 540.

[0125] The reference determination module 510 is used to determine historical reference information in response to receiving interaction information from the target object, wherein the historical reference information includes content belonging to the same session as the interaction information.

[0126] The update module 520 is used to update the interaction information based on historical reference information to obtain the target interaction information.

[0127] The node determination module 530 is used to determine the target service node from multiple service nodes in the distributed service cluster for executing the target interaction task indicated by the target interaction information.

[0128] The sending module 540 is used to send historical reference information and target interaction information to the target service node so that the target service node can perform the target interaction task and obtain the interaction result.

[0129] According to embodiments of this disclosure, the update module includes at least one of the following sub-modules: an entity update sub-module and a rewrite sub-module.

[0130] The entity update submodule is used to update the predetermined entity information in the interaction information based on the reference entity information in the historical reference information. The predetermined entity information is determined by entity recognition of the interaction information.

[0131] The rewrite submodule is used to rewrite the predetermined interactive sub-information of the interactive information based on historical reference information. The predetermined interactive sub-information is determined by semantic recognition of the interactive information.

[0132] According to embodiments of this disclosure, the rewriting submodule includes a document determination unit and a rewriting unit.

[0133] The document determination unit is used to determine a target document that matches the interaction information from historical reference information when a predetermined interaction sub-information is determined to exist in the interaction information, wherein the target document is input by the target object during the session.

[0134] The rewriting unit is used to rewrite predetermined interactive sub-information using the document summary of the target document, wherein the document summary is determined by semantic extraction of the target document.

[0135] According to embodiments of this disclosure, the interactive device further includes a text determination module and an entity determination module.

[0136] The text determination module is used to determine the reference text that matches the interaction information from historical reference information when the interaction information contains predetermined entity information.

[0137] The entity determination module is used to determine reference entity information from the reference text.

[0138] According to embodiments of this disclosure, historical reference information includes at least one of the following: target context information of the target document and interaction information entered by the target object during the session.

[0139] According to embodiments of this disclosure, the reference determination module includes at least one of the following sub-modules: a document determination sub-module and an information determination sub-module.

[0140] The document identification submodule is used to identify target documents from the document set that match the interactive information.

[0141] The Information Determination submodule is used to determine the target context information that matches the interactive information from the context information set of the conversation.

[0142] According to embodiments of this disclosure, the document determination submodule includes: a keyword matching unit, a vector matching unit, and a combination unit.

[0143] The keyword matching unit is used to determine the first document in the document set that matches the keywords of the interactive information using keyword matching methods.

[0144] The vector matching unit is used to determine a second document from the document set that semantically matches the interactive information vector using vector matching.

[0145] The combination unit is used to determine the target document based on the first document and the second document.

[0146] According to embodiments of this disclosure, the information determination submodule includes: a first information determination unit and a second information determination unit.

[0147] The first information determination unit is used to take the conversation context information set as the target context information when the amount of information in the conversation context information set is less than the information amount threshold.

[0148] The second information determination unit is used to determine the target context information based on the summary information of the context information set and the historical context information that semantically matches the interaction information when the information content of the context information set is greater than or equal to the information content threshold. The summary information is determined by semantic extraction of the context information set.

[0149] According to embodiments of this disclosure, the interactive device further includes a streaming receiving module and a collection update module.

[0150] The streaming receiving module is used to receive feedback information streamed from the target service node, where the feedback information is a portion of the interaction result.

[0151] The set update module is used to update the target interaction information and interaction results to the conversation context set after confirming that the interaction results have been fully received.

[0152] According to embodiments of this disclosure, the node determination module includes a node determination submodule.

[0153] The node determination submodule is used to determine the target service node from multiple service nodes based on the task type of the target interactive task and the tool attribute information of the tools deployed on each of the multiple service nodes. The tool attribute information includes the task type of the interactive task that the tool can execute.

[0154] According to embodiments of this disclosure, the node determination submodule includes: a first node determination unit and a second node determination unit.

[0155] The first node determination unit is used to determine multiple candidate service nodes from multiple service nodes based on the task type of the target interaction task and the tool attribute information of the tools deployed by each of the multiple service nodes.

[0156] The second node determination unit is used to determine the target service node from multiple candidate service nodes based on the load information of each candidate service node.

[0157] Figure 6 A schematic block diagram of a smart agent according to an embodiment of the present disclosure is shown.

[0158] In embodiments of this disclosure, such as Figure 6 As shown, the intelligent agent 600 may include an input module 610, a processing module 620, and an output module 630.

[0159] Input module 610 is used to receive input information.

[0160] The processing module 620 is used to determine the target task based on the input information received by the input module, determine the large model based on the target task, and obtain output information by calling the large model to execute the interaction method provided according to the embodiments of this disclosure.

[0161] Output module 630 is used to output the output information obtained by the processing module.

[0162] According to embodiments of this disclosure, the input module 610 is responsible for receiving or sensing information such as queries, requests, instructions, signals, or data from the outside world (e.g., users or the external environment), and converting it into a format that the intelligent agent 600 can understand and process. The input module 610 is the primary link for the intelligent agent 600 to interact with the outside world, enabling the intelligent agent 600 to efficiently and accurately obtain necessary "sensory" information from the outside world and respond to this information.

[0163] In the example, input module 610 can input the interactive information described above.

[0164] In the example, processing module 620 is the core support for the ability of agent 600 to handle complex tasks. Processing module 620 can execute the interaction methods described above.

[0165] In the example, the performance of processing module 620 is closely related to the large model on which agent 600 is based. To fully leverage the capabilities of the large model, the internal structure of processing module 620 can be designed to be highly configurable and scalable to handle various types of tasks and requirements in real-world scenarios.

[0166] In the example, after the agent 600 acquires the interaction information, the processing module 620 can use the large model to determine the historical reference information, update the interaction information based on the historical reference information, obtain the target interaction information, determine the target service node from multiple service nodes in the distributed service cluster to execute the target interaction task indicated by the target interaction information, and pass the historical reference information, the target interaction information, and the node identification information corresponding to the target service node to the output module 630.

[0167] Understandably, while large models possess excellent language understanding and generation capabilities, like humans, their ability to solve tasks is limited without the aid of any tools. When Agent 600 is given the ability to invoke tools, it can perform tasks such as using a calculator to perform mathematical calculations, using Python to perform data analysis, and using a search engine to create weather forecasts.

[0168] In the example, the output module 630 can send historical reference information and target interaction information to the target service node based on the node identification information corresponding to the target service node, so as to use the target service node to execute the target interaction task and obtain the interaction result.

[0169] The intelligent agent 600 according to the embodiments of this disclosure can simply and effectively improve the level of intelligence, and enhance flexibility and versatility.

[0170] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0171] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.

[0172] According to embodiments of the present disclosure, a non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions are used to cause a computer to perform the method described above.

[0173] According to an embodiment of this disclosure, a computer program product includes a computer program that, when executed by a processor, implements the method described above.

[0174] Figure 7 A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0175] like Figure 7 As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded into random access memory (RAM) 703 from storage unit 708. The RAM 703 may also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0176] Multiple components in device 700 are connected to input / output (I / O) interface 705, including: input unit 706, such as a keyboard, mouse, etc.; output unit 707, such as various types of displays, speakers, etc.; storage unit 708, such as a disk, optical disk, etc.; and communication unit 709, such as a network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0177] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as interactive methods. For example, in some embodiments, the interactive method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the interactive method described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform interactive methods by any other suitable means (e.g., by means of firmware).

[0178] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0179] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0180] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0181] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0182] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0183] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, distributed system servers, or servers incorporating blockchain technology.

[0184] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0185] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. An interaction method, comprising: In response to receiving interaction information from a target object, historical reference information is determined, wherein the historical reference information includes content belonging to the same session as the interaction information; Based on the historical reference information, the interaction information is updated to obtain the target interaction information; From multiple service nodes in a distributed service cluster, determine the target service node for executing the target interaction task indicated by the target interaction information; and The historical reference information and the target interaction information are sent to the target service node so that the target service node can be used to execute the target interaction task and obtain the interaction result.

2. The method according to claim 1, wherein, The step of updating the interaction information based on the historical reference information to obtain the target interaction information includes at least one of the following: Based on the reference entity information in the historical reference information, the predetermined entity information in the interaction information is updated, wherein the predetermined entity information is determined by entity recognition of the interaction information; and Based on the historical reference information, the predetermined interaction sub-information of the interaction information is rewritten, wherein the predetermined interaction sub-information is determined by semantic recognition of the interaction information.

3. The method according to claim 2, wherein, The step of rewriting the predetermined interaction sub-information of the interaction information based on the historical reference information includes: If it is determined that the predetermined interaction sub-information exists in the interaction information, a target document matching the interaction information is determined from the historical reference information, wherein the target document is input by the target object during the session; and The predetermined interactive sub-information is rewritten using the document summary of the target document, wherein the document summary is determined by semantic extraction of the target document.

4. The method according to claim 2 or 3, further comprising: If it is determined that the predetermined entity information exists in the interaction information, a reference text matching the interaction information is determined from the historical reference information; as well as The reference entity information is determined from the reference text.

5. The method according to any one of claims 1 to 4, wherein, The historical reference information includes at least one of the following: the target document entered by the target object during the session and the target context information of the interaction information; The determination of historical reference information includes at least one of the following: Identify the target document from the document set that matches the interaction information; as well as The target context information that matches the interaction information is determined from the context information set of the conversation.

6. The method according to claim 5, wherein, The step of determining the target document that matches the interaction information from the document set includes: Using keyword matching, determine the first document from the document set that matches the keywords of the interactive information; Using vector matching, a second document that semantically matches the interaction information vector is determined from the document set; and The target document is determined based on the first document and the second document.

7. The method according to claim 5 or 6, wherein, Determining the target context information that matches the interaction information from the conversation context information set includes: If the amount of information in the aforementioned conversation context is less than the information amount threshold, then the aforementioned conversation context is used as the target context; and If the amount of information in the conversation context set is greater than or equal to the information amount threshold, the target context information is determined based on the summary information of the conversation context set and the historical context information that semantically matches the interaction information. The summary information is determined by semantic extraction of the conversation context set.

8. The method according to any one of claims 1 to 7, further comprising: Receive feedback information streamed from the target service node, wherein the feedback information is a portion of the interaction result; and If it is determined that the interaction result has been fully received, the target interaction information and the interaction result are updated to the session context information set.

9. The method according to any one of claims 1 to 8, wherein, The step of determining the target service node from multiple service nodes in the distributed service cluster to execute the target interaction task indicated by the target interaction information includes: Based on the task type of the target interaction task and the tool attribute information of the tools deployed by each of the multiple service nodes, the target service node is determined from the multiple service nodes, wherein the tool attribute information includes the task type of the interaction task that the tool can execute.

10. The method according to claim 9, wherein, The process of determining the target service node from among the multiple service nodes based on the task type of the target interaction task and the tool attribute information of the tools deployed on each of the multiple service nodes includes: Based on the task type of the target interaction task and the tool attribute information of the tools deployed by each of the multiple service nodes, multiple candidate service nodes are determined from the multiple service nodes; and The target service node is determined from the candidate service nodes based on the load information of each candidate service node.

11. An interactive device, comprising: A reference determination module is used to determine historical reference information in response to receiving interaction information from a target object, wherein the historical reference information includes content belonging to the same session as the interaction information; An update module is used to update the interaction information based on the historical reference information to obtain the target interaction information; A node determination module is used to determine, from multiple service nodes in a distributed service cluster, a target service node for executing the target interaction task indicated by the target interaction information; and The sending module is used to send the historical reference information and the target interaction information to the target service node, so as to use the target service node to execute the target interaction task and obtain the interaction result.

12. An intelligent agent, comprising: The input module is used to receive input information; The processing module is configured to determine a target task based on the input information received by the input module, determine a large model based on the target task, and obtain output information by calling the large model to execute the method of any one of claims 1 to 10. An output module is used to output the output information obtained by the processing module.

13. An electronic device, comprising: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1 to 10.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1 to 10.

15. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 10.

Images