Intelligent dialogue processing method and device, electronic equipment and storage medium

Abstract

The present disclosure discloses a kind of intelligent conversation processing method, device, electronic equipment and storage medium, it is related to computer technical field, specifically it is related to artificial intelligence technical field such as intelligent conversation and natural language processing.The specific implementation scheme includes: the first intention of the first query sentence in the first session of user is identified;Based on the first intention and the function of each agent in multiple agents, the first query sentence is routed to the matched first target agent, to respond to the first query sentence by the first target agent;If the conversation round attribute of the first target agent is multi-round conversation, first intention binding relationship between the session identifier of the first session and the identifier of the first target agent is constructed;Based on the first intention binding relationship, the first session is managed.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, specifically to the field of artificial intelligence technology such as intelligent dialogue and natural language processing, and particularly to an intelligent dialogue processing method, apparatus, electronic device and storage medium. Background Technology

[0002] With the development of artificial intelligence (AI) technology, the application of large language models (LLM) is becoming increasingly widespread.

[0003] For example, in intelligent dialogue applications, a large LLM can be used to handle all types of queries and provide responses. Summary of the Invention

[0004] This disclosure provides an intelligent dialogue processing method, apparatus, electronic device, and storage medium.

[0005] According to one aspect of this disclosure, an intelligent dialogue processing method is provided, comprising:

[0006] Identify the first intent of the first query in the user's first session;

[0007] Based on the first intent and the functions of each agent among the multiple agents, the first query statement is routed to the matching first target agent so that the first target agent can respond to the first query statement.

[0008] If the dialogue turn attribute of the first target agent is a multi-turn dialogue, a first intent binding relationship is constructed between the session identifier of the first session and the identifier of the first target agent;

[0009] Based on the first intent binding relationship, manage the first session.

[0010] According to another aspect of this disclosure, an intelligent dialogue processing method is provided, comprising:

[0011] Receive query statements from the user's current session routed by the intelligent dialogue processing device;

[0012] Respond to the query statement to obtain the response content;

[0013] Obtain the status identifier of the response content; the status identifier is used to identify the relationship between the current session and the current agent, including related, terminated, or unrelated;

[0014] The system returns a response carrying the status identifier to the intelligent dialogue processing device, so that the intelligent dialogue processing device can manage the current session based on the status identifier.

[0015] According to another aspect of this disclosure, an intelligent dialogue processing apparatus is provided, comprising:

[0016] The recognition module is used to identify the first intent of the first query statement in the user's first session;

[0017] The routing module is used to route the first query statement to a matching first target agent based on the first intent and the functions of each agent among the multiple agents, so that the first target agent can respond to the first query statement.

[0018] The construction module is used to construct a first intent binding relationship between the session identifier of the first session and the identifier of the first target intelligent agent when the dialogue turn attribute of the first target intelligent agent is a multi-turn dialogue;

[0019] The management module is used to manage the first session based on the first intent binding relationship.

[0020] According to another aspect of this disclosure, an intelligent agent is provided, comprising:

[0021] The receiving module is used to receive query statements from the current session of the user routed by the intelligent dialogue processing device;

[0022] The response generation module is used to respond to the query statement and obtain the response content;

[0023] The acquisition module is used to acquire the status identifier of the response content; the status identifier is used to identify the relationship between the current session and the current agent, including related, terminated, or unrelated.

[0024] The sending module is used to return response content carrying the status identifier to the intelligent dialogue processing device, so that the intelligent dialogue processing device can manage the current session based on the status identifier.

[0025] According to another aspect of this disclosure, an intelligent dialogue processing system is provided, including an intelligent dialogue processing device and a plurality of intelligent agents, each of the intelligent agents being communicatively connected to the intelligent dialogue processing device, the intelligent dialogue processing device being an intelligent dialogue processing device employing the aspects and any possible implementations described above, the intelligent agents employing the aspects and any possible implementations described above, and the intelligent dialogue processing is implemented using a method employing the aspects and any possible implementations described above.

[0026] According to yet another aspect of this disclosure, an electronic device is provided, comprising:

[0027] At least one processor; and

[0028] A memory communicatively connected to the at least one processor; wherein,

[0029] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the methods described above and any possible implementations.

[0030] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing the computer to perform the methods described above and any possible implementation thereof.

[0031] According to yet another aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the aspects and any possible implementations described above.

[0032] According to the technology disclosed herein, dialogue in a session can be processed accurately and effectively, thereby effectively improving dialogue processing efficiency.

[0033] 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

[0034] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0035] Figure 1 This is a schematic diagram based on the first embodiment of the present disclosure;

[0036] Figure 2 This is a schematic diagram according to the second embodiment of the present disclosure;

[0037] Figure 3 This is a schematic diagram according to the third embodiment of the present disclosure;

[0038] Figure 4 This is a schematic diagram according to the fourth embodiment of the present disclosure;

[0039] Figure 5 This is a schematic diagram according to the fifth embodiment of the present disclosure;

[0040] Figure 6 This is a schematic diagram according to the sixth embodiment of the present disclosure;

[0041] Figure 7 This is a schematic diagram according to the seventh embodiment of the present disclosure;

[0042] Figure 8 This is a schematic diagram according to the eighth embodiment of the present disclosure;

[0043] Figure 9 This is a schematic diagram according to the ninth embodiment of the present disclosure;

[0044] Figure 10 This is a block diagram of an electronic device used to implement the methods of the embodiments of this disclosure. Detailed Implementation

[0045] 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.

[0046] Obviously, the described embodiments are only some, not all, of the embodiments disclosed herein. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0047] It should be noted that the terminal devices involved in the embodiments of this disclosure may include, but are not limited to, smart devices such as mobile phones, personal digital assistants (PDAs), wireless handheld devices, and tablet computers; the display devices may include, but are not limited to, personal computers, televisions, and other devices with display functions.

[0048] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0049] Figure 1 This is a schematic diagram based on the first embodiment of the present disclosure; as shown Figure 1 As shown, this embodiment provides an intelligent dialogue processing method, which may specifically include the following steps:

[0050] S101. Identify the first intent of the first query statement in the user's first session;

[0051] The executing entity of the intelligent dialogue processing method in this embodiment is an intelligent dialogue processing device, which can be an electronic entity or a software-integrated application that runs on a computer device when in use.

[0052] In this embodiment, the intelligent dialogue processing device processes user dialogues at the session level. Specifically, during a single session with a user, the user's problem can be resolved in one round of dialogue or through multiple rounds. In scenarios where multiple rounds of dialogue are used to resolve the user's problem, the user needs to input multiple query statements through these rounds. For example, taking the user's first session as an example, the first query statement can be the first query statement entered by the user in the first session. In this embodiment, at the beginning of the first session, it is necessary to identify the initial intent of the first query statement in the first session.

[0053] S102. Based on the first intent and the functions of each agent among the multiple agents, the first query statement is routed to the matching first target agent so that the first target agent can respond to the first query statement.

[0054] In this embodiment, the dialogue management processing device can schedule multiple intelligent agents, each with different functions. Specifically, this step involves detecting the degree of matching between the first intent and the functions of each intelligent agent based on their functions and the first intent. Ultimately, a first target intelligent agent that best matches the first intent is selected, deemed capable of accurately and effectively responding to the user's query statement in the first session. Further, the first query statement is sent to the matched first target intelligent agent, which then responds to the query statement.

[0055] S103. If the dialogue turn attribute of the first target agent is multi-turn dialogue, construct the first intent binding relationship between the session identifier of the first session and the identifier of the first target agent;

[0056] Specifically, for each agent, not only can its functionality be known in advance, but also its dialogue turn attribute. This dialogue turn attribute can include one-turn or multi-turn dialogue. A one-turn dialogue limits the agent to responding to the user's query in a single interaction. A multi-turn dialogue limits the agent to responding to the user's query through multiple rounds of dialogue in a single interaction. In multi-turn dialogues, the agent can guide the user to input more queries to compensate for the shortcomings of the first query in the current session, thus providing a more accurate response to the user's question.

[0057] S104. Based on the first intent binding relationship, manage the first session.

[0058] Based on the built-in first intent binding management, the first session can be effectively managed to efficiently process the user's first session dialogue.

[0059] For agents with a single-turn dialogue attribute, after the first query is routed to the matching target agent, the target agent directly responds to the first query, and the dialogue ends.

[0060] For agents with a dialogue turn attribute of multi-turn dialogue, the method in this embodiment can be adopted. After routing the first query statement to the matching first target agent, it is necessary to construct a first intent binding relationship between the session identifier of the first session and the identifier of the first target agent, and manage the first session based on the first intent binding relationship.

[0061] The intelligent dialogue processing method of this embodiment can route the first query statement to the matching first target intelligent agent based on the first intent of the first query statement in the user's first session and with reference to the functions of each intelligent agent. At the same time, when the dialogue round attribute of the first target intelligent agent is multi-turn dialogue, it can also construct a first intent binding relationship between the session identifier of the first session and the identifier of the first target intelligent agent. Based on the first intent binding relationship, the first session can be managed, and the dialogue in the first session can be processed accurately and effectively, thereby effectively improving the dialogue processing efficiency.

[0062] Figure 2 This is a schematic diagram based on the second embodiment of this disclosure; the intelligent dialogue processing method of this embodiment, in the above... Figure 1 Based on the technical solutions of the illustrated embodiments, the technical solutions of this disclosure will be described in further detail. For example... Figure 2 As shown, the intelligent dialogue processing method of this embodiment may specifically include the following steps:

[0063] S201. Identify the first intent of the first query statement in the user's first session;

[0064] Specifically, in this embodiment, a pre-trained intent recognition model or a pre-set intent recognition rule can be used to identify the first intent of the first query statement in the user's first session.

[0065] S202. Based on the first intent and the functions of each agent among the multiple agents, the first query statement is routed to the matching first target agent so that the first target agent can respond to the first query statement.

[0066] S203. Obtain the dialogue turn attribute of the first target agent from the attribute information of the first target agent;

[0067] In this embodiment, the dialogue turn attribute is used to identify whether the first target agent responds to a question in a session using a single-turn dialogue or a multi-turn dialogue.

[0068] S204. Detect whether the dialogue turn attribute of the first target agent is a single-turn dialogue or a multi-turn dialogue. If it is a multi-turn dialogue, proceed to step S205; if it is a single-turn dialogue, proceed to step S206.

[0069] S205. Construct a first intent binding relationship between the session identifier of the first session and the identifier of the first target agent; execute step S206;

[0070] In practice, the first intent binding relationship can be constructed by generating a mapping relationship between the session identifier of the first session and the identifier of the first target agent, and storing it in the intent binding list.

[0071] S206. Receive the response content of the first query statement returned by the first target agent; execute step S207;

[0072] S207. Based on the response content of the first query statement, respond to the user's first query statement;

[0073] By employing steps S206 and S207, a complete round of dialogue can be completed, ensuring the integrity and accuracy of dialogue processing.

[0074] It should be noted that if the dialogue round attribute of the first target agent is multi-turn dialogue, the response content of the first query statement can also carry a status identifier. This status identifier is marked by the first target agent in the response content of the first query statement and is used to identify the relationship between the current first session and the first target agent. The status identifier includes relevant, ended, or irrelevant. In this case, the first session only includes the first query statement. Correspondingly, relevant means that the first query statement of the first session is relevant to the function of the first target agent. Irrelevant means that the first query statement of the first session is irrelevant to the function of the first target agent, but only one round of dialogue has been conducted, and the first session will not end directly. The first target agent can still conduct multiple rounds of interaction with the user through guidance. End means that the first target agent believes that the user's question in the first session has been answered, and the user's first session can be ended at this time; or after multiple rounds of guidance from the first target agent, if the user's supplementary query statements are still irrelevant to the function of the first target agent, an end identifier can be forcibly added to the corresponding response content when the preset tolerance threshold is exceeded, thus forcibly ending the first session. Because the dialogue turn attribute of the first target agent is multi-turn dialogue, the response content of other query statements after the first query statement in the first session can usually carry an end status indicator, while the response content of the first query statement usually does not carry an end status indicator.

[0075] In addition, if the dialogue turn attribute of the first target agent is single-turn dialogue, the first dialogue ends at step S207 after the user's first session response is completed.

[0076] S208. If the dialogue turn attribute of the first target agent is multi-turn dialogue, continue to obtain other query statements in the user's first session; execute step S209.

[0077] For example, other query statements in the first session of this embodiment can be considered as any query statement other than the first query statement in the first session. For example, in a multi-turn dialogue, the first query statement is usually incomplete, causing the first target agent to be unable to respond accurately. Therefore, other query statements in a multi-turn dialogue can be considered as supplementary query statements to make up for the incompleteness of the first query statement.

[0078] S209. Based on the first intent binding relationship, route other query statements to the first target agent so that the first target agent can continue to respond to other query statements in the user's first session; execute step S210;

[0079] Optionally, steps S203-S205 and steps S206-S207 in this embodiment can be performed simultaneously, but step S209 must be performed after step S205.

[0080] By using steps S208 and S209, other query statements in the first session can be directly routed to the corresponding first target agent based on the first intent binding relationship, without having to identify the intent of other query statements, which can effectively improve the dialogue processing efficiency in the first session.

[0081] S210, Receive the response content of other query statements returned by the first target agent; execute step S211;

[0082] S211. Obtain the status identifier from the response content of other query statements; the status identifier includes relevant, ended, or irrelevant; proceed to step S212;

[0083] S212. Based on the response content of other query statements, respond to the user's other query statements; execute step S213;

[0084] S213. In response to the state flag indicating the end, unbind the first intent binding relationship between the session flag and the flag of the first target agent, and end the first session; execute step S214.

[0085] Specifically, in practical applications, the first intent binding relationship can be unbound by deleting it from the intent binding list; correspondingly, the intelligent dialogue processing device will no longer maintain the session identifier of the first session, thus ending the first session. In this way, after the first session ends, when a query statement sent by the user is received again, it will be automatically identified as a second session, different from the first session.

[0086] In cases where the status is marked as relevant or irrelevant, the first intention binding relationship will not be unbound.

[0087] It should be noted that in practical applications, there is no sequential relationship between steps S212 and S213; they can be executed simultaneously.

[0088] In this embodiment, by employing steps S210 and S212 as described above, a complete round of dialogue corresponding to other query statements can be completed, ensuring the integrity and accuracy of dialogue processing.

[0089] Furthermore, by employing steps S211 and S213, the first intent binding relationship can be unbound in a timely manner at the end of the session, effectively ensuring the accuracy of subsequent sessions.

[0090] S214. Obtain the second query statement of the user's second session;

[0091] Referring to the above description, it can be seen that during the existence of the first intent binding relationship, there are two situations. The first situation is that the status indicators of the response content of multiple query statements in the first session are all irrelevant. In this case, the first target agent can detect whether the number of irrelevant dialogue rounds in the first session has reached a preset tolerance threshold. If it has, the first target agent can forcibly terminate the first session, carrying an end status indicator in the response content of the last query statement in the first session. Correspondingly, in the current intelligent session processing device, it is also possible to obtain that during the existence of the first intent binding relationship, the status indicator in the response content of the last query statement is "end," and the status indicators in the response content of the previous query statements are all irrelevant.

[0092] The second scenario is that in the first session, some responses to multiple queries have status indicators that are irrelevant, while others have status indicators that are relevant. For example, at the beginning of the first session, the user might enter an incomplete query, causing the first target agent to believe that the first session is irrelevant to its function. However, as the first target agent guides the user to gradually enter more detailed queries, the first target agent determines that the queries are relevant until the user's problem is finally resolved, at which point the first target agent ends the session.

[0093] In the first scenario, if the problem remains unresolved after the first session ends, the user can initiate a second session. In this case, the second session can have the same problem as the first session. In the second scenario, the user can also initiate a second session with a new problem after the first session ends. In this case, the problem in the second session can be related to, or unrelated to, the problem in the first session.

[0094] S215. Based on the second query statement and the constructed user state pool, identify the second intent of the second query statement; the user state pool is constructed based on multiple query statements corresponding to the first session identifier during the interaction between the first target agent and the user, and the user state pool is stored independently of each agent, including structured information of the entire domain.

[0095] Based on the above analysis, if the user's problem in the first session remains unresolved, the second intent can be re-identified based on the user's second query statement. To improve the accuracy of second intent identification, information from the first session can be combined for identification. In this embodiment, during the first session of interaction between the first target agent and the user, a user state pool can be constructed based on the user's multiple query statements. This user state pool is stored independently of each agent, but all agents can access it. The user state pool includes structured information across the entire domain. For example, during the dialogue, an entity extraction model can be used to extract entities and their corresponding values ​​from the user's query statements, convert them into structured data, and store them in the user's state pool. Therefore, the second intent of the second query statement can be accurately and efficiently identified based on the second query statement and the constructed user state pool.

[0096] If a user's problem is resolved in the first session, the user can also raise other questions related to the problem in the first session in a second session. At this time, the user's state pool that has been built can be referenced to accurately and effectively identify the intent of the second query statement.

[0097] In other words, regardless of whether the first session ends because the problem is solved or is forcibly ended because the problem is irrelevant, the user's state pool generated in the first session can be referenced to accurately and efficiently identify the second intent of the second query statement.

[0098] For example, in the medical field, all key medical data of a user can be extracted during the first session, such as chief complaint, duration, medication history, and recommended departments, and stored in the user's state pool. After the first session ends, when the user initiates a second session, the user's state pool can be directly accessed, eliminating the need for the user to re-enter the information already entered in the first session. This enables "one-time input, full reuse" of information, greatly improving user satisfaction and dialogue processing efficiency.

[0099] Alternatively, in this embodiment, the intelligent dialogue processing device, the first target intelligent agent, or other detection devices can also detect whether there are conflicts in the information in the user's state pool. For example, if different rounds of query statements attempt to modify the value of the same slot in the user's state pool, the confidence level of each value can be determined based on a preset strategy, or the priority of the timestamp can be combined to make judgments and resolve conflicts, so as to ensure the rigor and accuracy of the data in the user's state pool.

[0100] S216. Based on the second intent, route the second query statement to the matching second target agent so that the second target agent responds to the second query statement; execute step S217.

[0101] Additionally, it should be noted that when the second target agent responds to the second query statement, it can also refer to the data in the user's state pool to respond accurately and effectively.

[0102] Specifically, when the second target agent is initialized, it first accesses the user's state pool and automatically performs slot mapping according to the slot requirements of the second target agent. The existing structured data in the user's state pool is directly filled into the slots of the second target agent, which allows the second target agent to start with inherited data, skips repeated queries for already collected information, and effectively improves the response accuracy and response efficiency of the second target agent.

[0103] S217. Obtain the dialogue turn attribute of the second target agent; the dialogue turn attribute is used to identify whether the second target agent responds to the second query statement using a single-turn dialogue or a multi-turn dialogue; execute step S218;

[0104] S218. In response to the second target agent's dialogue turn attribute being a multi-turn dialogue agent, construct a second intent binding relationship between the second session's session identifier and the second target agent's identifier; execute step S219.

[0105] S219. Manage the second session based on the second intent binding relationship.

[0106] Specifically, the implementation method for managing the second session based on the second intent binding relationship is the same as the implementation method for managing the first session based on the first intent binding relationship described above. For details, please refer to the relevant content such as steps S206-S213 above, which will not be repeated here.

[0107] The intelligent dialogue processing method of this embodiment can be applied to a medical question-and-answer system. In this system, multiple intelligent agents with specialized functions can be pre-configured, as well as at least one general-purpose intelligent agent. For example, the multiple specialized intelligent agents can include agents corresponding to various sub-fields within the medical field, such as those for dentistry, cardiovascular medicine, and traditional Chinese medicine. Since users may also input non-medical queries during the dialogue, such as "What's the weather like today?" or "What did I eat today?", at least one general-purpose intelligent agent is also needed to uniformly process non-specialized query statements input by users.

[0108] The intelligent dialogue processing method in this embodiment can classify and process dialogue turns based on the dialogue turn attributes of the target intelligent agent matched with the intent. This avoids the use of a unified model and a single process in existing technologies, enabling decoupling of stateless and stateful services at the architectural level. It can effectively ensure the efficiency of simple and high-frequency queries in single-turn dialogues. Furthermore, in multi-turn dialogue scenarios, it can manage sessions by building intent binding relationships, effectively reducing redundant resource consumption and improving dialogue efficiency in multi-turn dialogue scenarios.

[0109] The intelligent dialogue processing method in this embodiment can effectively avoid the problem of data silos between different intelligent agents by introducing a user state pool. It can also avoid users inputting repetitive information and solve the token over-limit problem when intelligent agents input context information, which significantly improves reasoning efficiency. At the same time, it provides a safe and reliable data foundation for subsequent data inheritance, thereby effectively improving the processing efficiency of intelligent dialogue.

[0110] Figure 3 This is a schematic diagram according to the third embodiment of the present disclosure; the intelligent dialogue processing method of this embodiment may specifically include the following steps:

[0111] S301, Receive the query statement from the user's current session routed by the intelligent dialogue processing device;

[0112] In this embodiment, the query statement can be any query statement in the current session.

[0113] S302. Respond to the query statement and obtain the response content;

[0114] S303. Obtain the status flag of the response content; the status flag is used to identify the relationship between the current session and the current agent, including related, terminated, or unrelated;

[0115] For example, "relevant" indicates that the current query in the current session is relevant to the function of the current agent. "Irrelevant" indicates that the current query in the current session is irrelevant to the function of the current agent, but the number of irrelevant dialogue rounds has not yet reached the preset tolerance threshold, so the current session cannot be terminated directly. "Terminate" indicates that the current agent believes the user's question in the current session has been answered, and the current session can be terminated; or, after multiple rounds of guidance from the current agent, if the user's supplementary queries are still irrelevant to the function of the current agent, the current session can be forcibly terminated by adding an end marker to the corresponding response content if the preset tolerance threshold is exceeded.

[0116] Because the current agent's dialogue turn attribute is multi-turn, the response to the first query in the current session typically does not include an end status indicator. However, the response to the second and subsequent queries in the current session may include an end status indicator.

[0117] S304. Return response content carrying a status identifier to the intelligent dialogue processing device so that the intelligent dialogue processing device can manage the current session based on the status identifier.

[0118] With the above Figure 1 The difference between the embodiments shown is that the above-described embodiments are... Figure 1 The embodiments shown describe the technical solutions of this disclosure on the side of the intelligent dialogue processing device, while this embodiment describes the technical solutions of this disclosure on the side of the intelligent agent whose dialogue round attribute is multi-turn dialogue.

[0119] Specifically, please refer to the above. Figure 2 According to the illustrated embodiment, if the status indicator is "End," the intelligent dialogue processing device can unbind the intent binding relationship corresponding to the current session and end the current session. If the status indicator is "Relevant" or "Irrelevant," the intent binding relationship corresponding to the current session is not unbound, and the device continues to respond to other query statements in the user's current session.

[0120] The intelligent dialogue processing method in this embodiment obtains the status identifier of the response content of the query statement and returns the response content carrying the status identifier to the intelligent dialogue processing device. This enables the intelligent dialogue processing device to effectively manage the current session based on the status identifier, accurately and effectively process the dialogue in the current session, and thus effectively improve the dialogue processing efficiency.

[0121] Figure 4 This is a schematic diagram based on the fourth embodiment of this disclosure; the intelligent dialogue processing method of this embodiment, in the above... Figure 3 Based on the technical solutions of the illustrated embodiments, the technical solutions of this disclosure will be described in further detail. For example... Figure 4 As shown, the intelligent dialogue processing method of this embodiment may specifically include the following steps:

[0122] S401, Receive the query statement from the user's current session routed by the intelligent dialogue processing device;

[0123] S402. Respond to the query statement and obtain the response content;

[0124] S403. Based on the query statement and the current agent's function, determine the relevance between the query statement and the current agent;

[0125] S404. The correlation result is either correlated or not correlated. If the correlation result is correlated, proceed to step S405; if the correlation result is not correlated, proceed to step S406.

[0126] S405. Determine the status indicator of the response content as relevant; proceed to step S409.

[0127] S406. Detect whether the number of irrelevant responses in the current session has reached the preset tolerance threshold. If it has, proceed to step S407; if it has not, proceed to step S408.

[0128] S407. Determine the status indicator of the response content as "end"; proceed to step S409.

[0129] S408. Determine the status of the response content as irrelevant; proceed to step S409.

[0130] By using the above steps S403-S408, the status identifier of the response content of the query statement can be accurately and effectively determined.

[0131] S409. Return response content carrying a status identifier to the intelligent dialogue processing device so that the intelligent dialogue processing device can manage the current session based on the status identifier.

[0132] S410. Based on the query statement, construct or update the user's state pool. The user's state pool is independent of the current agent's storage and includes structured information from the entire domain.

[0133] Refer to the above Figure 2 According to the relevant description of the embodiment shown, during the dialogue process, an entity extraction model can be used to extract entities and their corresponding values ​​from the query statement, and convert them into structured data, which is then stored in the user's state pool.

[0134] For example, in the medical field, all key medical data of a user, such as chief complaint, duration, medication history, and recommended departments, can be extracted from the query statement of the current session and stored in the user's state pool. After the current session ends, when the user initiates a subsequent session, the user's state pool can be directly accessed, without requiring the user to re-enter the information already entered in the previous session. This enables "one-time input, full reuse" of information, greatly improving user satisfaction and dialogue processing efficiency.

[0135] Optionally, when responding to the query statement in step S402, if the user's state pool already exists, the user's state pool can be referenced to respond to the first query statement and obtain the response content.

[0136] For example, when the current agent initializes, it can first access the user's state pool and automatically perform slot mapping according to the current agent's own slot requirements. The existing structured data in the user's state pool can be directly filled into the current agent's slots, allowing the current agent to start with inherited data, skipping repeated queries for already collected information, and effectively improving the current agent's response accuracy and efficiency.

[0137] Therefore, by referencing the user's state pool and responding to the first query, the system can inherit the user's historical information, conduct dialogue accurately and effectively, and significantly improve the accuracy and efficiency of dialogue processing.

[0138] Alternatively, in this embodiment, the intelligent dialogue processing device, any intelligent agent, or other detection subject can detect whether there are conflicts in the information in the user's state pool. For example, if different rounds of query statements attempt to modify the value of the same slot in the user's state pool, the confidence level of each value can be determined based on a preset strategy, or the priority of the timestamp can be combined to make judgments and resolve conflicts, so as to ensure the rigor and accuracy of the data in the device's state pool.

[0139] The intelligent dialogue processing method in this embodiment can effectively avoid the problem of data silos between different intelligent agents by introducing a user state pool. It can also avoid users inputting repetitive information and solve the token over-limit problem when intelligent agents input context information, which significantly improves reasoning efficiency. At the same time, it provides a safe and reliable data foundation for subsequent data inheritance, thereby effectively improving the processing efficiency of intelligent dialogue.

[0140] This disclosed intelligent dialogue processing method can be applied to medical question-and-answer systems. It addresses common technical problems in complex, multi-turn, cross-domain, and high-risk medical dialogues, such as chaotic context management, rigid intent transitions, redundant information collection, and insufficient system robustness. By constructing a user state pool, it manages user context information, avoiding redundant information collection, thus saving redundant resources and effectively improving dialogue processing efficiency. Furthermore, it effectively manages each session by establishing intent binding relationships between them. This disclosed intelligent dialogue processing method is applicable to various general-purpose medical intelligent question-and-answer platforms, virtual doctor assistants, triage and guidance robots, electronic medical record front-end entry systems, and other projects.

[0141] Figure 5 This is a schematic diagram according to the fifth embodiment of this disclosure; as shown Figure 5 As shown, this embodiment provides an intelligent dialogue processing device 500, including:

[0142] The identification module 501 is used to identify the first intent of the first query statement in the first session of a user.

[0143] The routing module 502 is used to route the first query statement to a matching first target agent based on the first intent and the functions of each agent among the multiple agents, so that the first target agent can respond to the first query statement.

[0144] The construction module 503 is used to establish a first intent binding relationship between the session identifier for constructing the first session and the identifier of the first target intelligent agent when the dialogue turn attribute of the first target intelligent agent is a multi-turn dialogue.

[0145] Management module 504 is used to manage the first session based on the first intent binding relationship.

[0146] The intelligent dialogue processing device 500 in this embodiment achieves the same implementation principle and technical effect as the above-mentioned related method embodiments by using the above-mentioned modules. For details, please refer to the description of the above-mentioned related method embodiments, which will not be repeated here.

[0147] Figure 6 This is a schematic diagram according to the sixth embodiment of the present disclosure; the intelligent dialogue processing device 600 of this embodiment, in the above... Figure 5 Based on the technical solutions of the illustrated embodiments, the technical solutions of this disclosure will be described in further detail. For example... Figure 5 As shown, the intelligent dialogue processing device 600 of this embodiment includes... Figure 5 The modules with the same name and function shown are: identification module 601, routing module 602, construction module 603, and management module 604.

[0148] like Figure 6 As shown, the intelligent dialogue processing device 600 in this embodiment further includes:

[0149] The receiving module 605 is used to receive the response content of the first query statement returned by the first target intelligent agent;

[0150] The response module 606 is used to respond to the user's first query statement based on the response content of the first query statement.

[0151] Further optionally, the management module 604 in this embodiment can be used for:

[0152] Retrieve other query statements from the user's first session;

[0153] The routing module is further configured to route the other query statements to the first target agent based on the first intent binding relationship, so that the first target agent can continue to respond to the other query statements in the user's first session.

[0154] Further optional, such as Figure 6 As shown, the intelligent dialogue processing device 600 in this embodiment further includes:

[0155] The acquisition module 607 is used to acquire a status identifier from the response content of the other query statements; the status identifier is used to identify the relationship between the first session and the first target agent, including related, ended, or unrelated.

[0156] The unbinding module 608 is used to unbind the first intent binding relationship between the session identifier and the identifier of the first target intelligent agent in response to the state identifier being terminated; and to terminate the first session.

[0157] Further optionally, in one embodiment of this disclosure, the receiving module 605 is also configured to obtain the second query statement of the user's second session;

[0158] The identification module 601 is further configured to identify the second intent of the second query statement based on the second query statement and the constructed state pool of the user; the state pool of the user is constructed based on multiple query statements corresponding to the first session identifier during the interaction between the first target agent and the user, and the state pool of the user is stored independently of each agent and includes structured information of the entire domain.

[0159] The routing module 602 is further configured to route the second query statement to a matching second target agent based on the second intent, so that the second target agent responds to the second query statement.

[0160] Further optionally, in one embodiment of this disclosure, the acquisition module 607 is further configured to acquire the dialogue turn attribute of the second target agent; the dialogue turn attribute is used to identify whether the second target agent responds to the question of the second query statement using a single-turn dialogue or a multi-turn dialogue;

[0161] The construction module 603 is also configured to, in response to the second target agent's dialogue turn attribute being a multi-turn dialogue agent, establish a second intent binding relationship between the session identifier for constructing the second session and the identifier of the second target agent;

[0162] The management module 604 is also used to manage the second session based on the second intent binding relationship.

[0163] Alternatively, in one embodiment of this disclosure, the acquisition module 607 is further configured to acquire the dialogue turn attribute of the first target intelligent agent from the attribute information of the first target intelligent agent.

[0164] The intelligent dialogue processing device 600 in this embodiment achieves the same implementation principle and technical effect as the above-mentioned related method embodiments by using the above-mentioned modules. For details, please refer to the description of the above-mentioned related method embodiments, which will not be repeated here.

[0165] Figure 7 This is a schematic diagram according to the seventh embodiment of the present disclosure; as shown Figure 7 As shown, this embodiment provides an intelligent agent 700, including:

[0166] The receiving module 701 is used to receive query statements from the current session of a user routed by the intelligent dialogue processing device;

[0167] The response generation module 702 is used to respond to the query statement and obtain the response content;

[0168] The acquisition module 703 is used to acquire the status identifier of the response content; the status identifier is used to identify the relationship between the current session and the current agent, including related, terminated, or unrelated.

[0169] The sending module 704 is used to return response content carrying the status identifier to the intelligent dialogue processing device, so that the intelligent dialogue processing device can manage the current session based on the status identifier.

[0170] The intelligent agent 700 in this embodiment achieves the same implementation principle and technical effect of intelligent dialogue processing by using the above-mentioned modules. For details, please refer to the description of the above-mentioned related method embodiments, which will not be repeated here.

[0171] Figure 8 This is a schematic diagram according to the eighth embodiment of the present disclosure; this embodiment provides an intelligent agent 800, in the above... Figure 7 Based on the technical solutions of the illustrated embodiments, the technical solutions of this disclosure will be described in further detail. For example... Figure 8 As shown, the intelligent agent 800 in this embodiment includes the above-mentioned Figure 7 The modules with the same name and function shown are: receiving module 801, response generation module 802, acquisition module 803, and sending module 804.

[0172] The acquisition module 803 is used for:

[0173] Based on the first query statement and the current agent's function, determine the correlation result between the first query statement and the current agent; the correlation result includes whether it is related or not.

[0174] If the correlation result is irrelevant, and the number of irrelevant responses in the first session reaches a preset tolerance threshold, the status identifier of the response content is determined to be "end".

[0175] Further, optionally, in one embodiment of this disclosure, the acquisition module 803 is also used for:

[0176] If the correlation result is irrelevant, and the number of irrelevant responses in the first session does not reach the preset tolerance threshold, the status identifier of the response content is determined to be irrelevant.

[0177] Further, optionally, in one embodiment of this disclosure, the acquisition module 803 is also used for:

[0178] If the correlation result is positive, the status identifier of the response content is determined to be positive.

[0179] Further optional, such as Figure 8 As shown, in one embodiment of this disclosure, the intelligent agent 800 further includes:

[0180] The update module 805 is used to construct or update the user's state pool based on the first query statement. The user's state pool is independent of the current agent storage and includes structured information across the entire domain.

[0181] Further optionally, in one embodiment of this disclosure, the response generation module 802 is used to respond to the first query statement with reference to the user's state pool to obtain the response content.

[0182] The intelligent agent 800 in this embodiment achieves the same implementation principle and technical effect of intelligent dialogue processing by adopting the above-mentioned modules. For details, please refer to the description of the above-mentioned related method embodiments, which will not be repeated here.

[0183] Figure 9 This is a schematic diagram based on the ninth embodiment of this disclosure; as shown Figure 9 As shown, this embodiment provides an intelligent dialogue processing system 900, including an intelligent dialogue processing device 901 and multiple intelligent agents 902. Each intelligent agent 902 is communicatively connected to the intelligent dialogue processing device 901. The intelligent dialogue processing device 901 adopts the above-described... Figure 5 or Figure 6 The intelligent dialogue processing device described above employs the intelligent agent as described above. Figure 7 or Figure 8The aforementioned intelligent agent; and specifically, it can be adopted as described above. Figures 1-4 The intelligent dialogue processing method of the embodiment shown implements intelligent dialogue processing. For details, please refer to the relevant descriptions of the above-mentioned related embodiments, which will not be repeated here.

[0184] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0185] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0186] Figure 10 A schematic block diagram of an example electronic device 1000 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.

[0187] like Figure 10 As shown, device 1000 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in read-only memory (ROM) 1002 or a computer program loaded into random access memory (RAM) 1003 from storage unit 1008. The RAM 1003 may also store various programs and data required for the operation of device 1000. The computing unit 1001, ROM 1002, and RAM 1003 are interconnected via bus 1004. Input / output (I / O) interface 1005 is also connected to bus 1004.

[0188] Multiple components in device 1000 are connected to I / O interface 1005, including: input unit 1006, such as keyboard, mouse, etc.; output unit 1007, such as various types of monitors, speakers, etc.; storage unit 1008, such as disk, optical disk, etc.; and communication unit 1009, such as network card, modem, wireless transceiver, etc. Communication unit 1009 allows device 1000 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0189] The computing unit 1001 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1001 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 1001 performs the various methods and processes described above, such as the methods of this disclosure. For example, in some embodiments, the methods of this disclosure can be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 1008. In some embodiments, part or all of the computer program can be loaded and / or installed on device 1000 via ROM 1002 and / or communication unit 1009. When the computer program is loaded into RAM 1003 and executed by the computing unit 1001, one or more steps of the methods of this disclosure described above can be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to perform the methods described above in this disclosure by any other suitable means (e.g., by means of firmware).

[0190] 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.

[0191] 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.

[0192] 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.

[0193] 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).

[0194] 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.

[0195] 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, servers in distributed systems, or servers incorporating blockchain technology.

[0196] 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.

[0197] 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 intelligent dialogue processing method, comprising: Identify the first intent of the first query in the user's first session; Based on the first intent and the functions of each agent among the multiple agents, the first query statement is routed to the matching first target agent so that the first target agent can respond to the first query statement. If the dialogue turn attribute of the first target agent is a multi-turn dialogue, a first intent binding relationship is constructed between the session identifier of the first session and the identifier of the first target agent; Based on the first intent binding relationship, manage the first session.

2. The method according to claim 1, wherein, The method further includes: Receive the response content of the first query statement returned by the first target intelligent agent; Based on the response content of the first query statement, respond to the user's first query statement.

3. The method according to claim 1, wherein, Based on the first intent binding relationship, manage the first session, including: Retrieve other query statements from the user's first session; Based on the first intent binding relationship, the other query statements are routed to the first target agent, so that the first target agent can continue to respond to the other query statements in the user's first session.

4. The method according to claim 3, wherein, The method further includes: Obtain a status identifier from the response content of the other query statements; the status identifier is used to identify the relationship between the first session and the first target agent, including related, terminated, or unrelated; In response to the state identifier indicating an end, the first intent binding relationship between the session identifier and the identifier of the first target agent is unbound; the first session is terminated.

5. The method according to claim 4, wherein, After unbinding the first intent binding relationship between the session identifier and the identifier of the first target agent, the method further includes: Obtain the second query statement of the user's second session; Based on the second query statement and the constructed user state pool, the second intent of the second query statement is identified; the user state pool is constructed based on multiple query statements corresponding to the first session identifier during the interaction between the first target agent and the user, and the user state pool is stored independently of each agent, including structured information of the entire domain; Based on the second intent, the second query statement is routed to a matching second target agent, so that the second target agent responds to the second query statement.

6. The method according to claim 5, wherein, The method further includes: Obtain the dialogue turn attribute of the second target agent; the dialogue turn attribute is used to identify whether the second target agent responds to the second query statement using a single-turn dialogue or a multi-turn dialogue. When the dialogue turn attribute of the second target agent is a multi-turn dialogue agent, a second intent binding relationship is constructed between the session identifier of the second session and the identifier of the second target agent; The second session is managed based on the second intent binding relationship.

7. The method according to claim 1, wherein, The method further includes: Obtain the dialogue turn attribute of the first target agent from the attribute information of the first target agent.

8. An intelligent dialogue processing method, comprising: Receive query statements from the user's current session routed by the intelligent dialogue processing device; Respond to the query statement to obtain the response content; Obtain the status identifier of the response content; The status identifier is used to identify the relationship between the current session and the current agent, including related, terminated, or unrelated; The system returns a response carrying the status identifier to the intelligent dialogue processing device, so that the intelligent dialogue processing device can manage the current session based on the status identifier.

9. The method according to claim 8, wherein, Obtaining the status identifier of the response content includes: Based on the first query statement and the current agent's function, determine the correlation result between the first query statement and the current agent; the correlation result includes whether it is related or not. If the correlation result is irrelevant, and the number of irrelevant responses in the first session reaches a preset tolerance threshold, the status identifier of the response content is determined to be "end".

10. The method according to claim 9, wherein, Obtaining the status indicator of the response content also includes: If the correlation result is irrelevant, and the number of irrelevant responses in the first session does not reach the preset tolerance threshold, the status identifier of the response content is determined to be irrelevant.

11. The method according to claim 9, wherein, Obtaining the status indicator of the response content also includes: If the correlation result is positive, the status identifier of the response content is determined to be positive.

12. The method according to any one of claims 8-11, wherein, The method further includes: Based on the first query statement, the user's state pool is constructed or updated. The user's state pool is independent of the current agent storage and includes structured information across the entire domain.

13. The method according to claim 12, wherein, The response to the first query statement yields the response content, including: Referring to the user's state pool, the system responds to the first query statement and obtains the response content.

14. An intelligent dialogue processing device, comprising: The recognition module is used to identify the first intent of the first query statement in the user's first session; The routing module is used to route the first query statement to a matching first target agent based on the first intent and the functions of each agent among the multiple agents, so that the first target agent can respond to the first query statement. The construction module is used to construct a first intent binding relationship between the session identifier of the first session and the identifier of the first target intelligent agent when the dialogue turn attribute of the first target intelligent agent is a multi-turn dialogue; The management module is used to manage the first session based on the first intent binding relationship.

15. An intelligent agent, comprising: The receiving module is used to receive query statements from the current session of the user routed by the intelligent dialogue processing device; The response generation module is used to respond to the query statement and obtain the response content; The acquisition module is used to acquire the status identifier of the response content; The status identifier is used to identify the relationship between the current session and the current agent, including related, terminated, or unrelated; The sending module is used to return response content carrying the status identifier to the intelligent dialogue processing device, so that the intelligent dialogue processing device can manage the current session based on the status identifier.

16. An intelligent dialogue processing system, comprising an intelligent dialogue processing device and a plurality of intelligent agents, each of the intelligent agents being communicatively connected to the intelligent dialogue processing device, the intelligent dialogue processing device being the intelligent dialogue processing device as described in claim 14 above, the intelligent agents being the intelligent agents as described in claim 15 above; and employing the method as described in any one of claims 1-13 above to realize intelligent dialogue processing.

17. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method according to any one of claims 1-13.

18. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-13.

19. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method according to any one of claims 1-13.

Images