Information processing method, information processing device, computer device, and storage medium

Abstract

The application relates to an information processing method, an information processing device, computer equipment and a storage medium. The method comprises the following steps: obtaining an information processing request triggered by a user account; the information processing request carries a dialogue sentence and account information of the user account; the dialogue sentence represents a request intention of the information processing request; based on the dialogue sentence and the account information, a reference processing path is determined from a plurality of candidate processing paths stored in a preset path library; the candidate processing path is a workflow in which a plurality of task nodes are connected in a preset order; the information processing request and the reference processing path are input into a pre-trained large language model to obtain a predicted processing path for the request intention output by the large language model; and task nodes of a workflow corresponding to the predicted processing path are executed to obtain a processing result for the information processing request. The method can improve the rationality of the predicted processing path and reduce the deviation between the final processing result and the actual request intention of the user.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an information processing method, an information processing apparatus, a computer device, and a computer-readable storage medium. Background Technology

[0002] Large Language Models (LLMs, which are generative models), such as ChatGPT (Chat Generative Pre-trained Transformer, a chatbot program developed by OpenAI), can generate and execute many downstream tasks in response to user requests for information processing (i.e., demand tasks). For example, for user demand tasks related to holiday travel, a series of task nodes are generated and executed.

[0003] However, the current methods for applying LLM to the real world to generate and execute task nodes are not yet mature, resulting in generated task nodes that do not quite match the user's actual needs, and consequently, the processing results obtained from the subsequent execution of task nodes deviate significantly from the user's actual needs. Summary of the Invention

[0004] Therefore, it is necessary to provide an information processing method, an information processing device, a computer device, a computer-readable storage medium, and a computer program product to address the aforementioned technical problems.

[0005] Firstly, this application provides an information processing method. The method includes:

[0006] Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements and account information of the user accounts; the dialogue statements represent the request intent of the information processing requests.

[0007] Based on the dialogue statement and the account information, a reference processing path is determined from multiple candidate processing paths stored in the preset path library; the candidate processing path is a workflow that connects multiple task nodes in a preset order.

[0008] The information processing request and the reference processing path are input into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model.

[0009] Execute the task nodes of the workflow corresponding to the predicted processing path to obtain the processing results for the information processing request.

[0010] In one embodiment, determining a reference processing path from multiple candidate processing paths stored in a preset path library based on the dialogue statement and the account information includes:

[0011] Based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each of the candidate processing paths, a first reference level is determined for each of the candidate processing paths; and

[0012] Based on the degree of correlation between the account information and the author information corresponding to each candidate processing path, a second reference degree for each candidate processing path is determined; the author information represents the account information of the creator or user corresponding to the candidate processing path.

[0013] Based on the first reference level and the second reference level, a reference processing path is determined from each of the candidate processing paths.

[0014] In one embodiment, prior to determining the first reference level for each of the candidate processing paths, the method further includes:

[0015] The dialogue statement is matched with multiple standard statements stored in a preset first routing table to determine the target standard statement that matches the dialogue statement.

[0016] The target intent associated with the target standard statement is determined in the preset second routing table, and the target intent is used as the request intent for the dialogue statement.

[0017] In one embodiment, determining the first reference level for each of the candidate processing paths based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each of the candidate processing paths includes:

[0018] The request intent is converted into a target space vector, and the candidate space vectors of the processing intents corresponding to each of the candidate processing paths are obtained.

[0019] The similarity between each candidate spatial vector and the target spatial vector is determined, and the similarity is used as the first reference degree of the candidate processing path corresponding to each candidate spatial vector.

[0020] In one embodiment, determining the second reference level of each candidate processing path based on the degree of correlation between the account information and the author information corresponding to each candidate processing path includes:

[0021] Based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path, a confidence score for each candidate processing path is determined; the confidence score characterizes the reliability of the local user account in selecting a candidate processing path as a reference processing path.

[0022] The confidence scores of each candidate processing path are input into a preset amplification function for amplification processing to obtain a second reference level for each candidate processing path.

[0023] In one embodiment, the degree of association is characterized based on weighting coefficients;

[0024] The step of determining the confidence score of each candidate processing path based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path includes:

[0025] Obtain the preset account level of each peer user account; the account level represents the rationality of the candidate processing path for the creation or application corresponding to the peer user account;

[0026] The confidence score of each candidate processing path is determined based on the product of the account level of each peer user account and the corresponding weight coefficient.

[0027] In one embodiment, the first degree of reference is characterized based on a first reference score, and the second degree of reference is characterized based on a second reference score.

[0028] The step of determining a reference processing path from each of the candidate processing paths based on the first reference level and the second reference level includes:

[0029] Obtain the time decay score for each of the candidate processing paths; the time decay score represents the reasonableness of applying the candidate processing path as the reference processing path at the current time;

[0030] A comprehensive reference score is determined for each candidate processing path based on the weighted sum of the first reference score, the second reference score, and the time decay score.

[0031] Based on the comprehensive reference score, the candidate processing paths are sorted in order of size, and the target processing path with the highest preset ranking is selected as the reference processing path.

[0032] In one embodiment, obtaining the time decay score of each of the candidate processing paths includes:

[0033] Obtain the creation time or the most recent application time of each of the candidate processing paths;

[0034] Determine the time distance between the current time and the creation time or the application time;

[0035] Based on a preset time decay function, the time distance of each candidate processing path is evaluated for decay, and a time decay score is obtained for each candidate processing path.

[0036] The time decay fraction is negatively correlated with the time distance.

[0037] In one embodiment, after obtaining the processing result for the information processing request, the method further includes:

[0038] Obtain the user account's rating score for the predictive processing path, the creation time of the predictive processing path, and the user account's account level;

[0039] If the evaluation score is greater than the preset score, the request intent of the information processing request is modified based on the predicted processing path to obtain a new request intent;

[0040] The predicted processing path, the creation time, the new request intent, and the account level are combined into a storage unit, and the storage unit is stored in the path library to update the path library.

[0041] Secondly, this application also provides an information processing apparatus. The apparatus includes:

[0042] The request acquisition unit is configured to execute an information processing request triggered by acquiring a user account; the information processing request carries a dialogue statement and account information of the user account; the dialogue statement represents the request intent of the information processing request.

[0043] The path reference unit is configured to determine a reference processing path from multiple candidate processing paths stored in a preset path library based on the dialogue statement and the account information; the candidate processing path is a workflow that connects multiple task nodes in a preset order.

[0044] The path prediction unit is configured to input the information processing request and the reference processing path into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model.

[0045] The path execution unit is configured to execute the task node of the workflow corresponding to the predicted processing path to obtain the processing result for the information processing request.

[0046] Thirdly, this application also provides a computer device, comprising:

[0047] processor;

[0048] Memory for storing the executable instructions of the processor;

[0049] The processor is configured to execute the executable instructions to implement the information processing method described above.

[0050] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium includes program data, which, when executed, implements the information processing method described above.

[0051] Fifthly, this application also provides a computer program product. The computer program product includes program instructions that, when executed, implement the information processing method described above.

[0052] The aforementioned information processing method, information processing device, computer equipment, computer-readable storage medium, and computer program product first obtain an information processing request triggered by a user account; the information processing request carries a dialogue statement and account information of the user account; the dialogue statement represents the request intent of the information processing request; based on the dialogue statement and account information, a reference processing path is determined from multiple candidate processing paths stored in a preset path library; the candidate processing path is a workflow that connects multiple task nodes in a preset order; the information processing request and the reference processing path are input into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model; the task nodes of the workflow corresponding to the predicted processing path are executed to obtain the processing result for the information processing request. In this way, on the one hand, this solution determines the reference processing path through the user's dialogue statements and account information, thereby facilitating the prediction of the predicted processing path using the reference processing path and the pre-trained language model. The workflow is then executed according to the predicted processing path to obtain the processing result, thus optimizing the information processing flow, effectively improving the efficiency of the predicted processing path, and reducing the consumption of human and material resources. On the other hand, by using a method different from existing technologies, this solution predicts the processing path based on the user-triggered information processing request and the pre-trained language model, thereby optimizing the prediction method of the processing path, effectively improving the rationality of the predicted processing path, and thus reducing the deviation between the final processing result and the user's actual request intent, resulting in a better user experience.

[0053] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0054] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure, and are not intended to unduly limit this disclosure.

[0055] Figure 1 This is an application environment diagram illustrating an information processing method according to an exemplary embodiment.

[0056] Figure 2 This is a flowchart illustrating an information processing method according to an exemplary embodiment.

[0057] Figure 3 This is a flowchart illustrating a step for determining the request intent of a dialogue statement according to an exemplary embodiment.

[0058] Figure 4 This is a flowchart illustrating a step for determining a confidence score of a candidate processing path according to an exemplary embodiment.

[0059] Figure 5 This is a flowchart illustrating a step of determining a reference processing path from candidate processing paths according to an exemplary embodiment.

[0060] Figure 6 This is a flowchart illustrating a step of updating a path library according to an exemplary embodiment.

[0061] Figure 7 This is a flowchart illustrating an information processing method according to another exemplary embodiment.

[0062] Figure 8 This is a block diagram illustrating an information processing method according to another exemplary embodiment.

[0063] Figure 9 This is a block diagram of an information processing apparatus according to an exemplary embodiment.

[0064] Figure 10 This is a block diagram illustrating a computer device for information processing according to an exemplary embodiment. Detailed Implementation

[0065] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0066] The information processing method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located in the cloud or on other network servers. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0067] In some embodiments, reference Figure 1 First, server 104 obtains an information processing request triggered by a user account; the information processing request carries the user account's dialogue statement and account information; the dialogue statement represents the request intent of the information processing request; then, server 104 determines a reference processing path from multiple candidate processing paths stored in a preset path library based on the dialogue statement and account information; the candidate processing path is a workflow that connects multiple task nodes in a preset order; finally, server 104 inputs the information processing request and the reference processing path into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model; then, it executes the task nodes of the workflow corresponding to the predicted processing path to obtain the processing result for the information processing request.

[0068] In one embodiment, such as Figure 2 As shown, an information processing method is provided, which is applied to... Figure 1 Taking the server in the example, the specific steps are as follows:

[0069] Step S11: Obtain the information processing request triggered by the user account.

[0070] In some embodiments, the information processing request is triggered by a server in response to a dialog operation performed by a user account in an end application for a target processing intent.

[0071] In one embodiment, the information processing request carries a dialog statement of the user account and account information. The dialog statement represents the request intent of the information processing request.

[0072] In some embodiments, the dialogue statement is a natural dialogue language statement input by the user account to the user terminal, and the request intent is the processing content expressed in the natural dialogue language statement that the user account expects to achieve.

[0073] As an example, if a user account inputs a dialogue statement to the user terminal as "Arrange a travel activity from Guangzhou to Beijing on XX month XX day", then its request intent can include planning travel routes, modes of transportation, purchasing transportation orders, booking hotel rooms, recommending scenic spots, purchasing scenic spot tickets, etc.

[0074] In some embodiments, the information processing request may also include various information about the user account. For example, this may include the user account's current location information, the user account's name, and the current date information.

[0075] Step S12: Based on the dialogue statements and account information, determine the reference processing path from multiple candidate processing paths stored in the preset path library.

[0076] In one embodiment, the candidate processing path is a workflow that connects multiple task nodes in a preset order.

[0077] For example, candidate processing path A1 is a workflow related to "food", and its task nodes include "target restaurant recommendation", "target restaurant availability query", "target restaurant availability reservation", "target restaurant menu recommendation", and "target restaurant menu reservation" in sequence; candidate processing path A2 is a workflow related to "travel", and its task nodes include "travel destination recommendation", "travel mode query", "purchase transportation order", "book hotel room", "scenic spot recommendation", and "purchase scenic spot ticket" in sequence.

[0078] In one embodiment, the server determines a reference processing path from a plurality of candidate processing paths stored in a preset path library, including the following steps:

[0079] Step 1: Based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each candidate processing path, determine the first reference level for each candidate processing path.

[0080] In some embodiments, the server determines the first reference degree for each candidate processing path, which may include: first converting the request intent into a target space vector and obtaining the candidate space vector of the processing intent corresponding to each candidate processing path; then determining the similarity between each candidate space vector and the target space vector; and finally using the similarity as the first reference degree of the candidate processing path corresponding to each candidate space vector.

[0081] Each candidate processing path has its corresponding processing intent, that is, the working purpose of the corresponding workflow.

[0082] In one embodiment, the request intent of the dialogue statement and the processing intent of the candidate processing path can both be represented based on the corresponding keywords, and each keyword is pre-converted into a corresponding spatial vector.

[0083] For example, if the keyword of request intent X1 is "planning travel routes", it has a corresponding spatial vector P1; if the keyword of processing intent X2 is "booking hotel rooms", it has a corresponding spatial vector P2; and if the keyword of request intent X3 is "recommended scenic spots", it has a corresponding spatial vector P3.

[0084] Step 2: Based on the degree of correlation between account information and the author information corresponding to each candidate processing path, determine the second reference degree of each candidate processing path; whereby author information represents the account information of the creator or user corresponding to the candidate processing path;

[0085] In one embodiment, the degree of correlation between account information and author information corresponding to each candidate processing path can be understood as the degree of correlation between the local user account and the peer user account. This degree of correlation can be represented by a preset weight coefficient or correlation score.

[0086] As an example, if the peer user account to which the candidate processing path belongs is the same as the user account on this end, it indicates that the correlation between the user account on this end and the peer user account is complete, and its preset weight coefficient is 1 (or the correlation score is 100 points); if the peer user account to which the candidate processing path belongs is an associated account of the user account on this end (such as a family account, sub-account, etc.), it indicates that the correlation between the user account on this end and the peer user account is very high, and its preset weight coefficient is 0.8 (or the correlation score is 80 points); if the peer user account to which the candidate processing path belongs is a favorite / subscription account of the user account on this end, it indicates that the correlation between the user account on this end and the peer user account is relatively high, and its preset weight coefficient is 0.6 (or the correlation score is 60 points); if the peer user account to which the candidate processing path belongs has not had any account interaction with the user account on this end, it indicates that the correlation between the user account on this end and the peer user account is very low, and its preset weight coefficient is 0.1 (or the correlation score is 10 points).

[0087] In some embodiments, the server determines a second reference level for each candidate processing path, which may include: first, determining a confidence score for each candidate processing path based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path; and then inputting the confidence scores of each candidate processing path into a preset amplification function for amplification processing to obtain a second reference level for each candidate processing path.

[0088] The confidence score represents the reliability of the local user account in selecting a candidate processing path as a reference processing path, and the confidence score is positively correlated with the reliability of the reference processing path.

[0089] In one embodiment, the server can directly use the weight coefficient or correlation score between the local user account and the peer user account as the confidence score of the corresponding candidate processing path, and then input the confidence scores of each candidate processing path into a preset amplification function for amplification processing.

[0090] Step 3: Based on the first and second reference levels, determine the reference processing path from each candidate processing path.

[0091] In one embodiment, the first reference degree is represented based on a first reference score, and the second reference degree is represented based on a second reference score. The server calculates a weighted sum of the first and second reference scores for each candidate processing path to obtain a comprehensive reference score for each candidate processing path. Finally, the target processing path with the higher comprehensive reference score is selected from the candidate processing paths as the reference processing path.

[0092] Step S13: Input the information processing request and the reference processing path into the pre-trained large language model to obtain the predicted processing path for the request intent output by the large language model.

[0093] In some embodiments, the large language model performs a predictive action on the processing path based on the user account's request intent and account information, and with reference to the task nodes in the input reference processing path, thereby obtaining a predicted processing path for the request intent.

[0094] Among them, large language models can be pre-trained generative models, such as the ChatGPT (Chat Generative Pre-trained Transformer, a chatbot program developed by OpenAI) model, which can generate human-like fluent responses for many downstream tasks (such as task-oriented dialogue and question answering).

[0095] Step S14: Execute the task node of the workflow corresponding to the predictive processing path to obtain the processing result for the information processing request.

[0096] In one embodiment, the server sends the predicted processing path to a pre-trained Executing LLM to execute the task nodes in the corresponding workflow.

[0097] The execution model supports various extended execution plugins, such as data retrieval plugins, script generation plugins, script execution plugins, and data analysis plugins, to complete various task nodes.

[0098] As an example, a predictive processing path includes: task node 1 "querying flight tickets from Guangzhou to Beijing on XX month XX day", task node 2 "purchasing flight tickets from Guangzhou to Beijing on XX month XX day", task node 3 "booking a hotel room in Beijing on XX month XX day", and task node 4 "analyzing recommended tourist attractions in Beijing on XX month XX day". The execution model, through its various extended execution plugins, first queries "flight ticket X1", then purchases "flight ticket X1", then books "hotel room X2", and finally analyzes and obtains "travel recommendation report X3".

[0099] In one embodiment, before the server sends the prediction processing path to the execution model, the prediction processing path can be sent to the client where the user account is located for display, so that the user can modify, edit and confirm the prediction processing path in sequence. After the user triggers the confirmation operation on the prediction processing path, the prediction processing path is then sent to the execution model.

[0100] In the aforementioned information processing process, the server obtains an information processing request triggered by a user account; the information processing request carries the user account's dialogue statement and account information; the dialogue statement represents the request intent of the information processing request; based on the dialogue statement and account information, a reference processing path is determined from multiple candidate processing paths stored in a preset path library; the candidate processing path is a workflow that connects multiple task nodes in a preset order; the information processing request and the reference processing path are input into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model; the task nodes of the workflow corresponding to the predicted processing path are executed to obtain the processing result for the information processing request. In this way, on the one hand, this solution determines the reference processing path through the user's dialogue statements and account information, thereby facilitating the prediction of the predicted processing path using the reference processing path and the pre-trained language model. The workflow is then executed according to the predicted processing path to obtain the processing result, thus optimizing the information processing flow, effectively improving the efficiency of the predicted processing path, and reducing the consumption of human and material resources. On the other hand, by using a method different from existing technologies, this solution predicts the processing path based on the user-triggered information processing request and the pre-trained language model, thereby optimizing the prediction method of the processing path, effectively improving the rationality of the predicted processing path, and thus reducing the deviation between the final processing result and the user's actual request intent, resulting in a better user experience.

[0101] Those skilled in the art will understand that the methods disclosed in the above-described specific embodiments can be implemented in more concrete ways. For example, the implementation described above, in which the server executes the task nodes of the workflow corresponding to the predicted processing path to obtain the processing results for the information processing request, is merely illustrative.

[0102] For example, the server determines the reference processing path from multiple candidate processing paths stored in the preset path library; or the server outputs a predicted processing path for the request intent through a large language model, etc. This is just one way of setting up the process. In actual implementation, there may be other ways of dividing the process. For example, the reference processing path and the predicted processing path may be combined or integrated into another system, or some features may be ignored or not executed.

[0103] In one exemplary embodiment, see Figure 3 , Figure 3 The request to determine the dialogue statements in this application Figure 1 A flowchart illustrating an embodiment. That is, before determining the first reference level for each candidate processing path, the server may also perform the following:

[0104] Step a1: Match the dialogue statement with multiple standard statements stored in the preset first routing table to determine the target standard statement that matches the dialogue statement.

[0105] In some embodiments, a preset routing table stores index information for multiple standard statements, meaning that each standard statement can be extracted through the preset routing table. These standard statements are various common user intent statements configured by the design engineer.

[0106] In some embodiments, dialogue statements are represented based on a first statement vector, and each standard statement is represented based on a corresponding second statement vector.

[0107] In one embodiment, the server first converts the dialogue statement into a first statement vector. Then, it calculates the similarity between the first statement vector and the second statement vector corresponding to each standard statement to determine the degree of similarity between the dialogue statement and each standard statement. Finally, the standard statement corresponding to the second statement vector with the highest similarity is taken as the target standard statement that matches the dialogue statement.

[0108] In other embodiments, the server may also extract keywords from the dialogue statements, then convert the keywords into a first statement vector, and then perform similarity calculations between the first statement vector and the second statement vector corresponding to each standard statement, so as to reduce the influence of too many non-keywords (i.e. words that are not related to the request intent of the dialogue statements) on the converted first statement vector.

[0109] In one embodiment, if the highest similarity is greater than a preset similarity, the standard statement corresponding to the highest similarity is taken as the target standard statement.

[0110] If the similarity between a dialogue statement and a certain standard statement is the highest and greater than the preset similarity, it indicates that the request intent corresponding to the dialogue statement can be represented by the standard statement, that is, the standard statement is taken as the target standard statement relative to the dialogue statement.

[0111] In another embodiment, if the highest similarity is less than or equal to a preset similarity, a response message for the information processing request is generated.

[0112] If the similarity between the dialogue statement and each standard statement does not reach the preset similarity level, it indicates that each standard statement cannot express the request intent corresponding to the dialogue statement. That is, the response information corresponding to the dialogue statement includes one situation: the design engineer has not yet configured a standard statement about the request intent corresponding to the dialogue statement, nor has he configured a candidate processing path corresponding to the request intent.

[0113] Therefore, the server generates response information for the dialogue statements, including response information indicating that the information processing request cannot be processed.

[0114] Step a2: Determine the target intent associated with the target standard statement in the preset second routing table, and use the target intent as the request intent for the dialogue statement.

[0115] In some embodiments, a path library corresponding to the standard statement stores multiple candidate processing paths, and a second routing table stores a target intent associated with each candidate processing path. Therefore, after determining the target standard statement that matches the dialogue statement, the server directly indexes the target intent bound to the target standard statement from the second routing table and uses the target intent as the request intent for the dialogue statement.

[0116] In one exemplary embodiment, see Figure 4 , Figure 4 This is a flowchart illustrating an embodiment of determining the confidence score of candidate processing paths in this application. Specifically, the process by which the server determines the confidence score of each candidate processing path based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path can be executed in the following manner:

[0117] Step b1: Obtain the preset account level of each peer user account.

[0118] In one embodiment, the account level represents the reasonableness of the candidate processing path for the creation or application corresponding to the peer user account.

[0119] In this process, design engineers configure account levels for peer user accounts corresponding to each candidate processing path in the path library based on some prior knowledge.

[0120] As an example, the configured account level is an integer ranging from 0 to 10. A higher account level indicates a higher degree of rationality for the candidate processing path created or applied by the corresponding user account. Specifically, for candidate processing path A1, if the corresponding user account S1 is a recognized technical "expert" in the relevant industry (i.e., the user believes the candidate processing path created or applied by the corresponding user account is highly rational), then the design engineer configures user account S1's account level to 10. For candidate processing path A2, if the corresponding user account S2 is a relatively active non-professional in the relevant industry (i.e., the user believes the candidate processing path created or applied by the corresponding user account is relatively rational), then the design engineer configures user account S2's account level to 8. For candidate processing path A3, if the corresponding user account S2 is a "novice" in the relevant industry (i.e., the user believes the candidate processing path created or applied by the corresponding user account is relatively skewed), then the design engineer configures user account S3's account level to 2.

[0121] Step b2: Determine the confidence score of each candidate processing path based on the product of the account level of each peer user account and the corresponding weight coefficient.

[0122] As an example, if the peer user account S1 has an account level of 2 and its weight coefficient with the local user account S0 is 0.8, then the confidence score for the candidate processing path belonging to the peer user account S1 is 2 × 0.8 = 1.6; if the peer user account S2 has an account level of 5 and its weight coefficient with the local user account S0 is 0.5, then the confidence score for the candidate processing path belonging to the peer user account S2 is 5 × 0.5 = 2.5; if the peer user account S3 has an account level of 8 and its weight coefficient with the local user account S0 is 0.1, then the confidence score for the candidate processing path belonging to the peer user account S3 is 8 × 0.1 = 0.8.

[0123] In one exemplary embodiment, see Figure 5 , Figure 5 This is a flowchart illustrating an embodiment of determining a reference processing path from candidate processing paths in this application. Specifically, the process by which the server determines the reference processing path from each candidate processing path based on a first reference level and a second reference level can be executed in the following manner:

[0124] Step c1: Obtain the time decay score for each candidate processing path.

[0125] In one embodiment, the time decay score characterizes how reasonable it is to apply a candidate processing path as a reference processing path at the current time.

[0126] The process by which the server determines the time decay score for each candidate processing path includes the following steps:

[0127] Step 1: Obtain the creation time or the most recent application time for each candidate processing path.

[0128] Step 2: Determine the time interval between the current time and the creation or application time.

[0129] Step 3: Based on the preset time decay function, evaluate the time distance of each candidate processing path to obtain the time decay score for each candidate processing path. The time decay score is negatively correlated with the time distance.

[0130] As an example, the creation time X1 of candidate processing path P1 is T1, and its time distance from the current time is T2; the creation time X2 of candidate processing path P2 is T2, and its time distance from the current time is T3; the creation time X3 of candidate processing path P3 is T3. Here, T1 > T2 > T3, indicating that candidate processing path P1 has the longest creation time and the least reference significance for the current predicted processing path; candidate processing path P3 has the shortest creation time and the most reference significance for the current predicted processing path. Therefore, the time decay score M1 obtained by evaluating candidate processing path P1 based on the preset time decay function is the smallest, the time decay score M2 obtained by evaluating candidate processing path P2 based on the preset time decay function is at a median size, and the time decay score M3 obtained by evaluating candidate processing path P3 based on the preset time decay function is the largest.

[0131] Step c2: Determine the comprehensive reference score for each candidate processing path based on the weighted sum of the first reference score, the second reference score, and the time decay score.

[0132] The server performs a weighted sum of the first reference score, the second reference score, and the time decay score for each candidate processing path to obtain a comprehensive reference score for each candidate processing path.

[0133] Step c3: Based on the comprehensive reference score, sort the candidate processing paths in order of size, and select the target processing path with the highest preset ranking as the reference processing path.

[0134] The target processing path ranked first in the preset order can be the processing path with the highest comprehensive reference score among all candidate processing paths, or it can be one of the X processing paths with the highest comprehensive reference score. No specific limitation is made here.

[0135] In one exemplary embodiment, see Figure 6 , Figure 6 This is a schematic diagram illustrating an embodiment of updating the path library in this application. Specifically, after the server receives the processing result for the information processing request, it can further perform the following actions:

[0136] Step d1: Obtain the user account's rating score for the predicted processing path, the creation time of the predicted processing path, and the user account's account level.

[0137] The evaluation score is a score manually assigned by the user account based on the processing results of the predicted processing path. That is, if the user account is relatively satisfied with the processing results of the predicted processing path, its evaluation score is higher; if the user account is relatively dissatisfied with the processing results of the predicted processing path, its evaluation score is lower.

[0138] Step d2: If the evaluation score is greater than the preset score, the request intent of the information processing request is modified based on the predicted processing path to obtain a new request intent.

[0139] If the evaluation score is greater than the preset score, it indicates that the user account is relatively satisfied with the processing result of the predicted processing path. Therefore, the predicted processing path can be considered as a qualified prediction with certain reference value. Thus, the server can modify the request intent of the information processing request corresponding to the predicted processing path to obtain a new request intent.

[0140] Step d3 combines the predicted processing path, creation time, new request intent, and account level into a storage unit and stores the storage unit in the path library to update the path library.

[0141] To more clearly illustrate the information processing method provided in the embodiments of this disclosure, the following describes the information processing method in detail with reference to a specific embodiment. In an exemplary embodiment, referring to... Figure 7 and Figure 8 , Figure 7 This is a flowchart illustrating an information processing method according to an exemplary embodiment. Figure 8 This is a block diagram illustrating an information processing method according to an exemplary embodiment. The information processing method is used in a server and specifically includes the following:

[0142] Step S31: Obtain the user's request dialogue input on the control terminal.

[0143] Step S32: Based on the user's needs and dialogue, infer the user's intent to obtain the inferred user intent.

[0144] Intent inference can be obtained based on the classic process of initializing data, which includes the following steps: ① Locate the case library table that needs to be used; ② Generate an executable script for intent inference based on user needs and the case library table; ③ Run the executable script for intent inference to obtain the inferred user intent.

[0145] Step S33: Determine the similarity score, time decay score, and author expertise score for each candidate case in the case library table.

[0146] Candidate cases are pre-made prior workflows.

[0147] The process of determining the similarity score of candidate cases includes: converting the inferred user intent into a target space vector, and calculating the similarity between the target space vector and the space vectors of each candidate case in the case library to obtain the similarity score of each candidate case.

[0148] The process of determining the time decay score of candidate cases includes: obtaining the creation time of each candidate case in the case library, and calculating the time decay score of each candidate case using the creation time and a preset time decay function.

[0149] The time decay function uses a decay rate of 0.01 to exponentiate time, resulting in the time decay score (Recency score).

[0150] The time decay function is a power product of a time variable T and a constant decay rate; the longer the case has been created, the lower its time decay score.

[0151] The process of determining the author's professional score for candidate cases includes: obtaining the professional level of the author of each candidate case in the case library, and calculating the author's professional score for each candidate case using the author's professional level and a preset amplification function.

[0152] Among them, the professional score is the author's score obtained by amplifying the author's professional level through an amplification function, which is the score given to the author in the case.

[0153] The amplification function can be expressed as f(X) = (X×10) / (X×10+1)+1 / 11, X∈(0,10); X refers to the professional level of the author to which the candidate case belongs, that is, the interval from 0 to 10 in which the author's professional level is located.

[0154] In particular, the higher the author's professional score, the higher the confidence level and reference value of the candidate case.

[0155] The author's professional level is preset. For example, if the author of the case study is the user themselves, the author's professional level is level 10; if the author of the case study is an expert in the industry, the author's professional level is level 9; if the author of the case study is an author that the user has saved and followed, the author's professional level is level 8.

[0156] Step S34: Convert all scores of each candidate case into scores in the range of [0-1], and calculate the comprehensive score of each candidate case.

[0157] The overall score of each candidate case is represented by the sum of "similarity score", "time decay score" and "professional score".

[0158] Step S35: Select the top k candidate cases with the highest comprehensive scores from the candidate cases as reference cases, and input these k reference cases into the planning model to output the prediction workflow.

[0159] The Planning LLM is used to integrate and adjust the input reference cases (i.e., reference workflows) and output the predicted workflow.

[0160] Step S36: Present the forecast workflow to the user so that the user can modify and edit the forecast workflow.

[0161] Step S37: In response to the user's confirmation operation on the modified and edited prediction workflow, send the confirmed application workflow to the execution model for execution, and obtain the execution result of the application workflow.

[0162] The Executing LLM supports extended execution plugins such as metadata retrieval plugins, script generation plugins, script execution plugins, and data analysis plugins, which can be used to complete various application tasks in the application workflow.

[0163] Step S38: Return the execution result of the application workflow to the user for viewing, and obtain the user's evaluation score for the application workflow.

[0164] Step S39: If the evaluation score of the application workflow is greater than the threshold, import the application workflow into the case library table to update the case library.

[0165] When updating the case library, the process involves three steps: First, it requires re-inferring user intent based on user needs, dialogues, and confirmed application workflows, and then converting the re-inferred user intent into a corresponding spatial vector. Second, it requires calculating the user's author professional score based on their current professional level. Third, it requires importing the complete application workflow (including the re-inferred user intent and its spatial vector, the application workflow and its creation time, author information, and author professional score) into the case library table for storage.

[0166] In this way, on the one hand, this solution determines the reference processing path through the user's dialogue statements and account information, thereby facilitating the prediction of the predicted processing path using the reference processing path and the pre-trained language model. The workflow is then executed according to the predicted processing path to obtain the processing result, thus optimizing the information processing flow, effectively improving the efficiency of the predicted processing path, and reducing the consumption of human and material resources. On the other hand, by using a method different from existing technologies, this solution predicts the processing path based on the user-triggered information processing request and the pre-trained language model, thereby optimizing the prediction method of the processing path, effectively improving the rationality of the predicted processing path, and thus reducing the deviation between the final processing result and the user's actual request intent, resulting in a better user experience.

[0167] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0168] Based on the same inventive concept, this application also provides an information processing apparatus for implementing the information processing method described above. The solution provided by this information processing apparatus is similar to the solution described in the above method; therefore, the specific limitations in one or more apparatus embodiments provided below can be found in the limitations of the user information identification method and the multi-terminal user interaction method described above, and will not be repeated here.

[0169] In one embodiment, such as Figure 9As shown, an information processing device 10 is provided, including: a request acquisition unit 11, a path reference unit 12, a path prediction unit 13, and a path execution unit 14, wherein:

[0170] The request acquisition unit 11 is configured to execute an information processing request triggered by acquiring a user account; the information processing request carries a dialogue statement and account information of the user account; the dialogue statement represents the request intent of the information processing request.

[0171] The path reference unit 12 is configured to determine a reference processing path from multiple candidate processing paths stored in a preset path library based on the dialogue statement and the account information; the candidate processing path is a workflow that connects multiple task nodes in a preset order.

[0172] The path prediction unit 13 is configured to input the information processing request and the reference processing path into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model.

[0173] The path execution unit 14 is configured to execute the task node of the workflow corresponding to the predicted processing path to obtain the processing result for the information processing request.

[0174] In some embodiments, determining a reference processing path from a plurality of candidate processing paths stored in a preset path library based on the dialogue statement and the account information includes:

[0175] Based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each of the candidate processing paths, a first reference level is determined for each of the candidate processing paths; and

[0176] Based on the degree of correlation between the account information and the author information corresponding to each candidate processing path, a second reference degree for each candidate processing path is determined; the author information represents the account information of the creator or user corresponding to the candidate processing path.

[0177] Based on the first reference level and the second reference level, a reference processing path is determined from each of the candidate processing paths.

[0178] In some embodiments, prior to determining the first reference level for each of the candidate processing paths, the method further includes:

[0179] The dialogue statement is matched with multiple standard statements stored in a preset first routing table to determine the target standard statement that matches the dialogue statement.

[0180] The target intent associated with the target standard statement is determined in the preset second routing table, and the target intent is used as the request intent for the dialogue statement.

[0181] In some embodiments, determining a first reference level for each candidate processing path based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each candidate processing path includes:

[0182] The request intent is converted into a target space vector, and the candidate space vectors of the processing intents corresponding to each of the candidate processing paths are obtained.

[0183] The similarity between each candidate spatial vector and the target spatial vector is determined, and the similarity is used as the first reference degree of the candidate processing path corresponding to each candidate spatial vector.

[0184] In some embodiments, determining the second reference level of each candidate processing path based on the degree of correlation between the account information and the author information corresponding to each candidate processing path includes:

[0185] Based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path, a confidence score for each candidate processing path is determined; the confidence score characterizes the reliability of the local user account in selecting a candidate processing path as a reference processing path.

[0186] The confidence scores of each candidate processing path are input into a preset amplification function for amplification processing to obtain a second reference level for each candidate processing path.

[0187] In some embodiments, the degree of association is characterized based on weighting coefficients;

[0188] The step of determining the confidence score of each candidate processing path based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path includes:

[0189] Obtain the preset account level of each peer user account; the account level represents the rationality of the candidate processing path for the creation or application corresponding to the peer user account;

[0190] The confidence score of each candidate processing path is determined based on the product of the account level of each peer user account and the corresponding weight coefficient.

[0191] In some embodiments, the first degree of reference is characterized based on a first reference score, and the second degree of reference is characterized based on a second reference score.

[0192] The step of determining a reference processing path from each of the candidate processing paths based on the first reference level and the second reference level includes:

[0193] Obtain the time decay score for each of the candidate processing paths; the time decay score represents the reasonableness of applying the candidate processing path as the reference processing path at the current time;

[0194] A comprehensive reference score is determined for each candidate processing path based on the weighted sum of the first reference score, the second reference score, and the time decay score.

[0195] Based on the comprehensive reference score, the candidate processing paths are sorted in order of size, and the target processing path with the highest preset ranking is selected as the reference processing path.

[0196] In some embodiments, obtaining the time decay score of each of the candidate processing paths includes:

[0197] Obtain the creation time or the most recent application time of each of the candidate processing paths;

[0198] Determine the time distance between the current time and the creation time or the application time;

[0199] Based on a preset time decay function, the time distance of each candidate processing path is evaluated for decay, and a time decay score is obtained for each candidate processing path.

[0200] The time decay fraction is negatively correlated with the time distance.

[0201] In some embodiments, after obtaining the processing result for the information processing request, the method further includes:

[0202] Obtain the user account's rating score for the predictive processing path, the creation time of the predictive processing path, and the user account's account level;

[0203] If the evaluation score is greater than the preset score, the request intent of the information processing request is modified based on the predicted processing path to obtain a new request intent;

[0204] The predicted processing path, the creation time, the new request intent, and the account level are combined into a storage unit, and the storage unit is stored in the path library to update the path library.

[0205] In one embodiment, a computer device is provided, which, depending on the program logic to be applied, can be an electronic device or a server, and its internal structure diagram can be as follows: Figure 10 As shown. The computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus.

[0206] The processor of this computer device provides computing and control capabilities. The memory of the computer device includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed, it implements an information processing method. The display screen of the computer device can be an LCD screen or an e-ink display screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, a trackball, or a touchpad located on the casing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0207] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer equipment (including servers and electronic devices) to which the present application is applied. Specific computer equipment may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0208] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed:

[0209] Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements and account information of the user accounts; the dialogue statements represent the request intent of the information processing requests.

[0210] Based on the dialogue statement and the account information, a reference processing path is determined from multiple candidate processing paths stored in the preset path library; the candidate processing path is a workflow that connects multiple task nodes in a preset order.

[0211] The information processing request and the reference processing path are input into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model.

[0212] Execute the task nodes of the workflow corresponding to the predicted processing path to obtain the processing results for the information processing request.

[0213] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, performs the following steps:

[0214] Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements and account information of the user accounts; the dialogue statements represent the request intent of the information processing requests.

[0215] Based on the dialogue statement and the account information, a reference processing path is determined from multiple candidate processing paths stored in the preset path library; the candidate processing path is a workflow that connects multiple task nodes in a preset order.

[0216] The information processing request and the reference processing path are input into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model.

[0217] Execute the task nodes of the workflow corresponding to the predicted processing path to obtain the processing results for the information processing request.

[0218] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0219] Those skilled in the art will understand that embodiments of this application can provide information processing methods, information processing apparatus, servers, computer equipment, computer-readable storage media, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer program instructions (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0220] This application is described with reference to flowchart illustrations and / or block diagrams of information processing methods, information processing apparatuses, servers, computer devices, computer-readable storage media, or computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by the computer program product. These computer program products can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that program instructions executable by the processor of the computer or other programmable data processing apparatus generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0221] These computer program products may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that program instructions stored in the computer program product produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0222] These program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing program instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0223] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0224] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0225] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

[0226] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An information processing method, characterized in that, The method includes: Obtain information processing requests triggered by user accounts; the information processing requests carry dialogue statements and account information of the user accounts; the dialogue statements represent the request intent of the information processing requests. Based on the dialogue statement and the account information, a reference processing path is determined from multiple candidate processing paths stored in a preset path library; the candidate processing path is a workflow that connects multiple task nodes in a preset order; including: determining a first reference degree for each candidate processing path based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each candidate processing path; and determining a second reference degree for each candidate processing path based on the degree of association between the account information and the author information corresponding to each candidate processing path; the author information represents the account information of the creator or user corresponding to the candidate processing path; based on the first reference degree and the second reference degree, a reference processing path is determined from each candidate processing path; The information processing request and the reference processing path are input into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model. Execute the task nodes of the workflow corresponding to the predicted processing path to obtain the processing results for the information processing request.

2. The method according to claim 1, characterized in that, Before determining the first reference level for each of the candidate processing paths, the method further includes: The dialogue statement is matched with multiple standard statements stored in a preset first routing table to determine the target standard statement that matches the dialogue statement. The target intent associated with the target standard statement is determined in the preset second routing table, and the target intent is used as the request intent for the dialogue statement.

3. The method according to claim 2, characterized in that, The determination of a first reference level for each candidate processing path based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each candidate processing path includes: The request intent is converted into a target space vector, and the candidate space vectors of the processing intents corresponding to each of the candidate processing paths are obtained. The similarity between each candidate spatial vector and the target spatial vector is determined, and the similarity is used as the first reference degree of the candidate processing path corresponding to each candidate spatial vector.

4. The method according to claim 1, characterized in that, The determination of the second reference level for each candidate processing path based on the degree of correlation between the account information and the author information corresponding to each candidate processing path includes: Based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path, a confidence score for each candidate processing path is determined; the confidence score characterizes the reliability of the local user account in selecting a candidate processing path as a reference processing path. The confidence scores of each candidate processing path are input into a preset amplification function for amplification processing to obtain a second reference level for each candidate processing path.

5. The method according to claim 4, characterized in that, The degree of correlation is characterized based on weighting coefficients; The step of determining the confidence score of each candidate processing path based on the degree of association between the local user account corresponding to the account information and the peer user account corresponding to the author information of each candidate processing path includes: Obtain the preset account level of each peer user account; the account level represents the rationality of the candidate processing path for the creation or application corresponding to the peer user account; The confidence score of each candidate processing path is determined based on the product of the account level of each peer user account and the corresponding weight coefficient.

6. The method according to claim 1, characterized in that, The first level of reference is characterized based on a first reference score, and the second level of reference is characterized based on a second reference score. The step of determining a reference processing path from each of the candidate processing paths based on the first reference level and the second reference level includes: Obtain the time decay score for each of the candidate processing paths; the time decay score represents the reasonableness of applying the candidate processing path as the reference processing path at the current time; A comprehensive reference score is determined for each candidate processing path based on the weighted sum of the first reference score, the second reference score, and the time decay score. Based on the comprehensive reference score, the candidate processing paths are sorted in order of size, and the target processing path with the highest preset ranking is selected as the reference processing path.

7. The method according to claim 6, characterized in that, The step of obtaining the time decay score for each of the candidate processing paths includes: Obtain the creation time or the most recent application time of each of the candidate processing paths; Determine the time distance between the current time and the creation time or the application time; Based on a preset time decay function, the time distance of each candidate processing path is evaluated for decay, and a time decay score is obtained for each candidate processing path. The time decay fraction is negatively correlated with the time distance.

8. The method according to claim 1, characterized in that, After obtaining the processing result for the information processing request, the method further includes: Obtain the user account's rating score for the predictive processing path, the creation time of the predictive processing path, and the user account's account level; If the evaluation score is greater than the preset score, the request intent of the information processing request is modified based on the predicted processing path to obtain a new request intent; The predicted processing path, the creation time, the new request intent, and the account level are combined into a storage unit, and the storage unit is stored in the path library to update the path library.

9. An information processing device, characterized in that, include: The request retrieval unit is configured to execute an information processing request triggered by retrieving a user account; The information processing request carries the user account's dialogue statements and account information; The dialogue statements represent the intent of the information processing request; The path reference unit is configured to determine a reference processing path from multiple candidate processing paths stored in a preset path library based on the dialogue statement and the account information. The candidate processing path is a workflow that connects multiple task nodes in a preset order; it includes: determining a first reference degree for each candidate processing path based on the degree of matching between the request intent corresponding to the dialogue statement and the processing intent corresponding to each candidate processing path; and determining a second reference degree for each candidate processing path based on the degree of association between the account information and the author information corresponding to each candidate processing path; the author information represents the account information of the creator or user corresponding to the candidate processing path; and determining a reference processing path from each candidate processing path based on the first reference degree and the second reference degree. The path prediction unit is configured to input the information processing request and the reference processing path into a pre-trained large language model to obtain a predicted processing path for the request intent output by the large language model. The path execution unit is configured to execute the task node of the workflow corresponding to the predicted processing path to obtain the processing result for the information processing request.

10. A computer device, characterized in that, include: processor; Memory for storing the executable instructions of the processor; The processor is configured to execute the executable instructions to implement the information processing method as described in any one of claims 1 to 8.

11. A computer-readable storage medium comprising program data, characterized in that, When the program data is executed by the processor of a computer device, the computer device is able to perform the information processing method as described in any one of claims 1 to 8.

Images