Information determination method, device and equipment, computer storage medium and computer program product

Abstract

The embodiment of the invention discloses an information determination method, and the method comprises the steps: obtaining intention information of a user for a to-be-processed object, and obtaining experience information based on the intention information; wherein the experience information is information concerned by the user and related to the to-be-processed object; and determining target prompt information based on the experience information, so that a target large language model determines response information corresponding to the intention information based on the target prompt information. The embodiment of the invention further discloses an information determination device and equipment, a computer storage medium and a computer program product.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a method, apparatus, device, computer storage medium, and computer program product for information determination.

[0002] Scenario Technology

[0003] With the development of technology, intelligent question-and-answer systems based on the "Large Language Model (LLM) + Prompt" model have been increasingly widely used in daily life. In this case, the prompt, as an important tool for guiding the large language model to generate high-quality answers, becomes extremely important. Currently, in related technologies, usually, relevant expert experience information is obtained based on the user's original question, and then the prompt is manually constructed according to the expert experience information and input into the LLM to generate corresponding response information. However, this method of manually constructing the prompt has poor accuracy, which causes the LLM to fail to understand the user's true question intention, thereby affecting the accuracy of the response information generated by the LLM. Summary of the Invention

[0004] To solve the above technical problems, embodiments of this application are expected to provide a method, apparatus, device, computer storage medium, and computer program product for information determination, which solves the problem of inaccurate determined prompts in the process of generating response information in related technologies.

[0005] To achieve the above objective, the technical solution of this application is implemented as follows:

[0006] A method for information determination, the method includes:

[0007] Obtain the intention information of the user for the object to be processed, and obtain the experience information based on the intention information; wherein, the experience information is the information that the user is concerned about and is related to the object to be processed;

[0008] Determine the target prompt information based on the experience information, so that the target large language model determines the response information corresponding to the intention information based on the target prompt information.

[0009] In the above solution, the determining the target prompt information based on the experience information includes:

[0010] Determine the target parameter from multiple candidate parameters; wherein, the experience information includes the multiple candidate parameters;

[0011] Determine the target information of each target parameter from multiple pieces of information of each target parameter;

[0012] In response to the target policy for the prompt information input by the user, based on each target parameter and the target information of each target parameter, determine the target prompt information.

[0013] In the above solution, determining the target parameter from multiple candidate parameters includes:

[0014] Determine the importance level of each candidate parameter;

[0015] Based on the importance level, determine the target parameter from the multiple candidate parameters.

[0016] In the above solution, determining the target parameter from multiple candidate parameters includes:

[0017] Based on the scenario information of the input information corresponding to the intent information, determine the target parameter from the multiple candidate parameters.

[0018] In the above solution, determining the target information of each target parameter from multiple pieces of information of each target parameter includes:

[0019] Obtain the generation time of each piece of information of each target parameter;

[0020] For each target parameter, based on the generation time of each piece of information, determine the target information from the multiple pieces of information of each target parameter.

[0021] In the above solution, in response to the target policy for the prompt information input by the user, based on each target parameter and the target information of each target parameter, determining the target prompt information includes:

[0022] Perform format conversion on each target parameter and the target information of each target parameter according to the target format to obtain the parameters to be used and the information to be used;

[0023] According to the target policy, perform splicing processing on the intent information, the parameters to be used, and the information to be used to obtain the target prompt information.

[0024] An information determination device, the device includes:

[0025] An acquisition unit, configured to acquire the intent information of the user for the object to be processed, and acquire experience information based on the intent information; wherein, the experience information is the information that the user is concerned about and is related to the object to be processed;

[0026] A determination unit, configured to determine the target prompt information based on the experience information, so that the target large language model determines the response information corresponding to the intent information based on the target prompt information.

[0027] An information determination device, the device comprising: a processor, a memory, and a communication bus;

[0028] The communication bus is used to implement a communication connection between the processor and the memory;

[0029] The processor is used to execute an information determination program in the memory to implement the steps of the above information determination method.

[0030] A computer-readable storage medium, the computer-readable storage medium storing one or more programs, the one or more programs being executable by one or more processors to implement the steps of the above information determination method.

[0031] A computer program product, the computer program product comprising a computer program, the computer program implementing the steps of the above information determination method when executed by a processor.

[0032] The information determination method, apparatus, device, computer storage medium, and computer program product provided by the embodiments of the present application can obtain the intention information of the user for the object to be processed, and based on the intention information, obtain the experience information that the user is concerned about and is related to the object to be processed, and determine the target prompt information based on the experience information, so that the target large language model determines the response information corresponding to the intention information based on the target prompt information. In this way, the experience information that the user is concerned about can be obtained according to the user's intention information first, and then the prompt word can be automatically generated according to the experience information, so that the target large language model generates accurate response information based on the prompt word. That is, not only can the prompt word be automatically generated, but also the user's true intention information is considered, rather than constructing the prompt word manually as in the related art. This can not only enable the LLM to fully understand the user's true intention, but also solve the problem of inaccurate determined prompt words in the process of generating response information in the related art. Brief Description of the Drawings

[0033] Figure 1 It is a schematic flow chart of an information determination method provided by an embodiment of the present application;

[0034] Figure 2 It is a schematic flow chart of another information determination method provided by an embodiment of the present application;

[0035] Figure 3 It is a schematic structural diagram of an information determination apparatus provided by an embodiment of the present application;

[0036] Figure 4 It is a schematic structural diagram of an information determination device provided by an embodiment of the present application. Detailed Description of the Embodiments

[0037] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application.

[0038] It should be understood that the "embodiments of the present application" or "the foregoing embodiments" mentioned throughout the specification mean that specific features, structures, or characteristics related to the embodiments are included in at least one embodiment of the present application. Therefore, the "in the embodiments of the present application" or "in the foregoing embodiments" that appear throughout the specification do not necessarily refer to the same embodiments. In addition, these specific features, structures, or characteristics can be combined in one or more embodiments in any suitable manner. In various embodiments of the present application, the magnitude of the serial numbers of the above processes does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application. The serial numbers of the embodiments of the present application above are only for description and do not represent the advantages or disadvantages of the embodiments.

[0039] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0040] The embodiments of the present application provide an information determination method. Referring to Figure 1 as shown, the method includes the following steps:

[0041] Step 101, obtain the intention information of the user for the object to be processed, and obtain the experience information based on the intention information.

[0042] Among them, the experience information is the information that the user is concerned about and is related to the object to be processed;

[0043] In the embodiments of the present application, the object to be processed can be a device, a commodity, or other objects; the intention information can refer to the information that the user wants to know.

[0044] For example: if the object to be processed is a device, the intention information can refer to the system vulnerability information of the device that the user wants to know; if the object to be processed is a commodity, the intention information can refer to the basic information of the commodity that the user wants to know.

[0045] In the embodiments of the present application, the experience information can include multiple parameters related to the object to be processed. For example, if the intention information is to know the vulnerability information of the device, the experience information can include parameters such as the number of attacks on the device, the IP addresses that attacked the device, and the attribute information of the device; if the intention information is to know the basic information of sports shoes, the experience information can include parameters such as the size, material, origin, and color of the sports shoes.

[0046] It should be noted that the experience information can be stored in a database.

[0047] Step 102: Determine target prompt information based on empirical information, so that the target large language model determines the response information corresponding to the intent information based on the target prompt information.

[0048] In the embodiments of the present application, the target prompt information may refer to the prompt words input into the target large language model; the target large language model is the LLM, which may be the Qianwen large model or the Jiutian large model, etc. Specifically, multiple parameters included in the empirical information can be screened first to obtain target parameters with higher value, and then the target prompt information for the intent information can be determined according to the target parameters and the information of the target parameters. After that, the generated target prompt information can be input into the target large language model to obtain the response information corresponding to the intent information.

[0049] For example: If the intent information is that the user wants to know the system vulnerability information of the device, the target prompt information may be "Do you know the following information: 1. The number of times the device has been attacked; 2. The IP addresses that attacked the device; 3. The attribute information of the device; etc. xx information. Please determine what the system vulnerabilities of the device are based on this information".

[0050] The information determination method provided by the embodiments of the present application can first obtain the empirical information concerned by the user according to the intent information of the user, and then automatically generate prompt words according to the empirical information, so that the target large language model generates accurate response information based on the prompt words. That is, it can not only automatically generate prompt words, but also consider the real intent information of the user, rather than constructing prompt words manually as in the related art. This can not only make the LLM fully understand the real intent of the user, but also solve the problem of inaccurate determined prompt words in the process of generating response information in the related art.

[0051] The embodiments of the present application provide another information determination method, which can be applied to an information determination device. Refer to Figure 2 As shown, this method includes the following steps:

[0052] Step 201: The information determination device obtains the intent information of the user for the object to be processed, and obtains empirical information based on the intent information.

[0053] Among them, the empirical information is the information concerned by the user and related to the object to be processed.

[0054] In the embodiments of the present application, the input information of the user for the device to be processed can be obtained first, and then the input information can be analyzed and processed to obtain the real intent information of the user. After that, the intent information of the user can be processed to obtain the empirical information concerned by the user. Furthermore, the corresponding empirical information can be obtained from the database through the application programming interface (API) of the information determination device.

[0055] It should be noted that the empirical information can refer to the specific ideas for answering the intention information. That is to say, prompt words can be constructed based on the obtained specific ideas so that the target large language model can generate corresponding response information according to the empirical information (i.e., specific ideas). And the empirical information can be pre-determined by experts based on historical data.

[0056] Step 202: The information determination device determines the target parameter from multiple candidate parameters.

[0057] Among them, the empirical information includes multiple candidate parameters.

[0058] In the embodiments of the present application, if the intention information is to understand the system vulnerability information of the device, then the multiple candidate parameters may include the number of attacks on the device, the IP addresses that attacked the device, the file headers of the attack files, etc.; if the intention information is to understand the basic information of the sports shoes, then the multiple candidate parameters may include the size, material, origin, color, etc. of the sports shoes.

[0059] In an implementable manner, the target parameter can be directly screened out from the multiple candidate parameters according to the scenario information corresponding to the user's intention information.

[0060] In another implementable manner, the priority of each candidate parameter can be determined first, and then the target parameter can be screened out from the multiple candidate parameters according to the priority of each candidate parameter. Among them, the priority can represent the importance of each candidate parameter.

[0061] In the embodiments of the present application, step 202 can be implemented through step 202a or steps 202b to 202c.

[0062] Step 202a: The information determination device determines the target parameter from the multiple candidate parameters based on the scenario information of the input information corresponding to the intention information.

[0063] In the embodiments of the present application, the input information may refer to information such as questions, instructions, or conversations input by the user; the scenario information may refer to the information of the scenario where the input information is generated. For example, if the intention information is to understand the system vulnerability information of the device, the input information may be "What are the system vulnerabilities of the device?" Then the scenario information may be that the device has been severely attacked by a virus recently, or it may be the daily maintenance of the device every month / week.

[0064] In an implementable manner, if the scenario information is that the device has been attacked by a virus recently, then the parameters related to the attack event can be determined as the target parameters from the multiple candidate parameters, which may include the number of attacks, the attack time, the IP address that initiated the attack, the geographical location distribution of the IP address, and the file headers of the attack files.

[0065] In another implementable manner, if the scenario information is the daily maintenance of the device on a monthly / weekly basis, then the parameter related to the attribute information of the device can be determined from multiple candidate parameters as the target parameter, which may include the system information and operation information of the device, etc.

[0066] In the embodiment of the present application, if the input information is "How many times has a certain IP address appeared?", then the scenario information can be that a certain IP address frequently accesses the device. The multiple candidate parameters may include the number of times the IP address appears and the geographical location where the IP address is located, etc. Obviously, the geographical location where the IP address is located has nothing to do with the scenario of the user's question. That is, the number of times the IP address appears can be directly determined as the target parameter from the multiple candidate parameters.

[0067] However, if the input information is "Has a certain IP address been maliciously exploited?", then the scenario information is that a certain IP address launches a virus attack on the device. In this case, the geographical location where the IP address is located is the necessary information. That is, the parameter of the geographical location where the IP address is located can be determined as the target parameter from the multiple candidate parameters.

[0068] It should be noted that there can be one or more target parameters.

[0069] Step 202b: The information determination device determines the importance level of each candidate parameter.

[0070] In the embodiment of the present application, the priority level, that is, the importance level, of each candidate parameter can be determined manually according to the pre-set evaluation strategy, or a target deep learning model can be used to process each candidate parameter and the scenario information to obtain the importance level corresponding to each candidate parameter.

[0071] In the embodiment of the present application, the importance level can represent the degree of association between each candidate parameter and the user's intention information. That is to say, the higher the degree of association between the candidate parameter and the intention information, the higher its importance level. For example: if the user's intention information is to understand the distribution of IP addresses that launch attacks on the device, then the candidate parameter of the geographical location where the IP address is located has the highest degree of association with the intention information, and its importance level is the highest; while the candidate parameter of the time when the IP address launches an attack has a weaker degree of association with the intention information, and its importance level is lower.

[0072] In the embodiment of the present application, step 202b can be implemented in the following manner:

[0073] a1: The information determination device inputs the scenario information and each candidate parameter into the target deep learning model to obtain the importance level corresponding to each candidate parameter.

[0074] In the embodiments of the present application, the target deep learning model may refer to a causal inference model, or may refer to other deep learning models, and specific limitations are not made here. Specifically, the scenario information and each candidate parameter may be used as input parameters, that is, the scenario information and each candidate parameter are input into the target deep model. Then, the target deep learning model may process the scenario information and each candidate parameter to obtain the importance degree corresponding to each candidate parameter.

[0075] It should be noted that the importance degree may be represented by a numerical value. For example, "1" may indicate the highest importance degree, and "3" may indicate a general importance degree.

[0076] In the embodiments of the present application, the target deep learning model may be trained in the following manner:

[0077] a11. Obtain the sample intention information of the user for the sample object, and determine the sample scenario information of the sample input information corresponding to the sample intention information.

[0078] In the embodiments of the present application, the sample object may refer to a sample device, a sample commodity, etc.; the sample intention information may refer to the information related to the sample object that the user wants to know. For example, if the sample object is a sample device, the sample intention information may be that the user wants to know how to repair the high-risk vulnerability of the device system, or the running status of the device that the user wants to know.

[0079] In the embodiments of the present application, the sample input information may refer to the questions, instructions, or conversations input by the user for the sample object; the sample scenario information may refer to the specific information of the scenario when the user inputs the sample input information.

[0080] Exemplarily, if the sample intention information is that the user wants to know the method of repairing the high-risk vulnerability of the device system, the sample input information may be "How to repair the high-risk vulnerability of the system?", and the sample scenario information may be that a high-risk vulnerability is found in the system during the daily maintenance of the device, or the device has been attacked by a virus.

[0081] a12. Obtain multiple sample parameters based on the sample intention information.

[0082] In the embodiments of the present application, the multiple sample parameters may refer to the information related to the sample object that the user is concerned about, which may be pre-stored in the database. Specifically, after obtaining the sample intention information, the multiple sample parameters corresponding to the sample intention information can be obtained from the database through the API.

[0083] a13. Train the initial deep learning model based on the sample scenario information and the multiple sample parameters to obtain the target deep learning model.

[0084] In an embodiment of the present application, sample scenario information and multiple sample parameters can be used as input parameters, that is, the sample scenario information and multiple sample parameters are input into an initial deep learning model to train the initial deep learning model, so as to obtain a target deep learning model for determining the importance degree corresponding to each candidate parameter.

[0085] Step 202c: The information determination device determines a target parameter from multiple candidate parameters based on the importance degree.

[0086] In an embodiment of the present application, multiple candidate parameters can be sorted according to the importance degree of each candidate parameter in descending order. After that, the top N candidate parameters in the sorted candidate parameters are determined as the target parameters.

[0087] Wherein, N is an integer greater than 0. It should be noted that N can be preset according to historical data and the actual needs of users.

[0088] Step 203: The information determination device determines the target information of each target parameter from multiple pieces of information of each target parameter.

[0089] In an embodiment of the present application, multiple pieces of information of a target parameter may refer to the specific content of the target parameter. For example, if the target parameter is the attacked time of a device, the multiple pieces of information may include 1999 / 1 / 20, 1999 / 2 / 5, 1999 / 3 / 18, etc.; if the target parameter is the IP address that initiates the attack, the multiple pieces of information may include 192.108.174.1x, 192.108.175.1x, 192.109.174.2x, 192.108.176.1x, etc.

[0090] In an embodiment of the present application, for each target parameter, the generation time of each piece of information can be obtained, and the target information is screened from multiple pieces of information according to the generation information of each piece of information.

[0091] In an embodiment of the present application, step 203 can be implemented through steps 203a to 203b.

[0092] Step 203a: The information determination device obtains the generation time of each piece of information of each target parameter.

[0093] In an embodiment of the present application, the generation time of each piece of information of each target parameter can be obtained from a database through an API.

[0094] Step 203b: For each target parameter, the information determination device determines the target information from multiple pieces of information of each target parameter based on the generation time of each piece of information.

[0095] In the embodiments of the present application, for each target parameter, multiple pieces of information can be sorted according to the generation time of each piece of information. After that, the target information can be determined from the multiple pieces of information according to the sorting result of the multiple pieces of information.

[0096] In the embodiments of the present application, step 203b can be implemented in the following manner.

[0097] b1. For each target parameter, the information determination device sorts the multiple pieces of information according to the generation time to obtain the arrangement order corresponding to the multiple pieces of information.

[0098] In the embodiments of the present application, for each target parameter, the multiple pieces of information can be sorted in the order from the latest to the earliest generation time to obtain the arrangement order corresponding to each piece of information.

[0099] Exemplarily, for the target parameter of the attacked time of the device, according to the generation time of this parameter, the multiple attack times can be sorted in the order from the latest to the earliest, that is, the latest attack time is sorted first and the earliest attack time is sorted last. Among them, the arrangement order of the multiple pieces of information can be as shown in Table 1 below.

[0100] Attack time of the device Sorting order 1999 / 1 / 5 4 1999 / 5 / 8 2 1999 / 7 / 16 1 1999 / 2 / 23 3

[0101] Table 1

[0102] b2. For each target parameter, the information determination device determines the target information from the multiple pieces of information based on the arrangement order and the scenario information.

[0103] In the embodiments of the present application, after determining the arrangement order of each piece of information for each target parameter, some candidate information can be first screened out from the multiple pieces of information according to the scenario information corresponding to the intention information. For example, if the scenario information is that the device has been attacked by a virus recently, for the target parameter of the attacked time of the device, the attacked time of the device within the recent week can be screened out from the multiple pieces of information as the candidate information; and if the scenario information is the daily maintenance of the device, the attacked time of the device within the recent month can be screened out from the multiple pieces of information as the candidate information.

[0104] After that, according to the arrangement order of the candidate information, M final target information can be screened out from the multiple candidate information in the order from high to low. Wherein, M is an integer greater than 0, and M can be preset according to historical data and the actual needs of the user.

[0105] Exemplarily, if the candidate information refers to the attack time of the device within the past week, then the top 3 attack times in the sorting order within the past week can be selected as the final target information. Specifically, taking Table 1 as an example, the target information may include July 16, 1999, May 8, 1999, and February 23, 1999.

[0106] In the embodiments of the present application, by screening the obtained multiple candidate parameters and the information of each candidate parameter, candidate parameters with higher value and the target information of the corresponding candidate parameters can be screened out from them. In this way, based on the screened candidate parameters with high value and target information, a more concise target prompt word can be determined, rather than directly adding multiple candidate parameters and the information of each candidate parameter to the target prompt information as in the related art. Thus, after inputting the target prompt word into the target large language model, the target large language model can more easily understand the true intention of the user, and thus generate more accurate response information.

[0107] Step 204: The information determination device, in response to the target policy for the prompt information input by the user, determines the target prompt information based on each target parameter and the target information of each target parameter, so that the target large language model determines the response information corresponding to the intention information based on the target prompt information.

[0108] In the embodiments of the present application, the target policy may refer to the style layout policy of the prompt information. In a feasible manner, the style layout policy may refer to the general - part policy, that is, when generating the prompt information, first write out the information that the user needs to understand, and then write out the screened target parameters and the target information of each target parameter in a general - part form.

[0109] Exemplarily, taking the parameter of the attack time of the device as an example, the target policy can be as shown in Table 2 below:

[0110]

[0111] Table 2

[0112] It should be noted that the target policy can be set manually according to the actual needs of the user.

[0113] In a feasible manner, the format of each target parameter and the target information of each target parameter can be converted first, and the target prompt information is determined according to the target policy, each target parameter, and the target information of each target parameter.

[0114] In another implementable manner, each target parameter and the target information of each target parameter can be used as input parameters, that is, each target parameter and the target information of each target parameter are input into the regression model. Then, the regression model can process each target parameter and the target information of each target parameter according to the target strategy to obtain the target prompt information input into the LLM.

[0115] It should be noted that an eXtreme Gradient Boosting (XGBoost) model or a deep reinforcement learning model can also be used to process each target parameter and each piece of target information according to the target strategy to obtain the target prompt information.

[0116] In the embodiments of the present application, step 204 can be implemented through steps 204a to 204b.

[0117] Step 204a: The information determination device performs format conversion on each target parameter and the target information of each target parameter according to the target format to obtain the parameters to be used and the information to be used.

[0118] In the embodiments of the present application, the target format can refer to the.JSON format. Specifically, each target parameter and the target information of each target parameter can be serialized respectively to obtain the parameters to be used and the information to be used that meet the requirements of the JSON format.

[0119] It should be noted that the target parameter and the parameter to be used only differ in format, and the content is the same.

[0120] Step 204b: The information determination device splices the intent information, the parameters to be used, and the information to be used according to the target strategy to obtain the target prompt information, so that the target large language model determines the response information corresponding to the intent information based on the target prompt information.

[0121] In the embodiments of the present application, after obtaining the parameters to be used and the information to be used, for each parameter to be used, each parameter to be used and the information to be used of each parameter to be used can be stored in the form of key-value pairs ("key": value) to obtain tabular data in the.JSON format.

[0122] It should be noted that each parameter to be used corresponds to a piece of tabular data. Exemplarily, taking the target parameter of the attack time of the device as an example, the converted tabular data can be as shown in Table 3 below:

[0123] Key (identifier) Value 1 1999 / 1 / 5 2 1999 / 5 / 8 3 1999 / 3 / 16 4 1999 / 2 / 23

[0124] Table 3

[0125] After that, the information to be used for each parameter to be used can be summarized to obtain the first information of each parameter to be used. For example: taking the parameter of the attacked time of the device as an example, the first information obtained after summarization is that the device has been attacked for 4 days in the past month; taking the parameter of the IP address initiating the attack as an example, the first information obtained after summarization is that there are 5 IP addresses initiating attacks on the device in the past month.

[0126] Furthermore, according to the pre-set target strategy, the total information of each parameter to be used and the tabular data corresponding to each parameter to be used can be concatenated to obtain the second information of each parameter to be used, which is also called the complete information of the parameter to be used. For example: taking the parameter of the attacked time of the device as an example, the complete information of each parameter to be used obtained after concatenation can be shown in Table 4 below:

[0127] "The device has been attacked for 4 days:

[0128] 1 1999 / 1 / 5 2 1999 / 5 / 8 3 1999 / 3 / 16 4 1999 / 2 / 23

[0129] Table 4"

[0130] After that, according to the target strategy, the intention information of the user and the complete information (i.e., the second information) of each parameter to be used can be concatenated to obtain the target prompt information. It should be noted that if there are multiple target parameters, that is, multiple parameters to be used, the complete information of each parameter to be used can be arranged in descending order according to the priority (i.e., importance) of the target parameters.

[0131] Exemplarily, if the intention information of the user is to understand the system vulnerabilities of the device, the generated target prompt information (i.e., the target prompt word) can be as follows:

[0132] "I want to know what the system vulnerabilities of the device are. Please refer to the following information:

[0133] There are 3 IP addresses initiating attacks on the device within a month:

[0134] 1 192.108.174.1x 2 192.108.175.1x 3 192.109.174.2x

[0135] …

[0136] The device has been attacked for 2 days:

[0137] 1 1999 / 1 / 5 2 1999 / 5 / 8

[0138] "

[0139] In the embodiments of the present application, by splicing the intent information and the complete information of the parameters to be used according to the style layout strategy (i.e., the target strategy), the target prompt information presented in a general-detail format can be obtained. In this way, after using the target prompt information as an input parameter and inputting it into the LLM, the LLM can process it to generate accurate response information for the intent information, that is, the accuracy of the generated response information can be improved.

[0140] The information determination method provided by the embodiments of the present application can first obtain the experience information concerned by the user according to the user's intent information, and then automatically generate a prompt word according to the experience information, so that the target large language model can generate accurate response information based on the prompt word. That is, it can not only automatically generate a prompt word, but also consider the user's true intent information, rather than constructing the prompt word manually as in the related art. This can not only enable the LLM to fully understand the user's true intent, but also solve the problem of inaccurate determined prompt words in the process of generating response information in the related art.

[0141] Based on the foregoing embodiments, the embodiments of the present application provide an information determination device, which can be applied to Figure 1 and 2 the information determination method provided in the corresponding embodiment, as shown in Figure 3 The information determination device 3 may include: an acquisition unit 31 and a determination unit 32, where:

[0142] The acquisition unit 31 is configured to acquire the intent information of the user for the object to be processed, and acquire experience information based on the intent information; wherein, the experience information is the information concerned by the user and related to the object to be processed;

[0143] The determination unit 32 is configured to determine target prompt information based on the experience information, so that the target large language model determines the response information corresponding to the intent information based on the target prompt information.

[0144] In other embodiments of the present application, the determination unit 32 is further configured to perform the following steps:

[0145] Determine a target parameter from multiple candidate parameters; wherein, the experience information includes multiple candidate parameters;

[0146] Determine the target information of each target parameter from the multiple pieces of information of each target parameter;

[0147] In response to the target strategy for the prompt information input by the user, determine the target prompt information based on each target parameter and the target information of each target parameter.

[0148] In other embodiments of the present application, the determination unit 32 is further configured to perform the following steps:

[0149] Determine the importance level of each candidate parameter;

[0150] Based on the importance level, determine the target parameter from multiple candidate parameters.

[0151] In other embodiments of the present application, the determination unit 32 is further configured to perform the following steps:

[0152] Based on the scenario information of the input information corresponding to the intent information, determine the target parameter from multiple candidate parameters.

[0153] In other embodiments of the present application, the determination unit 32 is further configured to perform the following steps:

[0154] Obtain the generation time of each piece of information of each target parameter;

[0155] For each target parameter, based on the generation time of each piece of information, determine the target information from multiple pieces of information of each target parameter.

[0156] In other embodiments of the present application, the determination unit 32 is further configured to perform the following steps:

[0157] Perform format conversion on each target parameter and the target information of each target parameter according to the target format to obtain the parameters to be used and the information to be used;

[0158] According to the target strategy, perform splicing processing on the intent information, the parameters to be used, and the information to be used to obtain the target prompt information.

[0159] It should be noted that the specific descriptions of the steps executed by each unit can be referred to Figure 1 and 2 In the information determination method provided in the corresponding embodiments, details are not described herein again.

[0160] The information determination device provided by the embodiments of the present application can first obtain the experience information concerned by the user according to the user's intent information, and then automatically generate a prompt word according to the experience information, so that the target large language model generates accurate response information based on the prompt word. That is, it can not only automatically generate a prompt word, but also consider the user's true intent information, rather than constructing the prompt word manually as in the related art. This can not only enable the LLM to fully understand the user's true intent, but also solve the problem of inaccurate determined prompt words in the process of generating response information in the related art.

[0161] Based on the foregoing embodiments, the embodiments of the present application provide an information determination device, which can be applied to Figure 1 and 2 In the information determination method provided in the corresponding embodiments, refer to Figure 4As shown in the figure, the information determination device 4 may include: a processor 41, a memory 42, and a communication bus 43, where:

[0162] The communication bus 43 is used to implement a communication connection between the processor 41 and the memory 42;

[0163] The processor 41 is used to execute the information determination program in the memory 42 to implement the following steps:

[0164] Obtain the intention information of the user for the object to be processed, and obtain the experience information based on the intention information; wherein, the experience information is the information that the user is concerned about and is related to the object to be processed;

[0165] Determine the target prompt information based on the experience information, so that the target large language model determines the response information corresponding to the intention information based on the target prompt information.

[0166] In other embodiments of the present application, the processor 41 is used to execute the information determination program in the memory 42 to determine the target prompt information based on the experience information to implement the following steps:

[0167] Determine the target parameter from multiple candidate parameters; wherein, the experience information includes multiple candidate parameters;

[0168] Determine the target information of each target parameter from multiple pieces of information of each target parameter;

[0169] In response to the target strategy for the prompt information input by the user, determine the target prompt information based on each target parameter and the target information of each target parameter.

[0170] In other embodiments of the present application, the processor 41 is used to execute the information determination program in the memory 42 to determine the target parameter from multiple candidate parameters to implement the following steps:

[0171] Determine the importance level of each candidate parameter;

[0172] Determine the target parameter from multiple candidate parameters based on the importance level.

[0173] In other embodiments of the present application, the processor 41 is used to execute the information determination program in the memory 42 to determine the target parameter from multiple candidate parameters to implement the following steps:

[0174] Determine the target parameter from multiple candidate parameters based on the scenario information of the input information corresponding to the intention information.

[0175] In other embodiments of the present application, the processor 41 is used to execute the information determination program in the memory 42 to determine the target information of each target parameter from multiple pieces of information of each target parameter to implement the following steps:

[0176] Obtain the generation time of each piece of information for each target parameter;

[0177] For each target parameter, determine the target information from multiple pieces of information of each target parameter based on the generation time of each piece of information.

[0178] In other embodiments of the present application, the processor 41 is used to execute the information determination program. In response to the target policy for the prompt information input by the user, based on each target parameter and the target information of each target parameter, determine the target prompt information to implement the following steps:

[0179] Perform format conversion on each target parameter and the target information of each target parameter according to the target format to obtain the parameters to be used and the information to be used;

[0180] Perform splicing processing on the intent information, the parameters to be used, and the information to be used according to the target policy to obtain the target prompt information.

[0181] It should be noted that the specific description of the steps executed by the processor can be referred to Figure 1 and 2 In the information determination method provided in the corresponding embodiments, it will not be elaborated here.

[0182] The information determination device provided in the embodiments of the present application can first obtain the experience information concerned by the user according to the user's intent information, and then automatically generate a prompt word according to the experience information, so that the target large language model generates accurate response information based on the prompt word. That is, it can not only automatically generate a prompt word, but also consider the user's true intent information, rather than constructing the prompt word manually as in the related art. This can not only enable the LLM to fully understand the user's true intent, but also solve the problem of inaccurate determined prompt words existing in the process of generating response information in the related art.

[0183] Based on the foregoing embodiments, the embodiments of the present application provide a computer-readable storage medium, which stores one or more programs, and the one or more programs can be executed by one or more processors to implement Figure 1 and 2 the steps of the information determination method provided in the corresponding embodiments.

[0184] Based on the foregoing embodiments, the embodiments of the present application provide a computer program product, which includes a computer program, and the computer program implements Figure 1 and 2 the steps of the information determination method provided in the corresponding embodiments when executed by a processor.

[0185] It should be noted that the above computer-readable storage medium may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a ferromagnetic random access memory (FRAM), a flash memory, a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM), etc.; it may also be various electronic devices including one or any combination of the above memories, such as a mobile phone, a computer, a tablet device, a personal digital assistant, etc.

[0186] It should be noted that in this article, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of another identical element in the process, method, article or device including that element.

[0187] The serial numbers of the embodiments of the present application above are only for description and do not represent the superiority or inferiority of the embodiments.

[0188] Through the description of the above embodiments, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disc) and includes several instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in the various embodiments of the present application.

[0189] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing device produce means for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0190] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0191] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one flow Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

Claims

1. An information determination method, characterized in that The method includes: Obtaining the intention information of the user for the object to be processed, and obtaining experience information based on the intention information; wherein, the experience information is the information that the user is concerned about and is related to the object to be processed; Determining target prompt information based on the experience information, so that the target large language model determines the response information corresponding to the intention information based on the target prompt information.

2. The method according to claim 1, wherein The determining the target prompt information based on the experience information includes: Determining target parameters from multiple candidate parameters; wherein, the experience information includes the multiple candidate parameters; Determining the target information of each target parameter from multiple pieces of information of each target parameter; In response to the target strategy for the prompt information input by the user, determining the target prompt information based on each target parameter and the target information of each target parameter.

3. The method according to claim 2, wherein The determining the target parameters from multiple candidate parameters includes: Determining the importance level of each candidate parameter; Determining the target parameters from the multiple candidate parameters based on the importance level.

4. The method according to claim 2, wherein The determining the target parameters from multiple candidate parameters includes: Determining the target parameters from the multiple candidate parameters based on the scenario information of the input information corresponding to the intention information.

5. The method according to claim 2, wherein The determining the target information of each target parameter from multiple pieces of information of each target parameter includes: Obtaining the generation time of each piece of information of each target parameter; For each target parameter, determining the target information from multiple pieces of information of each target parameter based on the generation time of each piece of information.

6. The method according to claim 2, wherein The in response to the target strategy for the prompt information input by the user, determining the target prompt information based on each target parameter and the target information of each target parameter includes: Performing format conversion on each target parameter and the target information of each target parameter according to the target format to obtain the parameters to be used and the information to be used; Performing splicing processing on the intention information, the parameters to be used and the information to be used according to the target strategy to obtain the target prompt information.

7. An information determination device, characterized in that, The device includes: An obtaining unit, configured to obtain the intention information of the user for the object to be processed, and obtain experience information based on the intention information; wherein, the experience information is the information that the user is concerned about and is related to the object to be processed; A determining unit, configured to determine target prompt information based on the experience information, so that the target large language model determines the response information corresponding to the intention information based on the target prompt information.

8. An information determination device, characterized in that, The device includes: a processor, a memory, and a communication bus; The communication bus is used to implement the communication connection between the processor and the memory; The processor is configured to execute the information determination program in the memory to implement the steps of the information determination method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the information determination method according to any one of claims 1 to 6.

10. A computer program product, the computer program product comprising a computer program, characterized in that, The computer program, when executed by a processor, implements the steps of the information determination method according to any one of claims 1 to 6.

Images