Statement processing method and system and related product

Abstract

The embodiment of the invention discloses a statement processing method and system and a related product, and compared with the prior art, the embodiment of the invention can automatically extract business indexes and related parameters in input statements, retrieve matched target parameters according to the related parameters, and improve the efficiency of statement processing. Therefore, the mapping relation between the input statement and the data required by the system is automatically and quickly established, and the feedback efficiency of the data query result is improved. Wherein the related parameters are grouped according to the combination original meaning of the input statements on the related parameters, so that misunderstanding of intentional demands of the input statements can be effectively avoided, user intentions are clarified in a targeted manner, and the situation that the query range of unneeded data is additionally expanded is avoided; in addition, the grouping basis between the related parameters is followed, and the target parameter set corresponding to each group of related parameters is retrieved according to groups, so that the isolation of each group of information can be kept from the source, and the influence on the reading experience of the user due to redundant feedback of the result is prevented.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of information processing technology, and in particular to a statement processing method, system, and related products. Background Art

[0002] With the popularization of generative artificial intelligence (AI), many businesses have also been integrated with AI to improve intelligence and production efficiency.

[0003] However, even the most intuitive user input still fails to effectively enable intelligent interaction with most businesses. This means it doesn't fundamentally advance the intelligence of the question-answering system. Current AI parsing of user input risks distorting the user's original intent, leading to erroneous results. Specifically, consider a user entering the ambiguous query "Analyze the profitability of a certain company (abbreviated name) in a certain year." Because this query is inherently ambiguous, current AI cannot accurately parse the specific company referred to by the abbreviation or the desired profitability metrics. This can easily lead to misunderstanding the user's original intent, resulting in incorrect results. Summary of the Invention

[0004] The embodiments of the present application provide a statement processing method, system and related products for clarifying the user's input intention and providing efficient, concise and accurate feedback on query results.

[0005] A first aspect of an embodiment of the present application provides a statement processing method, including:

[0006] Extracting the business indicators described by the input sentence and the relevant parameters associated with the business indicators;

[0007] Grouping the related parameters according to the original intention of the input sentence to combine the related parameters;

[0008] For each of the related parameters in each of the groups, a target parameter matching the related parameter is retrieved from a parameter library to form a target parameter set corresponding to the group;

[0009] Feedback to the input end of the input statement is a data query result of each group of the target parameter sets and / or the target parameter sets with respect to the business indicators.

[0010] Optionally, grouping the related parameters according to the original intention of the input sentence to combine the related parameters includes:

[0011] According to the original meaning of the combination between the relevant parameters of the input sentence and the prompt of the preset prompt word, the relevant parameters are arranged and combined following the original meaning of the combination; the combination between the relevant parameters of the example sentence in the preset prompt word conforms to the original meaning of the example sentence.

[0012] Optionally, retrieving a target parameter that matches the relevant parameter from a parameter library includes:

[0013] Retrieving candidate parameters from the main parameter library whose similarity to the relevant parameters meets a preset similarity condition;

[0014] If there are multiple candidate parameters, at least one candidate parameter is selected as the target parameter corresponding to the relevant parameter according to the preset prompt word; the preset prompt word is at least used to indicate the similarity requirement of the candidate parameter as the target parameter.

[0015] Optionally, if no target parameter matching the relevant parameter can be retrieved from the main parameter library, the method further includes:

[0016] Retrieving some parameters from the auxiliary parameter library as candidate parameters; the keyword similarity between the query sentence corresponding to the source of the candidate parameters and the input sentence exceeds the sentence similarity threshold;

[0017] At least one parameter among the candidate parameters is used as the target parameter.

[0018] Optionally, the construction process of the auxiliary parameter library includes:

[0019] Collect multiple query statements within a historical period, and establish a mapping relationship between the query statements and the corresponding parameters associated with the query statements to construct the auxiliary parameter library; wherein, the more times the query statement is selected as the source of the alternative parameter and / or is input in history, the higher the probability that the relevant parameters of the query statement are selected as the alternative parameter.

[0020] Optionally, feeding back to the input end of the input statement each group of the target parameter set and / or the data query result of the target parameter set on the business indicator includes:

[0021] Feedback each set of the target parameter sets to the input end, and collect feedback opinions of the input end on the target parameter sets;

[0022] If the feedback is that the target parameter set meets the expected result, feeding back to the input end the data query result of the target parameter set on the business indicator;

[0023] If the feedback is that the target parameter set does not meet the expected result, at least return to the step of retrieving a target parameter set that matches the relevant parameters from the parameter library until a new target parameter set that meets the expected result is obtained, or stop searching after multiple retrieval failures.

[0024] Optionally, feeding back to the input end of the input statement each group of the target parameter set and / or the data query result of the target parameter set on the business indicator includes:

[0025] Arranging the target parameters in each target parameter set in a tree structure according to the parent-child relationship between the dimensions to which the parameters belong, wherein the target parameters in the same dimension are arranged in the same layer of the tree structure;

[0026] For each group of the tree structures, if there are branches with the same target parameters in the direction from the child node to the root node of the tree structure, the target parameters in the branches with the same target parameters are integrated to obtain a new tree structure with fewer branches;

[0027] The original grouping form between each group of the target parameter sets is continued to be used, and the arrangement parameters in the new tree structure of each group of the target parameter sets are displayed to the input end in a differentiated manner, and / or the data query results corresponding to each group of the arrangement parameters are displayed in a differentiated manner.

[0028] When the method described in the first aspect of this application is specifically implemented, the content described in the second aspect of this application can be adopted to achieve it.

[0029] A second aspect of an embodiment of the present application provides a statement processing system, comprising: an acquisition unit and a processing unit;

[0030] The acquisition unit is used to extract the business indicators described in the input sentence and the relevant parameters associated with the business indicators;

[0031] The processing unit is configured to group the relevant parameters according to the original intention of the combination of the relevant parameters in the input sentence;

[0032] The processing unit is further configured to retrieve, for each of the related parameters in each of the groups, a target parameter that matches the related parameter from a parameter library to form a target parameter set corresponding to the group;

[0033] The processing unit is further configured to feed back to the input end of the input statement each group of the target parameter sets and / or the data query result of the target parameter sets regarding the business indicators.

[0034] A third aspect of an embodiment of the present application provides an electronic device, including: a processor and a memory;

[0035] The processor is configured to communicate with the memory and execute instructions in the memory to implement the method described in the first aspect of the embodiment of the present application or any specific implementation of the first aspect.

[0036] A fourth aspect of an embodiment of the present application provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a processor, they implement the method described in the first aspect of the embodiment of the present application or any specific implementation of the first aspect.

[0037] A fifth aspect of an embodiment of the present application provides a computer program product, which includes computer instructions. When the computer instructions are executed by a processor, they implement the method described in the first aspect of the embodiment of the present application or any specific implementation of the first aspect.

[0038] It can be seen from the above technical solutions that the embodiments of the present application have at least the following advantages:

[0039] Compared with the prior art, the embodiment of the present application can automatically extract the business indicators and their related parameters in the input statement, and retrieve the matching target parameters based on the related parameters, thereby automatically and quickly establishing a mapping relationship between the input statement and the data required by the system, and improving the feedback efficiency of the data query results. Among them, grouping the related parameters based on the original intention of the combination between the related parameters in the input statement can effectively avoid misunderstanding the original intention of the input statement and clarify the user's intention in a targeted manner, so as to avoid additional expansion of the query scope for non-required data; in addition, following the grouping basis between the related parameters, the target parameter set corresponding to each group of related parameters is retrieved by group, which helps to maintain the isolation of each group of information from the source, prevent redundant feedback results, and affect the user's reading experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0041] It should be noted that, although the steps in the process diagrams (if any) involved in the various embodiments are drawn in sequence as indicated by the arrows, unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times. The order of execution of these steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0042] Figure 1 A schematic diagram of a system architecture of the statement processing method according to an embodiment of the present application;

[0043] Figures 2 to 3 Some flowcharts of the statement processing method according to the embodiment of the present application;

[0044] Figures 4 to 6 Schematic diagram of the parameter optimization and integration process of the statement processing method according to an embodiment of the present application;

[0045] Figure 7 A schematic diagram of the structure of the sentence processing system according to an embodiment of the present application;

[0046] Figure 8 This is a structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0048] The terms "first," "second," "third," "fourth," and the like (if any) in the specification and claims of the present application and in the drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusions, e.g., a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product, or apparatus.

[0049] In the following description, reference is made to “a specific embodiment” or “a specific example” and similar expressions, which describe a subset of all possible embodiments, but it is understood that “a specific embodiment” or “a specific example” may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. In the following description, the term “a plurality” refers to at least two. In some specific examples, the numerical value mentioned in this application reaches a threshold value (if any), and the former may be greater than the latter if the latter is greater than the threshold value; if “any” or “at least one” and other similar expressions are mentioned, it may specifically refer to any one of the listed examples or any combination of these examples.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0051] See also Figure 1 , Figure 1 A schematic diagram of an application environment suitable for the embodiment of the present application is shown. The method provided in the embodiment of the present application can be applied to Figure 1 The interactive system 100 shown includes a terminal device 101 and a server 102, which is in communication with the terminal device 101. The server 102 can be a traditional server or a cloud server, which is not specifically limited here. The terminal device 101 can be various electronic devices with a display screen and supporting data input, including but not limited to smartphones, tablet computers, laptop computers, desktop computers, and wearable electronic devices. Specifically, data input can be based on voice input based on a voice module on the terminal device 101, character input based on a character input module, or image input based on an image input module, or it can be based on a gesture recognition module installed on the terminal device 101, so that the user can implement gesture input and other interactive methods. A client application can be installed on the terminal device 101, and the user can communicate with the server 102 based on the client application (such as an APP, a mini-program, etc.), or can also implement the aforementioned data input based on the client application. Similarly, the server 102 may also be deployed with a server application, which may cooperate with the aforementioned client application to implement the present method, for example, to perform relevant processing on data input by the client application, such as returning certain information to the terminal device 101 .

[0052] In the embodiment of the present application, the user can send an input statement to the server 102 through the terminal device 101. Of course, the first input of the user is not necessarily a statement. For example, if the input can be an image, the terminal device 101 can parse the image content to form an input statement to the server 102, or the server 102 can parse the image content to form an input statement. The server 102 can extract the business indicators described by the input statement and the relevant parameters corresponding to the business indicators, and group the relevant parameters according to the original intention of the combination between the relevant parameters in the input statement; for each relevant parameter in each group, the server 102 can retrieve the target parameter that matches the relevant parameter from the main parameter library to form a target parameter set corresponding to the group; the server 102 feeds back the data query results of each group of target parameter sets and / or target parameter sets on the business indicators to the input end of the input statement (i.e., the terminal device 101).

[0053] It should be noted that the method provided in the embodiment of the present application can be implemented jointly by the terminal device and the server as described above, or it can be implemented entirely on the server side, or it can be implemented entirely on the terminal device side. The specific method can be determined according to the actual application scenario and is not limited here.

[0054] The method of this application will be described in further detail below.

[0055] See also Figure 2 In a first aspect, the present application provides a specific embodiment of a statement processing method, which includes the following steps:

[0056] Step S21: extracting the business indicators described in the input sentence and the relevant parameters associated with the business indicators;

[0057] The above-mentioned input sentences can be natural language sentences, which can be specifically formed by the user's natural language habits and part-of-speech rules. The business indicators described by each input sentence and their related parameters can be a one-to-one or one-to-many relationship. Of course, the business indicators described by each input sentence can be one or more, that is, the number of business indicators and related parameters of each input sentence can be unlimited. The method of the embodiment of the present application can be summarized as the operation method of AIAgent (artificial intelligence agent). For example, the business task that the user wants to initiate can be summarized as "analyzing the profitability of ZX Company and HY Company in 2022 and 2023". The user can enter the sentence "Analyzing the profitability of ZX Company and HY Company in 2022 and 2023" in the AIAgent dialog box. After that, the device side or system program required for the operation of AIAgent can extract: the business indicator described by this input sentence is "profitability", and the corresponding related parameters associated with this business indicator are "ZX Company, HY Company, 2022, 2023".

[0058] The above-mentioned AI Agent is an intelligent entity that can perceive the environment, make decisions and perform actions. It can be understood as a computer program or system. It has autonomy, goal orientation, learning ability and logical reasoning ability, and can understand user intentions and perform complex tasks.

[0059] It should be noted that this extraction process can specifically include the word segmentation process of the input sentence. For example, the above-mentioned "profitability" can be used as a word segmentation result. Therefore, the above-mentioned business indicator can be the word segmentation result "profitability", which is a synonym of the word segmentation result, or the word segmentation result itself. That is, the business indicator can use the word segmentation result and its synonyms or one of them. The specific details can be determined by the actual situation and are not limited here. However, the applicant has summarized and found that the input sentence is generally an ambiguous expression, and the expression accuracy of the word segmentation result obtained by word segmentation may not be as good as its synonyms, so the synonyms of the word segmentation result are generally selected as business indicators. In some examples, if the synonyms of the word segmentation result are selected as business indicators, a mapping relationship between the word segmentation result and its synonyms (such as a mapping table) can be pre-built and maintained to quickly match the synonyms of a certain word segmentation result.

[0060] Step S22: group the related parameters according to the original intention of the combination of the related parameters in the input sentence;

[0061] Taking the above input statement "Analyze the profitability of ZX Company and HY Company in 2022 and 2023" as an example, we can analyze the original intention of the combination of relevant parameters in this input statement, that is, the original intention of the combination of multiple parameters. Therefore, in order not to misunderstand the user's original intention, these multiple parameters can form the following four groups of grouping results: ZX Company / 2022, ZX Company / 2023, HY Company / 2022, HY Company / 2023, where the symbol " / " can indicate the meaning of combination, is optional, or can be replaced by other symbols.

[0062] In actual situations, for the input statement "Analyze the profitability of ZX Company in 2022 and the profitability of HY Company in 2023", there are differences in the parameters between multiple indicators in this input statement (such as 2022 is different from 2023), and a short user input statement may have multiple permutations and combinations of dimensional values ​​(i.e., parameters). Therefore, the intention of the embodiment of the present application to follow the original intention of the combination between related parameters is also to help avoid: incorrectly combining unnecessary parameter combinations "ZX Company / 2023, HY Company / 2022", increasing the Cartesian product between the parameters, and thus wasting resources to retrieve relevant information of this unnecessary parameter combination, that is, causing the query scope to be ineffectively expanded, affecting the business process and user experience.

[0063] It can be seen that the embodiment of the present application can ensure that the user's intention is not distorted, and the Cartesian product between the multiple parameters before and after processing remains consistent, which can ensure that the target parameters and query results found subsequently can effectively and reliably meet the user's original intention needs.

[0064] Step S23: for each relevant parameter in each group, retrieve the target parameter that matches the relevant parameter from the main parameter library to form a target parameter set corresponding to the group;

[0065] For the input statement "Analyze the profitability of ZX Company in 2022 and HY Company in 2023," based on the grouping results above, each set of relevant parameters can be clarified separately, and the target parameters that match each relevant parameter can be retrieved from the parameter library. For example, for the grouping result "ZX Company / 2022," the target parameter that matches "ZX Company" might be the company's abbreviation "ZX International Company," and the target parameter that matches "2022" might be "2022-01 to 2022-12." Therefore, the target parameter set corresponding to this grouping result might be "ZX International Company / 2022-01 to 2022-12." Of course, the actual target parameter set might differ, such as "ZX Industrial Company / 2022-01 to 2023-12." In this case, further analysis can be performed to determine the target parameter set that best meets expectations, as detailed below.

[0066] Step S24: Feedback each target parameter set and / or the data query result of the target parameter set on the business indicator to the input end of the input statement.

[0067] As described above, the input statement "Analyze the profitability of ZX and HY companies in 2022 and 2023" can be fed back to the input terminal with the target parameter sets: ZX International Company / 2022-01-2022-12, ZX International Company / 2023-01-2023-12, HY Software Company / 2022-01-2022-12, and HY Software Company / 2023-01-2023-12. Optionally, the AI ​​Agent can use this target parameter set to index relevant data from internal and external sources, such as income and expenditure statements for ZX International Company and HY Software Company for the periods 2022-01-2022-12 and 2023-01-2023-12, and profitability ratings. The AI ​​Agent can then process this data to generate the profitability of ZX and HY companies for 2022 and 2023, respectively, and return it to the input terminal.

[0068] In summary, compared with the prior art, the embodiment of the present application can automatically extract the business indicators and their related parameters in the input statement, and retrieve the matching target parameters based on the related parameters, so as to automatically and quickly establish the mapping relationship between the input statement and the data required by the question-answering system (which can be regarded as quickly realizing interaction), and improve the feedback efficiency of the data query results. Among them, grouping the related parameters based on the original intention of the combination between the related parameters in the input statement can effectively avoid misunderstanding the original intention of the input statement and clarify the user's intention in a targeted manner, so as to avoid additionally expanding the query scope for non-required data and increasing the workload of non-required queries; in addition, following the grouping basis between the related parameters, the target parameter sets are distributed according to the original combination intention, and the target parameter sets corresponding to each group of related parameters are retrieved by group, which helps to keep the isolation of each group of information from the source, prevent redundant feedback results, and affect the user's reading experience.

[0069] Based on the above example descriptions, some specific possible implementation examples will be provided below. In actual applications, the implementation contents between these examples can be combined or implemented separately as needed according to the corresponding functional principles and application logic. If implemented in combination, the execution order between the combined examples can be determined according to their respective processing logic, which can be determined by the actual scenario.

[0070] In some examples, the embodiments of the present application can classify the input statements entered by the user into operational categories, such as query or analysis types. Of course, they can also be classified according to the indicator type to which the business indicator belongs, such as financial or weather scenario types. The query or analysis operation required for the financial input statement can be regarded as initiating a financial analysis business. Taking the financial analysis business as an example, the method of the embodiment of the present application can be summarized as including a clarification (i.e., identification) stage for identifying the user's intention. The specific operations of this stage are: clarifying the business indicators required for financial analysis and the parameters (or dimension values) associated with the indicator. First, the indicator extraction clarification can be performed, and then the dimension values ​​required for the indicator can be clarified. That is, the method of the embodiment of the present application needs to clarify which business indicators and target parameters the input statement wants to reflect or hit; based on the target parameters, a financial analysis business on this business indicator can be finally initiated.

[0071] With reference to the operating instructions for clarifying parameters in each embodiment of the present application, business indicators can be clarified similarly. For example, for the input sentence "Analyze the profitability of ZX Company and HY Company in 2022 and 2023", this input sentence can be segmented and understood in terms of meaning to parse out that the segmentation "profitability" can be used as a business indicator for this input sentence; this parsing result can be obtained because the segmentation "profitability" itself has been recorded in the indicator library in advance, that is, it is a pre-recognized business indicator; or, it is because "profitability" is similar to existing business indicators in the indicator library (such as the synonym "profitability"). Of course, this synonym "profitability" can also replace "profitability" as a business indicator for this input sentence, which can be determined by the needs.

[0072] It should be noted that the order in which the two clarification stages (clarifying business indicators and parameters) are executed can be arbitrary and can be performed simultaneously. For example, for the input statement "Analyze the profitability of ZX Company and HY Company in 2022 and 2023," the relevant parameters of the input statement can be clarified first, or even after clarifying the target parameters, and then the business indicators of this input statement can be clarified. In other words, it can be clarified later that the user intends to initiate a financial analysis of the business indicator "profitability."

[0073] based on Figure 2 Based on the example content, this application provides another specific embodiment of a statement processing method, which includes the following steps:

[0074] Step S21: extracting the business indicators described in the input sentence and the relevant parameters associated with the business indicators;

[0075] Step S22: group the related parameters according to the original intention of the combination of the related parameters in the input sentence;

[0076] In some examples, prompts can be used to better understand the intent of the input sentence and respond accordingly. Therefore, the actual operation process of step S22 may include: according to the original intention of the combination of relevant parameters in the input sentence and the prompt of the preset prompt, the relevant parameters are arranged and combined according to the original intention of the combination; the combination of relevant parameters of the example sentence in the preset prompt is consistent with the original intention of the example sentence. In actual applications, the prompt word and the input sentence can be input into a large language model, so that the recognition ability of the large language model itself can better understand and output the true intention of the input sentence, for example, output the arrangement and combination of relevant parameters according to the original intention of the combination. Of course, the process of selecting at least one prompt word from multiple alternative prompt words to be used in step S22 can also be implemented with the help of the large language model. For example, the alternative prompt word that prompts the model to output the best result can be used as the prompt word for step S22.

[0077] Specifically, the relevant parameters can be arranged and combined according to the original meaning of the combination of semantics and prompt words between the relevant parameters in the input sentence. The preset prompt words used to group the relevant parameters here can be as follows:

[0078] Input: Analysis of HY Company's solvency in January and September 2023

[0079] Parameters = ["org","period"]

[0080] Output: ````json

[0081] [{

[0082] "index":"Debt Solvency"

[0083] "parameter":[

[0084] {

[0085] "org":"HY Company",

[0086] "period":"January 2023"

[0087] },

[0088] {

[0089] "org":"HY Company"

[0090] "period":"September 2023"

[0091] } ]

[0093] ]}

[0094] Among them, index and Parameter represent the business indicators and related parameters described by the input statement, while org and period indicate that the parameter itself or the dimension to which the parameter belongs is organization or period. As can be seen, the above preset prompts provide a certain degree of guidance on information extraction and grouping. For example, the accurate example of combining two sets of related parameters, "HY Company / January 2023" and "HY Company / September 2023," helps AI Agent better understand user intent and accurately clarify the matching relationship between the user's desired related parameters, so as to avoid incorrectly combining unnecessary parameter combinations and ineffectively expanding the query scope.

[0095] Step S23: for each relevant parameter in each group, retrieve the target parameter that matches the relevant parameter from the main parameter library to form a target parameter set corresponding to the group;

[0096] In actual applications, if there are identical related parameters in different groups, and the target parameters corresponding to these related parameters have been retrieved previously, other groups do not need to repeatedly search for the target parameters corresponding to these related parameters, so as to avoid wasting resources and delays. Of course, in special scenarios such as when duplicate costs are not taken into account or when the parameter library has changed, each group can retrieve its own target parameters separately; for identical related parameters, each group can directly use the search results retrieved by itself, or the search results of each group can be combined to determine what target parameters are ultimately matched to this identical related parameter in each group. The specific settings can be made according to actual conditions or needs and are not limited here.

[0097] The parameter library of each embodiment of the present application can be a database consisting of vector results of each parameter, i.e., a vector library. Therefore, when faced with a vector library, one can first obtain the parameter vector results corresponding to the relevant parameters. Then, one can retrieve vector results similar to the parameter vector results from the vector library. The parameters corresponding to the retrieved vector results will have a chance to become the target parameters corresponding to the relevant parameters.

[0098] As described above, grouping related parameters and their target parameters allows subsequent operations to keep the information of each group isolated in groups, so as to avoid misunderstanding the user's original intention and incorrectly combining parameters, thereby ineffectively increasing the query workload and affecting the feedback effect.

[0099] In some examples, the actual operation process of step S22 may include: retrieving from the main parameter library alternative parameters whose similarity with the relevant parameters meets the preset similarity conditions; if there are multiple alternative parameters, according to the prompt of the preset prompt word, select at least one alternative parameter as the target parameter corresponding to the relevant parameter; the preset prompt word is at least used to indicate the similarity requirement of the alternative parameter as the target parameter.

[0100] like Figure 3 Specifically, for each relevant parameter within each group, similar candidate parameters can be retrieved from the main parameter library. These parameters (or options) qualify as candidates if they meet pre-defined similarity criteria: the similarity between this option and the relevant parameter is within the top n (n can be a self-defined positive number) and / or the similarity exceeds a similarity threshold. Thus, for each dimension value within each group, the top 10 options (i.e., candidate parameters) are obtained.

[0101] If there are multiple candidate parameters for each relevant parameter in each group, at least one candidate parameter can be selected as the target parameter corresponding to the relevant parameter using a preset prompt. For example, the top n candidate parameters for each relevant parameter can be used as custom input parameters to call the following prompt, so that at least one parameter that best meets the user's original intention can be selected from the top n candidate parameters as the target parameter. The at least one target parameter here can be fed back to the user in the form of a parameter clarification form, and the top 1 target parameter that best meets the user's original intention can be used as the default option in the parameter clarification form; of course, multiple default options can also be set, and the specific options can be customized and are not limited here.

[0102] In short, the relevant parameters in the user input statement can be extracted, and then the parameter list, i.e., the parameter library, can be traversed for searching to retrieve the alternative parameters that match each relevant parameter in the parameter library. If any relevant parameter has multiple alternative parameters, the result set (i.e., these multiple alternative parameters) that meet the similarity condition (i.e., the similarity with the relevant parameter exceeds the similarity threshold) can be given to the prompt word, so as to use the prompt word to select the alternative parameter that is most likely to meet the user's original intention as the default option (i.e., the target parameter). In some examples, in order to avoid lengthy data, the relevant parameters of the input statement or the clarified target parameters can be de-overlapped and merged separately. Please see below for details; the similarity threshold used for each relevant parameter can be the same or different.

[0103] The preset prompt word prompt for selecting the target parameter from multiple candidate parameters may be as follows:

[0104] Input: What are the profitability and budget capabilities of HY Company in 2022 and ZX Company in 2021?

[0105] parameter_similarity=[{"parameter":"org","similarity_list":[{"value":{"value":"ZX International Corporation","key":1},"similarity":"96%"},

[0106] {"value":{"value":"FT Motors","key":2},"similarity":"90%"},

[0107] {"value":{"value":"HY Software Company","key":3},"similarity":"96%"},

[0108] {"value":{"value":"BD Company","key":4},"similarity":"94%"}]},

[0109] {"parameter":"time","similarity_list":[{"value":{"value":"2021-01~2021-12","key":1},"similarity":"96%"},{"value":{"value":"2022-01~2022-12","key":2},"si milarity":"98%"},{"value":{"value":"2021-01~2023-12","key":3},"similarity":"90%"}]}]

[0110] Output: ```json

[0111] [{

[0112] "org":[{"value":"HY Software Company","key":3}],

[0113] "time":[{"value":"2022-01~2022-12","key":2}]

[0114] },

[0115] {

[0116] "org":[{"value":"ZX International Corporation","key":1}],

[0117] "time":[{"value":"2021-01~2021-12","key":1}]

[0118] }]

[0119] The similarity in the above prompt can be expressed as the similarity between two parameters. The similarity requirement indicated by the prompt in the above example can be summarized as selecting alternative parameters with high similarity as the target parameters corresponding to the relevant parameters. For example, this prompt prompts the selection of parameter combinations with high similarity as output: "HY Software Company / 2022-01~2022-12", "ZX International Company / 2021-01~2021-12".

[0120] The above prompt is only for illustrative purposes. In actual situations, the similarities indicated by parameters of different attributes (or dimensions) such as time and organization (org) may be the same or different. Of course, there may be more than one prompt used in the embodiments of the present application.

[0121] As a possible implementation method, if the target parameter that matches the relevant parameters cannot be retrieved from the main parameter library, the method of the embodiment of the present application may also include: retrieving some parameters from the auxiliary parameter library as alternative parameters; the keyword similarity between the query statement corresponding to the source of the alternative parameters and the input statement exceeds the phrase similarity threshold; and using at least one parameter among the alternative parameters as the target parameter.

[0122] like Figure 3 As shown in the figure, the concept of Retrieval-Augmented Generation (RAG) can be used to introduce external knowledge (such as an auxiliary parameter library) to compensate for the retrieval deficiencies of the main parameter library. For example, if the user input is ambiguous and the relevant parameters (hereinafter referred to as special parameters) cannot be matched in the main parameter library, the auxiliary parameter library can be searched to return the results. Similarly, the process of searching the main parameter library mentioned above can be used to find the top n options in the auxiliary parameter library that meet the similarity as each special parameter in each group as candidate parameters. After that, the target parameter can be selected from the top n candidate parameters in a similar way using the preset prompt words.

[0123] In some specific examples, the construction process of the above-mentioned auxiliary parameter library may include: collecting multiple query statements within a historical period, and establishing a mapping relationship between the query statements and the corresponding parameters associated with the query statements to construct the auxiliary parameter library; wherein, the more times the query statement has been selected as the source of alternative parameters and / or input in history, the higher the probability that the relevant parameters of the query statement will be selected as alternative parameters.

[0124] The auxiliary parameter library can include historically summarized parameters. For example, the auxiliary parameter library can exist in the form of a historical Q&A (Question and Answer) retrieval library, in which there is a mapping relationship between input sentences (or questions) and corresponding answers (i.e., related parameters), so that one can find the other. The relationship between each question and answer can be one-to-one or one-to-many, and the same answer can correspond to different questions. The specific setting can be determined according to the situation and is not detailed here.

[0125] In actual applications, the number of times the questions (i.e., query statements) in the historical Q&A library have been selected as sources of alternative parameters in history can be counted. This is equivalent to counting the number of times the parameters corresponding to each question are retrieved as alternative parameters. Of course, the number of times the parameters corresponding to each question are hit as target parameters can also be counted; in addition, the number of times each question has been input by users as questions in history can be counted. The statistical feedback here is based on the fact that in actual situations, the more query statements are hit and / or asked, the higher the probability that their related parameters are selected as alternative parameters. Furthermore, in some examples, these statistical results can be used to comprehensively determine which alternative parameters are ultimately adopted as target parameters through weighted summation and other forms.

[0126] Step S24: Feedback each target parameter set and / or the data query result of the target parameter set on the business indicator to the input end of the input statement.

[0127] In some specific examples, the specific operation process of step S24 may include (referring to the user feedback mechanism to return a form to the user to confirm the target parameters): feeding back each group of target parameter sets to the input end, and collecting the input end's feedback on the target parameter set; if the feedback is that the target parameter set meets the expected results, feeding back the data query results of the target parameter set about the business indicators to the input end; if the feedback is that the target parameter set does not meet the expected results, at least returning to the step of retrieving the target parameter set that matches the relevant parameters from the main parameter library, until a new target parameter set that meets the expected results is obtained, or stopping the search after multiple retrieval failures.

[0128] For example, a user can enter the sentence "Please help me analyze the profitability of ND Company in January 2023" in the AIAgent dialog box. After that, AIAgent analyzes, diagnoses, and identifies the user's intention, obtains the indicator extraction result "Profitability", and returns it to the user for confirmation: "The indicator "Profitability" that you may want has been matched from the indicator library. Are you sure it is this indicator?" If the user confirms it is this indicator, the corresponding parameter extraction and retrieval are performed, and the retrieved alternative parameters are returned to the user for confirmation: "Please confirm that this alternative parameter "January 2023, ND International Company" is your target parameter?" If the user confirms, the corresponding financial analysis business is initiated. The operational flow of this indicator clarification phase can be summarized as follows: AIAgent recognizes the word "profitability" for the input sentence, and the extracted business indicator can be "profitability." If the user confirms that this business indicator meets expectations, the process proceeds to the specific parameter clarification phase: extracting the relevant parameters described in the input sentence, such as "January 2023, ND International Company," and providing them for the user to confirm. After the user confirms, the AIAgent retrieves the corresponding alternative parameters from the parameter library, allowing the user to confirm the target parameters that meet expectations. Alternatively, AIAgent can directly provide the target parameters for the user to select, without providing alternative parameters for confirmation. The specific details are not limited here. Once the user confirms that the target parameters meet expectations, they can initiate a subsequent financial analysis operation, returning the data query results for the target parameter set related to the business indicator to the input terminal. Conversely, if the user does not confirm that the target parameters meet expectations, the process of the parameter clarification phase can be repeated, or even re-implemented from the beginning, until the target parameters that meet the user's expectations are provided, thereby maximally satisfying user needs.

[0129] Since financial analysis can analyze multiple indicators at the same time, and indicators and dimensions can be a one-to-many relationship, the clarification page display of multiple indicators and multiple dimensions can be optimized and integrated. Therefore, in some specific examples, the specific operation process of step S24 may include (deduplicate integration of each target parameter, or optimization): according to the parent-child relationship between the dimensions to which the parameters belong, the target parameters in each group of target parameter sets are arranged in a tree structure, wherein the target parameters of the same dimension are arranged at the same layer of the tree structure; for each group of tree structures, if there are branches with the same target parameters from the child node to the root node of the tree structure, the target parameters in the branches with the same parameters are integrated to obtain a new tree structure with reduced branches; using the original grouping form between each group of target parameter sets, the arrangement parameters in the new tree structure of each group of target parameter sets are displayed to the input end, and / or the data query results corresponding to each group of arrangement parameters are displayed. The "distinct display" here can refer to displaying different groups of information in different lines and / or colors, which can be set as needed and is not limited here.

[0130] As explained above, after completing the above parameter clarification operation process, that is, clarifying each target parameter, the clarified target parameters can be integrated and optimized in a targeted manner. For example, they can be optimized into a streamlined tree structure, thereby optimizing the display effect of the parameter clarification form, making it easier for users to review or confirm the multi-parameter results under the indicators, that is, improving the readability of large amounts of data in a user-friendly manner, and at the same time helping to reduce data storage space and retrieval chains, and facilitating back-end query and storage of data.

[0131] The principles of the above optimization and integration are: ensuring that the user's intention is not distorted before and after optimization, that is, the user's original intention is not misunderstood, and the Cartesian product between multiple dimensions remains consistent, that is, the parameter results before and after optimization remain unchanged, and no parameters are lost or added; this principle helps to minimize the number of clarification forms for parameter combinations, avoid lengthy forms, and improve the user's reading experience. Figures 4 to 6 The following example illustrates the above optimization and integration process by simulating a complex single-indicator, multi-dimensional value (i.e., multi-parameter) scenario:

[0132] User input: "Please help me analyze the X indicators of Organization 1 Region A, Region B, and Region C in 2023 and 2024, the X indicators of Organization 2 Region E in 2023 and 2024, the X indicators of Organization 3 Region A in 2023, and the X indicators of Organization 3 Region B and Region C in 2023 and 2024."

[0133] Step (1) Construct parameter tree: each layer has different parameter values ​​of the same dimension

[0134] For ease of illustration and understanding, Figures 4 to 6 The target parameters shown are simply taken as the relevant parameters described by the input sentence. According to the operation of step S11 to step S13, the target parameters (or dimension values) of the input sentence in three dimensions can be clarified; the parent-child relationship between the three dimensions can be as follows Figure 4 As shown, "Organization Dimension - Region Dimension - Period Dimension", where the period dimension is the lowest-level leaf node dimension; of course, the parent-child relationship between these three dimensions can be arranged separately as needed, such as "Organization Dimension - Period Dimension - Region Dimension", with no specific restrictions. Accordingly, each dimension value (such as Region A, 2023, etc.) can be arranged into a tree structure according to the parent-child relationship between the dimensions to which it belongs. Among them, each group of dimension values ​​can be arranged into a parameter tree on a group basis. For example, each group can be identified by organization number. These trees can be jointly maintained by a root node (the value of the root node can be empty) to establish a connection. Figure 4 The results of the parameter arrangement before optimization are as follows:

[0135] Organization 1 Region A: 2023, 2024; Organization 1 Region B: 2023, 2024; Organization 1 Region C: 2023, 2024; Organization 2 Region E: 2023, 2024; Organization 3 Region A: 2023; Organization 3 Region B: 2023, 2024; Region C: 2023, 2024

[0136] Step (2) Traverse the integrated subtree from the leaf node to the root node

[0137] Specifically, you can traverse from the leaf nodes (periods) toward the root node and compare whether the node values ​​are identical. Taking the leaf nodes (periods) as an example, each set of period values ​​under the tree, regardless of the number of period values, can be considered a single "leaf node value." For example, you can construct a period list of "2023, 2024" and another period list of "2023." If each value within each set of period lists is identical, then these period lists are considered identical. Similarly, identical region lists can be compared.

[0138] Furthermore, for subtrees composed of nodes of different dimensions, if the values ​​of the subtrees of the nodes under the same parent node are completely equal, then these multiple subtrees can be combined and regarded as the same subtree of the parent node, that is, these multiple subtrees can be regarded as each other's same reference branches. Figure 5 In the first group of trees shown, the subtree values ​​(periods) of the three nodes, region A, region B, and region C, under the parent node organization 1 are exactly the same. Therefore, the three subtrees formed by these three nodes and their child nodes can be regarded as each other's common reference branches, and these three subtrees can be regarded as a subtree of the parent node organization 1 as a whole; region E of the second group of trees is the only child node of the parent node, and its uniqueness does not require merging; the subtree values ​​(periods) of regions B and region C under the parent node organization 3 of the third group of trees are exactly the same. Therefore, the two subtrees formed by these two nodes and their child nodes can be regarded as a subtree of the parent node organization 1 as a whole.

[0139] After the optimization and integration in step (2), the organization 1 tree can be integrated into one, the organization 2 subtree can be integrated into one, and the organization 3 subtree can be integrated into two. If there are still parent nodes in the upper layer of the organization, the organization needs to be merged and optimized, that is, step (2) needs to be repeated in the same way; if there are no other dimensions in the upper layer of the organization, the subtrees can be traversed from the top layer of the tree downwards, and the final integration result is returned to the input end, that is, the following step (3) is implemented.

[0140] Step (3) Traverse the optimized tree from top to bottom to get feedback results

[0141] After the optimization and integration of step (2), we can get Figure 6 The new tree structure after branch reduction is shown. According to the arrangement style of parameter values ​​in each new tree structure, the optimized arrangement results of each group of parameters can be as follows:

[0142] (1) Organization 1 Region A, Region B, Region C 2023, 2024

[0143] (2) Organization 2 Region E 2023, 2024

[0144] (3) Organization 3 Region A2023

[0145] (4) Organization 3 Region B, Region C 2023, 2024

[0146] It can be seen that the results of the parameter arrangement after optimization are better than those of Figure 4 The arrangement results of each group of parameters before optimization are more concise; accordingly, the data query results corresponding to each group of parameters after optimization can also be arranged and displayed in a similar concise manner to facilitate users to review and confirm the results.

[0147] It should be supplemented that the number of indicators that can be extracted and clarified for an input statement is usually less than the number of related parameters. Therefore, in comparison, the parameter clarification stage often has one more deduplication and integration process than the indicator clarification stage. Of course, in actual scenarios or specific needs where the number of indicators is large enough, the indicators can also be deduplicated and integrated in a similar way to reduce the redundancy of indicator information. For example, in some examples, the above-mentioned optimization and integration steps can be implemented similarly between multiple indicators, between related parameters, or between their data query results, so as to reduce the resource loss of each operation flow and improve the query efficiency and reading experience. Among them, multiple indicators can share the same set of parameter integration conditions, such as leaf nodes can be uniformly defined as a certain dimension (such as the period dimension), and a unified optimization grouping principle can be adopted: to ensure that the Cartesian product of the parameters before and after optimization remains unchanged. The specific details of the present invention will not be elaborated in detail.

[0148] In summary, the embodiment of the present application can identify user intentions through the AIAgent intelligent dialog box, match the indicators required for the financial analysis business and the dimensional parameters required for the indicators, and complete the classification initiation of the financial analysis business. Specifically, the user input sentence can be segmented, and indicator extraction and parameter retrieval matching can be performed. The indicators and parameters obtained can be provided to the user for confirmation, which is equivalent to introducing an additional knowledge source of user feedback, helping to establish a mapping between user input and system effective values, simplifying user operations and improving experience. The establishment of a historical Q&A retrieval library can improve the adaptability of business scenarios, improve user experience, improve efficiency, and reduce unnecessary manual operations.

[0149] Mainly, in the parameter extraction stage mentioned above, information isolation can be completed step by step by simplifying the prompt word responsibilities to avoid the increase of the Cartesian product between parameters and expand the query scope; by constructing a dimension tree for flashback traversal, the parameter clarification form set can be optimized to simplify user operations and reading difficulty; establishing historical Q&A data retrieval can improve the adaptability of business scenarios and enhance user experience. In summary, the embodiment of the present application solves the problem of the inconvenience of quickly establishing a mapping between user input and data values ​​inside and outside the system, improves the efficiency of AI dialog box initiation of business and the intelligence level of the system, and at the same time, the information isolation operation avoids the clarification operation being too lengthy due to multi-dimensional parameters.

[0150] Compared to Figure 2 The examples shown indicate that the above-mentioned additional or detailed example contents or possible implementation methods do not necessarily have to be executed during specific implementation. If more than two examples or possible implementation methods are added, these examples or possible implementation methods can be implemented in combination or separately. If implemented in combination, the execution order of the combined examples can be determined according to their respective processing logic, which may depend on the actual scenario.

[0151] See also Figure 7 , a second aspect of the present application provides a specific embodiment of a sentence processing system, the system comprising: an acquisition unit 701 and a processing unit 702;

[0152] The acquisition unit 701 is used to extract the business indicators described by the input sentence and the relevant parameters associated with the business indicators;

[0153] The processing unit 702 is configured to group the related parameters according to the original meaning of the combination between the related parameters in the input sentence;

[0154] The processing unit 702 is further configured to retrieve, for each of the related parameters in each of the groups, a target parameter that matches the related parameter from a parameter library to form a target parameter set corresponding to the group;

[0155] The processing unit 702 is further configured to feed back to the input end of the input statement each group of the target parameter sets and / or the data query result of the target parameter set regarding the business indicator.

[0156] In some examples, the processing unit 702 is specifically configured to:

[0157] According to the original meaning of the combination between the relevant parameters of the input sentence and the prompt of the preset prompt word, the relevant parameters are arranged and combined following the original meaning of the combination; the combination between the relevant parameters of the example sentence in the preset prompt word conforms to the original meaning of the example sentence.

[0158] In some examples, the processing unit 702 is specifically configured to:

[0159] Retrieving candidate parameters from the main parameter library whose similarity to the relevant parameters meets a preset similarity condition;

[0160] If there are multiple candidate parameters, at least one candidate parameter is selected as the target parameter corresponding to the relevant parameter according to the preset prompt word; the preset prompt word is at least used to indicate the similarity requirement of the candidate parameter as the target parameter.

[0161] In some examples, if no target parameter matching the relevant parameter can be retrieved from the main parameter library, the processing unit 702 is further configured to:

[0162] Retrieving some parameters from the auxiliary parameter library as candidate parameters; the keyword similarity between the query sentence corresponding to the source of the candidate parameters and the input sentence exceeds the sentence similarity threshold;

[0163] At least one parameter among the candidate parameters is used as the target parameter.

[0164] In some examples, the processing unit 702 is specifically configured to:

[0165] Collect multiple query statements within a historical period, and establish a mapping relationship between the query statements and the corresponding parameters associated with the query statements to construct the auxiliary parameter library; wherein, the more times the query statement is selected as the source of the alternative parameter and / or is input in history, the higher the probability that the relevant parameters of the query statement are selected as the alternative parameter.

[0166] In some examples, the processing unit 702 is specifically configured to:

[0167] Feedback each set of the target parameter sets to the input end, and collect feedback opinions of the input end on the target parameter sets;

[0168] If the feedback is that the target parameter set meets the expected result, feeding back to the input end the data query result of the target parameter set on the business indicator;

[0169] If the feedback is that the target parameter set does not meet the expected result, at least return to the step of retrieving a target parameter set that matches the relevant parameters from the parameter library until a new target parameter set that meets the expected result is obtained, or stop searching after multiple retrieval failures.

[0170] In some examples, the processing unit 702 is specifically configured to:

[0171] Arranging the target parameters in each target parameter set in a tree structure according to the parent-child relationship between the dimensions to which the parameters belong, wherein the target parameters in the same dimension are arranged in the same layer of the tree structure;

[0172] For each group of the tree structures, if there are branches with the same target parameters in the direction from the child node to the root node of the tree structure, the target parameters in the branches with the same target parameters are integrated to obtain a new tree structure with fewer branches;

[0173] The original grouping form between each group of the target parameter sets is continued to be used, and the arrangement parameters in the new tree structure of each group of the target parameter sets are displayed to the input end in a differentiated manner, and / or the data query results corresponding to each group of the arrangement parameters are displayed in a differentiated manner.

[0174] In the embodiment of the present application, the operations performed by each unit of the statement processing system are similar to the operations described in the first aspect or any specific method embodiment of the first aspect, and will not be described in detail here. Of course, the specific implementation process of each operation in the first aspect of the present application can also be implemented by referring to the relevant description of the second aspect.

[0175] Among them, the parameter integration process is to construct the parameters of each dimension into a tree: between single indicators and multiple dimensions, branches with the same dimension values ​​are merged; similarly, between multiple indicators, branches are merged only when all dimension values ​​are equal.

[0176] See also Figure 8 The electronic device 800 of the embodiment of the present application may include one or more processors (such as central processing units (CPU)) 801 and a memory 805, in which one or more applications or data are stored.

[0177] Memory 805 may be volatile or persistent storage. The program stored in memory 805 may include one or more modules, each of which may include a series of instruction operations on the electronic device. Furthermore, processor 801 may be configured to communicate with memory 805 to execute the series of instruction operations in memory 805 on electronic device 800.

[0178] The electronic device 800 may also include one or more power supplies 802, one or more wired or wireless network interfaces 803, one or more input and output interfaces 804, and / or one or more operating systems, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0179] The processor 801 can execute the operations performed by the aforementioned first aspect or any specific method embodiment of the first aspect, and the details will not be repeated here.

[0180] The present application provides a computer-readable storage medium, comprising instructions. When the instructions are executed on a computer, the computer executes the method described in the first aspect or any specific implementation of the first aspect.

[0181] The present application provides a computer program product comprising instructions or a computer program. When the computer program product is run on a computer, the computer is caused to execute the method described in the first aspect or any specific implementation of the first aspect.

[0182] It is understood that in the various embodiments of the present application, the sequence number of each step does not mean the order of execution. The order of execution of each step should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The operation content added or refined in each example scheme of the above-mentioned method, system or device (if any) does not necessarily have to be executed in the specific implementation. If two or more operations are added, these operations can be implemented in combination or separately, depending on the actual scenario.

[0183] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system (if any) and device can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0184] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system or device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection of some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0185] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0186] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0187] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product (or computer program product) is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a business server or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

Claims

1. A statement processing method, characterized in that: include: Extracting the business indicators described by the input sentence and the related parameters associated with the business indicators; Grouping the related parameters according to the original intention of the input sentence to combine the related parameters; For each of the related parameters in each of the groups, a target parameter matching the related parameter is retrieved from a parameter library to form a target parameter set corresponding to the group; Feedback to the input end of the input statement is a data query result of each group of the target parameter sets and / or the target parameter sets with respect to the business indicator.

2. The sentence processing method according to claim 1, characterized in that: The grouping of the related parameters according to the original intention of the input sentence to combine the related parameters includes: According to the original meaning of the combination between the relevant parameters of the input sentence and the prompt of the preset prompt word, the relevant parameters are arranged and combined following the original meaning of the combination; the combination between the relevant parameters of the example sentence in the preset prompt word conforms to the original meaning of the example sentence.

3. The sentence processing method according to claim 1, characterized in that: The step of retrieving a target parameter matching the relevant parameter from a parameter library includes: Retrieving from the main parameter library candidate parameters whose similarity with the relevant parameters meets a preset similarity condition; If there are multiple candidate parameters, at least one of the candidate parameters is selected as the target parameter corresponding to the relevant parameter according to the preset prompt word; the preset prompt word is at least used to indicate the similarity requirement of the candidate parameter as the target parameter.

4. The sentence processing method according to claim 3, characterized in that: If no target parameter matching the relevant parameter can be retrieved from the main parameter library, the method further comprises: Retrieving some parameters from the auxiliary parameter library as candidate parameters; the keyword similarity between the query sentence corresponding to the source of the candidate parameters and the input sentence exceeds the sentence similarity threshold; At least one parameter among the candidate parameters is used as the target parameter.

5. The sentence processing method according to claim 4, characterized in that: The construction process of the auxiliary parameter library includes: Collect multiple query statements within a historical period, and establish a mapping relationship between the query statements and the corresponding related parameters associated with the query statements to construct the auxiliary parameter library; wherein, the more times the query statement is selected as the source of the candidate parameter and / or is input in history, the higher the probability that the related parameters of the query statement are selected as the candidate parameter.

6. The sentence processing method according to any one of claims 1 to 5, characterized in that: Feedback to the input end of the input statement of each group of the target parameter set and / or the data query result of the target parameter set on the business indicator includes: Feedback each group of the target parameter sets to the input end, and collect feedback opinions of the input end on the target parameter sets; If the feedback is that the target parameter set meets the expected result, feeding back to the input end the data query result of the target parameter set on the business indicator; If the feedback is that the target parameter set does not meet the expected result, at least return to the step of retrieving a target parameter set matching the relevant parameters from the parameter library until a new target parameter set meeting the expected result is obtained, or stop searching after multiple failed searches.

7. The sentence processing method according to any one of claims 1 to 5, characterized in that: Feedback to the input end of the input statement of each group of the target parameter set and / or the data query result of the target parameter set on the business indicator includes: Arranging the target parameters in each target parameter set in a tree structure according to the parent-child relationship between the dimensions to which the parameters belong, wherein the target parameters in the same dimension are arranged in the same layer of the tree structure; For each group of the tree structures, if there are branches with the same target parameters in the direction from the child node to the root node of the tree structure, the target parameters in the branches with the same target parameters are integrated to obtain a new tree structure with fewer branches; The original grouping form between each group of the target parameter sets is continued to be used, and the arrangement parameters in the new tree structure of each group of the target parameter sets are displayed to the input end in a differentiated manner, and / or the data query results corresponding to each group of the arrangement parameters are displayed in a differentiated manner.

8. A sentence processing system, characterized in that: include: Acquisition unit and processing unit; The acquisition unit is used to extract the business indicator described by the input sentence and the relevant parameters associated with the business indicator; The processing unit is used to group the related parameters according to the original intention of the combination between the related parameters in the input sentence; The processing unit is further configured to retrieve, for each of the related parameters in each of the groups, a target parameter matching the related parameter from a parameter library to form a target parameter set corresponding to the group; The processing unit is further configured to feed back to the input end of the input statement each group of the target parameter sets and / or the data query result of the target parameter set regarding the business indicator.

9. An electronic device, characterized in that: include: Processor and memory; The processor is configured to communicate with the memory and execute instructions in the memory to implement the method according to any one of claims 1 to 7.

10. A readable storage medium, characterized in that: The readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

11. A computer program product, characterized in that The computer program product comprises computer instructions, which implement the method according to any one of claims 1 to 7 when executed by a processor.

Images