A method and system for improving completion parameters for understanding user input
By generating semantic labels and using an improved Levenstein distance algorithm, combined with historical click data and association graphs, the problems of semantic disambiguation and parameter dependency management are solved, enabling accurate understanding and efficient sorting of user input, and improving the interpretability and efficiency of industrial manufacturing decisions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INFORMATION TECH BOTE INTELLIGENT TECH CO LTD
- Filing Date
- 2026-02-28
- Publication Date
- 2026-05-29
AI Technical Summary
Existing technologies have weak semantic disambiguation capabilities, resulting in inaccurate semantic labels generated in polysemous scenarios, untraceable derivation processes, inefficient parameter dependency management, inability to effectively identify key parameters and detect circular dependencies, and unbalanced candidate parameter ranking, which affects user experience.
By generating semantic tags, the similarity between candidate parameters and user sessions is calculated using an improved Levenstein distance algorithm. Parameters are sorted by combining historical click data and association graphs to construct a traceable derivation path, identify key parameters, detect circular dependencies, and optimize inference efficiency.
It achieves accurate semantic tag generation, improves the consistency of user input intent, meets the interpretability requirements of decision-making in industrial scenarios, optimizes parameter dependency identification and sorting, and improves the efficiency and accuracy of industrial manufacturing.
Smart Images
Figure CN122114174A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of industrial manufacturing technology, and more specifically, to a method and system for improving the understanding of user input completion parameters. Background Technology
[0002] Methods to improve the understanding of user input and complete parameters refer to optimizing algorithms and technologies to enhance the system's ability to understand and respond to user input. These methods typically involve technologies such as natural language processing, context analysis, and semantic understanding. The goal of parameter completion is to provide real-time suggestions and complete industrial parameters during the industrial manufacturing process, thereby improving the efficiency of industrial manufacturing and enabling the system to meet user needs more accurately and efficiently when interacting with users. By improving its ability to understand language and expressions, it can better serve users.
[0003] However, the main problem in existing technologies is the weak semantic disambiguation capability. Traditional technologies that rely on coarse-grained keyword matching are prone to ambiguity when dealing with polysemous words, resulting in inaccurate semantic labels and ambiguity in polysemous word scenarios. Secondly, the lack of traceability in the derivation process challenges the transparency and credibility of decision-making. In areas involving compliance, the lack of reverse constraint verification mechanisms makes it impossible to trace the source and derivation steps of the output results.
[0004] Meanwhile, the problem of inefficient parameter dependency management is also very prominent. Existing technologies cannot effectively identify key parameters and detect circular dependencies, which may lead to logical conflicts and inefficiency during the reasoning process. In addition, the problem of unbalanced candidate parameter ranking often causes the ranking results to deviate from the user's actual needs, and cannot effectively take into account the importance of the current session and the dynamic changes of historical preferences, thus affecting the user experience. Summary of the Invention
[0005] To overcome the above-mentioned deficiencies of the prior art, embodiments of the present invention provide a method and system for improving the understanding of user input completion parameters.
[0006] To achieve the above objectives, the present invention provides the following technical solution:
[0007] Firstly, a method is provided to improve the understanding of user input completion parameters, the method comprising:
[0008] S101: Input an industrial standardization document, generate semantic tags, perform reasoning based on the industrial standardization document and semantic tags to obtain an optimized industrial standardization document, and obtain an industrial rule set and a set of text parameters input by the user. Derive the first parameter set from the set of text parameters input by the user, and construct an association graph based on the first parameter set and historical sessions.
[0009] The set of text parameters input by the user is the set of user instructions and industrial parameters received by the method;
[0010] S102: Extract multiple candidate parameters from the first parameter set, filter and weight them using historical click data to obtain the weight values of the candidate parameters, sort them based on the weight values of the candidate parameters, and obtain the sorted candidate parameter table.
[0011] S103: Receive and store the user's dialogue content, parse the user's input text parameter set to obtain the second parameter set, and generate the context parameter set by linking the second parameter set and the association graph.
[0012] S104: Complete and provide feedback based on the sorted candidate parameter table and context parameter set.
[0013] Furthermore, the generation of semantic tags includes:
[0014] Extract the part-of-speech tags of user conversation words and label them to obtain conversation vectors with part-of-speech tags. Query the synonym list and calculate the path similarity between the conversation vector and the synonym set in the synonym list.
[0015] Take the maximum similarity of each word in the conversation vector, and then calculate the average to get the final similarity of the synonym set;
[0016] Compare the final similarity scores of all synonym sets and sort them from largest to smallest. Select the synonym set corresponding to the first final similarity score as the disambiguation result to obtain the semantic label.
[0017] Furthermore, the derivation of the set of text parameters input by the user includes:
[0018] Scan the industrial rule set, bind the set of text parameters input by the user with the rules in the industrial rule set to obtain the first parameter corresponding to the rule, and continue to trigger subsequent rules based on the first parameter to obtain the derivation parameter set;
[0019] Then, based on the set of text parameters input by the user, reverse derivation is performed to obtain the preconditions and verify them, thus obtaining the derivation path. The derivation parameter set and the derivation path constitute the first parameter set.
[0020] Further, obtaining the weight values of the candidate parameters includes:
[0021] The similarity between candidate parameters and user sessions is calculated using an improved Levenstein distance algorithm. A similarity threshold of n is set, and candidate parameters with similarity ≥ n are selected. The parameters are then filtered according to rule priority to generate a candidate parameter set.
[0022] Perform local and global calculations separately to obtain local and global calculation results. Multiply the local and global calculation results to obtain the initial weight values of the candidate parameters.
[0023] The click-through rate is calculated based on historical click data, and the time decay factor is derived using an exponential decay function.
[0024] The initial weight value, click-through rate, and time decay factor of the candidate parameter are multiplied together to obtain the weight value of the candidate parameter.
[0025] Furthermore, the calculation of the similarity between candidate parameters and user sessions using the improved Levenstein distance algorithm includes:
[0026] Extract the user's editing operation sequence for candidate parameters from historical click data, and calculate the frequency of each editing operation at different character positions based on the editing operation sequence to obtain the operation frequency matrix and operation weight table;
[0027] Each candidate parameter is split and represented as a two-dimensional vector to generate a parameter matrix;
[0028] Create a distance matrix between candidate parameters, and perform in-field editing calculations using the operation weight table to obtain the edit distance matrix. The specific calculation formula is as follows:
[0029]
[0030] In the formula: This represents the dynamic edit distance between the i-th field of parameter A and the j-th field of parameter B, where i and j are the field indices of the distance matrix between candidate parameters. , , These represent the weights for field deletion, field insertion, and field replacement operations, respectively. This indicates the specific character value of parameter A / B at field i / j;
[0031] The semantic features of the candidate parameters are extracted, the candidate parameters are converted into natural language descriptions, the semantic similarity between the candidate parameters is calculated using cosine similarity, and compensation is performed on the edit distance matrix based on the semantic similarity to obtain the compensation distance matrix.
[0032] Check whether the candidate parameters match historical high-frequency click data, and add preference scores to the candidate parameters that match;
[0033] The similarity is calculated by combining preference bonuses and compensation distance matrices, and the formula is as follows:
[0034]
[0035] The following are the separate local and global calculations:
[0036] The local calculation is: the number of times the candidate parameter appears in this session divided by the total number of parameters;
[0037] The overall calculation is as follows: the total number of sessions divided by the number of sessions including candidate parameters, and then the logarithm is taken.
[0038] Furthermore, the specific steps for parsing the user-input text parameter set include:
[0039] Type validation is performed on each parameter in the text parameter set input by the user to obtain the validated parameter set. The validated parameter set is then combined with the rules in the industrial rule set to derive the derivation result. The derivation result is then added to the validated parameter set to obtain the variable parameter set.
[0040] By querying the association results of the variable parameter set through the association graph, the association results are added to the variable parameter set to obtain the second parameter set.
[0041] Furthermore, the linkage based on the second parameter set and the correlation graph includes:
[0042] The historical session content is compared with the industrial rule set to obtain extended rules. The extended parameters corresponding to the extended rules are added to the second parameter set to obtain the extended parameter set.
[0043] By querying the attributes corresponding to the extended parameter set through the association graph and binding them to the extended parameter set, the context parameter set can be obtained.
[0044] Secondly, a system for improving the understanding of user input completion parameters is provided, which is implemented based on the above-mentioned method for improving the understanding of user input completion parameters, the system comprising:
[0045] The basic framework module takes an industrial standardization document as input, generates semantic tags, and performs inference using the industrial standardization document and semantic tags to derive an optimized industrial standardization document. It also obtains the industrial rule set and the text parameter set input by the user, derives the first parameter set from the text parameter set input by the user, and constructs an association graph based on the first parameter set and historical sessions.
[0046] The initial parameter module extracts multiple candidate parameters from the first parameter set, combines them with historical click data for filtering and weighted calculation to obtain the weight values of the candidate parameters, and sorts them based on the weight values of the candidate parameters to obtain the sorted candidate parameter table;
[0047] The text storage module receives and stores the user's dialogue content, parses the user's input text parameter set to obtain a second parameter set, and generates a context parameter set based on the second parameter set and the association graph.
[0048] The parameter completion module completes and provides feedback based on the sorted candidate parameter table and context parameter set.
[0049] Thirdly, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor, when executing the computer program, implements the aforementioned method for improving the understanding of user input completion parameters.
[0050] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed, implements the method described above for improving the understanding of user input completion parameters.
[0051] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0052] This application discloses a method and system for improving the understanding of user input completion parameters, including: inputting an industrial standardization document, generating semantic tags, performing inference through the industrial standardization document and semantic tags to obtain an optimized industrial standardization document, and obtaining an industrial rule set and a set of text parameters input by the user, deriving a first parameter set from the set of text parameters input by the user, and constructing a correlation graph based on the first parameter set and historical sessions. This invention achieves accurate semantic tag generation based on the WordNet thesaurus, resolving the ambiguity of polysemous words and avoiding misreading of equipment maintenance instructions or mismatching of process parameters caused by polysemous ambiguity. Simultaneously, by calculating the similarity between candidate parameters and user sessions through an improved Levenstein distance algorithm and compensation mechanism, it accurately captures the user's potential needs, improving consistency with the user's actual input intent. Combining forward rule matching and reverse constraint verification, a traceable derivation path is formed, meeting the stringent requirements for decision interpretability in industrial scenarios. The association graph uses the Kahn algorithm to construct parameter dependencies, identifies key parameters and detects cyclic dependencies through topological sorting, optimizing inference efficiency. Candidate parameters are sorted by accumulating initial weight values, click-through rates, and time decay factors, balancing immediate production needs and historical experience in industrial scenarios, achieving significant improvements in accuracy, adaptability, efficiency, and interpretability. Attached Figure Description
[0053] Figure 1 A flowchart illustrating a method for improving the understanding of user input completion parameters provided by the present invention;
[0054] Figure 2A schematic diagram of the module structure of a system for improving the understanding of user input completion parameters provided by the present invention;
[0055] Figure 3 A schematic diagram of the structure of an electronic device provided by the present invention;
[0056] Figure 4 The flowchart for generating semantic tags provided by this invention. Detailed Implementation
[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0058] It should be noted that the industrial rule set includes derivation rules, constraint rules, calculation rules, and priority configuration;
[0059] The derivation rule is defined as a logical chain that derives unknown parameters based on known parameters, and adopts an implied structure from premise to conclusion.
[0060] Constraint rules are formal constraints that verify the validity of parameters and ensure that the derivation results conform to business logic, such as time constraints and capacity constraints.
[0061] The calculation rules are functional rules that perform specific numerical or time calculations, and usually include mathematical operations or time operations;
[0062] Priority configuration sets the order in which rules are executed, resolving conflicts when multiple rules are matched. It includes explicit priority and implicit priority.
[0063] Explicit priorities directly indicate the rule priority, while implicit priorities are automatically adjusted based on user behavior feedback.
[0064] Example 1
[0065] Please see Figure 1 and Figure 4 As shown, this embodiment discloses a method for improving the understanding of user input completion parameters, the method comprising:
[0066] S101: Input an industrial standardization document, generate semantic tags, perform reasoning based on the industrial standardization document and semantic tags to obtain an optimized industrial standardization document, and obtain an industrial rule set and a set of text parameters input by the user. Derive the first parameter set from the set of text parameters input by the user, and construct an association graph based on the first parameter set and historical sessions.
[0067] The set of text parameters input by the user is the set of user instructions and industrial parameters received by the method;
[0068] Among them, the industrial standard documents define the basic class hierarchy and attribute range of electronic devices, which can support the derivation of basic relationships but do not cover new user requirements.
[0069] The optimized industry standardization documents expand upon the original documents by incorporating user feedback and rule optimization results. This is existing technology, and will not be elaborated upon further in this application.
[0070] The process of generating semantic tags includes: extracting the part-of-speech tags of user conversation words and tagging them to obtain conversation vectors with part-of-speech tags; querying a list of synonyms; and calculating the path similarity between the conversation vector and the set of synonyms in the list of synonyms.
[0071] It should be added that the synonym list in this embodiment uses the WordNet thesaurus database. Its thesaurus organizes the semantics of words as basic units—each thesaurus represents a specific concept (, polysemy is distinguished by part-of-speech tagging and unique numbering, and a three-dimensional knowledge network is constructed through semantic relationships such as hypernym / hypernym, whole / part, antonym, etc.).
[0072] The formula for calculating path similarity is as follows:
[0073]
[0074] In the formula: For the i-th thesaurus, Represents the set of context words, For a single word in the context set, express With vocabulary The shortest path length in the WordNet thesaurus database.
[0075] Take the maximum similarity of each word in the conversation vector, and then calculate the average to get the final similarity of the synonym set;
[0076] Compare the final similarity scores of all synonym sets and sort them from largest to smallest. Select the synonym set corresponding to the first final similarity score as the disambiguation result to obtain the semantic label.
[0077] For example, the derivation of the set of text parameters input by the user includes:
[0078] Scan the industrial rule set, bind the set of text parameters input by the user with the rules in the industrial rule set to obtain the first parameter corresponding to the rule, and continue to trigger subsequent rules based on the first parameter to obtain the derivation parameter set;
[0079] Then, based on the set of text parameters input by the user, reverse derivation is performed to obtain the preconditions and verify them, thus obtaining the derivation path. The derivation parameter set and the derivation path constitute the first parameter set.
[0080] Furthermore, the specific steps for constructing the association graph based on the first parameter set and historical sessions include:
[0081] The Kahn algorithm is used to perform topological sorting, identify the direct and indirect dependencies between parameters in the first parameter set, and then construct the association graph with parameters as nodes and dependencies as directional edges.
[0082] The specific steps for topological sorting using the Kahn algorithm are as follows:
[0083] Calculate the in-degree of each node, add all nodes with an in-degree of 0 to the zero-in-degree queue as the starting point for topological sorting, extract a node from the zero-in-degree queue, add it to the topological sequence, delete all directional edges of the node, and update the in-degree of the target node. If the in-degree of the target node becomes 0, add it to the zero-in-degree queue.
[0084] Repeat the above steps to process all nodes in the zero-in-degree queue to obtain a new topology sequence;
[0085] In the new topological sequence, the edges between adjacent nodes represent direct dependencies, and the dependencies between non-adjacent nodes formed through intermediate nodes are indirect dependencies, thus obtaining a set of dependency relationships until all nodes are added to the topological sequence.
[0086] S102: Extract multiple candidate parameters from the first parameter set, filter and weight them using historical click data to obtain the weight values of the candidate parameters, sort them based on the weight values of the candidate parameters, and obtain the sorted candidate parameter table.
[0087] It should be understood that the weight values of the candidate parameters include:
[0088] The similarity between candidate parameters and user sessions is calculated using an improved Levenstein distance algorithm. A similarity threshold of n is set, and candidate parameters with similarity ≥ n are selected. The parameters are then filtered according to rule priority, i.e., candidate parameters corresponding to high priority rules are retained, thus generating a candidate parameter set. In this embodiment, n is set to 0.8.
[0089] Perform local and global calculations separately to obtain local and global calculation results. Multiply the local and global calculation results to obtain the initial weight values of the candidate parameters.
[0090] The click-through rate is calculated based on historical click data, and the time decay factor is derived using an exponential decay function.
[0091] The initial weight value, click-through rate, and time decay factor of the candidate parameter are multiplied together to obtain the weight value of the candidate parameter.
[0092] The calculation of the similarity between candidate parameters and user sessions using the improved Levenstein distance algorithm includes:
[0093] Extract the user's editing operation sequence for candidate parameters from historical click data, and calculate the frequency of each editing operation at different character positions based on the editing operation sequence to obtain the operation frequency matrix and operation weight table;
[0094] The operation weight table can statistically analyze user operations, assigning low weight values to high-frequency operations and high weight values to low-frequency operations.
[0095] Each candidate parameter is split and represented as a two-dimensional vector to generate a parameter matrix. The field constraint rules are that the minute field only allows integers from 00 to 59, and the hour field only allows integers from 00 to 23.
[0096] Create a distance matrix between candidate parameters, and perform in-field editing calculations using the operation weight table to obtain the edit distance matrix. The specific calculation formula is as follows:
[0097]
[0098] In the formula: This represents the dynamic edit distance between the i-th field of parameter A and the j-th field of parameter B, where i and j are the field indices of the distance matrix between candidate parameters. , , These represent the weights for field deletion, field insertion, and field replacement operations, respectively. This indicates the specific character value of parameter A / B at field i / j;
[0099] It should be noted that when When =0, if any parameter is empty, all fields need to be inserted / deleted;
[0100] This formula integrates user behavior preferences (such as low weighting for high-frequency operations) through dynamic weighting and combines recursive calculation to quantify editing costs, thus preserving the rigor of character editing while incorporating semantic similarity. As the core mathematical engine, it supports subsequent semantic compensation and preference fusion, ultimately improving the accuracy of candidate parameter ranking.
[0101] The semantic features of the candidate parameters are extracted, the candidate parameters are converted into natural language descriptions, the semantic similarity between the candidate parameters is calculated using cosine similarity, and compensation is performed on the edit distance matrix based on the semantic similarity to obtain the compensation distance matrix.
[0102] Meanwhile, compensation is performed based on semantic similarity on the edit distance matrix, specifically as follows:
[0103] A compensation coefficient is generated, which is set to 0.3 in this embodiment. In other implementations, the value can be adjusted according to different situations; this application does not limit this. A semantic similarity threshold of 0.7 is set, and distance correction is applied to candidate parameters that exceed the semantic similarity threshold.
[0104]
[0105] Check whether the candidate parameters match historical high-frequency click data, and add preference scores to the candidate parameters that match;
[0106] The similarity is calculated by combining preference bonuses and compensation distance matrices, and the formula is as follows:
[0107]
[0108] The following are the separate local and global calculations:
[0109] The local calculation is: the number of times the candidate parameter appears in this session divided by the total number of parameters;
[0110] The overall calculation is as follows: the total number of sessions divided by the number of sessions including candidate parameters, and then the logarithm is taken.
[0111] As a concrete example, suppose that in the last 1000 sessions, x was clicked 800 times, y was clicked 150 times, and z was clicked 50 times, and the similarity of x, y, and z is 0.6, 0.2, and 0.2, respectively; in the last hour, x appeared 100 times and was clicked 80 times; y appeared 50 times and was clicked 10 times; z appeared 30 times and was clicked 3 times.
[0112] In this session, x appears twice, y appears once, z appears once, and the total number of parameters appears 5 times.
[0113] The local calculations are 0.4, 0.2, and 0.2 respectively;
[0114] The overall calculation is as follows:
[0115] X: ;
[0116] Y: ;
[0117] Z: ;
[0118] Click-through rates were 80%, 20%, and 10% respectively;
[0119] The time decay factor is (weight of the most recent hour = 1, weight of 2 hours ago = 0.8).
[0120] The weights of the candidate parameters are as follows:
[0121] X=0.0310, Y=0.0320, Z=0.0260;
[0122] S103: Receive and store the user's dialogue content, parse the user's input text parameter set to obtain the second parameter set, and generate the context parameter set by linking the second parameter set and the association graph.
[0123] It should be noted that the specific steps for receiving and storing user conversation content include:
[0124] Receive the user's session content, user session ID, and timestamp to create a session state stack, and use a wake-up buffer to store the number of the last 3 queries;
[0125] For example: Session ID=S202401061430, Timestamp=2024-01-06 14:30:00;
[0126] The specific steps for parsing the user-input text parameter set include:
[0127] Type validation is performed on each parameter in the text parameter set input by the user to obtain the validated parameter set. The validated parameter set is then combined with the rules in the industrial rule set to derive the derivation result. The derivation result is then added to the validated parameter set to obtain the variable parameter set.
[0128] It should be noted that this step enables precise parsing of user input and identifies implicit rule parameters through industry rule sets, thereby further expanding the user session.
[0129] By querying the association results of the variable parameter set through the association graph, the association results are added to the variable parameter set to obtain the second parameter set.
[0130] The linkage based on the second parameter set and the correlation graph includes:
[0131] The historical session content is compared with the industrial rule set to obtain extended rules. The extended parameters corresponding to the extended rules are added to the second parameter set to obtain the extended parameter set.
[0132] By querying the attributes corresponding to the extended parameter set through the association graph and binding them to the extended parameter set, the context parameter set can be obtained.
[0133] S104: Complete and provide feedback based on the sorted candidate parameter table and context parameter set;
[0134] As a specific implementation method, the process of completing and providing feedback based on the sorted candidate parameter table, context parameter set, and core parameters includes:
[0135] The context parameter set is treated as the initial completion parameter set, and the highest priority candidate parameter can be automatically selected for completion based on the sorted candidate parameter table.
[0136] The value to be completed is entered into the user interface, and the user's selection behavior and modification operation are recorded. This historical data is saved to optimize subsequent completion strategies.
[0137] Example 2
[0138] Please see Figure 2 As shown, based on the same inventive concept, this embodiment discloses a system for improving the understanding of user input completion parameters. For details not covered in this embodiment, please refer to the relevant sections of Embodiment 1. The system includes:
[0139] The basic framework module takes an industrial standardization document as input, generates semantic tags, and performs inference using the industrial standardization document and semantic tags to derive an optimized industrial standardization document. It also obtains the industrial rule set and the text parameter set input by the user, derives the first parameter set from the text parameter set input by the user, and constructs an association graph based on the first parameter set and historical sessions.
[0140] The initial parameter module extracts multiple candidate parameters from the first parameter set, combines them with historical click data for filtering and weighted calculation to obtain the weight values of the candidate parameters, and sorts them based on the weight values of the candidate parameters to obtain the sorted candidate parameter table;
[0141] The text storage module receives and stores the user's dialogue content, parses the user's input text parameter set to obtain a second parameter set, and generates a context parameter set based on the second parameter set and the association graph.
[0142] The parameter completion module completes and provides feedback based on the sorted candidate parameter table and context parameter set.
[0143] Example 3
[0144] Please see Figure 3As shown, this embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the methods provided by the above-described methods for improving the understanding of user input completion parameters.
[0145] Since the electronic device described in this embodiment is an electronic device used to implement the method for improving the understanding of user input completion parameters according to the embodiments of this application, those skilled in the art can understand the specific implementation and various variations of the electronic device in this embodiment based on the method for improving the understanding of user input completion parameters described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. Any electronic device used by those skilled in the art to implement the method for improving the understanding of user input completion parameters according to the embodiments of this application falls within the scope of protection of this application.
[0146] Example 4
[0147] This embodiment discloses a computer-readable storage medium, including a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the computer program, it implements the methods provided by the above-described methods for improving the understanding of user input completion parameters.
[0148] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters, weights, and thresholds in the formulas are set by those skilled in the art according to the actual situation.
[0149] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired or wireless network. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.
[0150] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0151] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0152] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only one method, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0153] The units described as separate components may or may not be physically separate. 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 the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0154] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0155] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
[0156] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for improving the understanding of user input completion parameters, the method being applied to intelligent dialogue scenarios in industrial manufacturing, characterized in that, The method includes: S101: Input an industrial standardization document, generate semantic tags, perform reasoning based on the industrial standardization document and semantic tags to obtain an optimized industrial standardization document, and obtain an industrial rule set and a set of text parameters input by the user. Derive the first parameter set from the set of text parameters input by the user, and construct an association graph based on the first parameter set and historical sessions. The set of text parameters input by the user is the set of user instructions and industrial parameters received by the method; S102: Extract multiple candidate parameters from the first parameter set, filter and weight them using historical click data to obtain the weight values of the candidate parameters, sort them based on the weight values of the candidate parameters, and obtain the sorted candidate parameter table. S103: Receive and store the user's dialogue content, parse the user's input text parameter set to obtain the second parameter set, and generate the context parameter set by linking the second parameter set and the association graph. S104: Complete and provide feedback based on the sorted candidate parameter table and context parameter set.
2. The method for improving the understanding of user input completion parameters according to claim 1, characterized in that, The generation of semantic tags includes: Extract the part-of-speech tags of user conversation words and label them to obtain conversation vectors with part-of-speech tags. Query the synonym list and calculate the path similarity between the conversation vector and the synonym set in the synonym list. Take the maximum similarity of each word in the conversation vector, and then calculate the average to get the final similarity of the synonym set; Compare the final similarity scores of all synonym sets and sort them from largest to smallest. Select the synonym set corresponding to the first final similarity score as the disambiguation result to obtain the semantic label.
3. The method for improving the understanding of user input completion parameters according to claim 1, characterized in that, The derivation of the user-input text parameter set includes: Scan the industrial rule set, bind the set of text parameters input by the user with the rules in the industrial rule set to obtain the first parameter corresponding to the rule, and continue to trigger subsequent rules based on the first parameter to obtain the derivation parameter set; Then, based on the set of text parameters input by the user, reverse derivation is performed to obtain the preconditions and verify them, thus obtaining the derivation path. The derivation parameter set and the derivation path constitute the first parameter set.
4. The method for improving the understanding of user input completion parameters according to claim 1, characterized in that, The weight values of the candidate parameters are obtained, including: The similarity between candidate parameters and user sessions is calculated using an improved Levenstein distance algorithm. A similarity threshold of n is set, and candidate parameters with similarity ≥ n are selected. The parameters are then filtered according to rule priority to generate a candidate parameter set. Perform local and global calculations separately to obtain local and global calculation results. Multiply the local and global calculation results to obtain the initial weight values of the candidate parameters. The click-through rate is calculated based on historical click data, and the time decay factor is derived using an exponential decay function. The initial weight value, click-through rate, and time decay factor of the candidate parameter are multiplied together to obtain the weight value of the candidate parameter.
5. The method for improving the understanding of user input completion parameters according to claim 4, characterized in that, The calculation of the similarity between candidate parameters and user sessions using the improved Levenstein distance algorithm includes: Extract the user's editing operation sequence for candidate parameters from historical click data, and calculate the frequency of each editing operation at different character positions based on the editing operation sequence to obtain the operation frequency matrix and operation weight table; Each candidate parameter is split and represented as a two-dimensional vector to generate a parameter matrix; Create a distance matrix between candidate parameters, and perform in-field editing calculations using the operation weight table to obtain the edit distance matrix. The specific calculation formula is as follows: In the formula: This represents the dynamic edit distance between the i-th field of parameter A and the j-th field of parameter B, where i and j are the field indices of the distance matrix between candidate parameters. , , These represent the weights for field deletion, field insertion, and field replacement operations, respectively. This indicates the specific character value of parameter A / B at field i / j; The semantic features of the candidate parameters are extracted, the candidate parameters are converted into natural language descriptions, the semantic similarity between the candidate parameters is calculated using cosine similarity, and compensation is performed on the edit distance matrix based on the semantic similarity to obtain the compensation distance matrix. Check whether the candidate parameters match historical high-frequency click data, and add preference scores to the candidate parameters that match; The similarity is calculated by combining preference bonuses and compensation distance matrices, and the formula is as follows: The following are the separate local and global calculations: The local calculation is: the number of times the candidate parameter appears in this session divided by the total number of parameters; The overall calculation is as follows: the total number of sessions divided by the number of sessions including candidate parameters, and then the logarithm is taken.
6. The method for improving the understanding of user input completion parameters according to claim 1, characterized in that, The specific steps for parsing the user-input text parameter set include: Type validation is performed on each parameter in the text parameter set input by the user to obtain the validated parameter set. The validated parameter set is then combined with the rules in the industrial rule set to derive the derivation result. The derivation result is then added to the validated parameter set to obtain the variable parameter set. By querying the association results of the variable parameter set through the association graph, the association results are added to the variable parameter set to obtain the second parameter set.
7. The method for improving the understanding of user input completion parameters according to claim 1, characterized in that, The linkage based on the second parameter set and the correlation graph includes: The historical session content is compared with the industrial rule set to obtain extended rules. The extended parameters corresponding to the extended rules are added to the second parameter set to obtain the extended parameter set. By querying the attributes corresponding to the extended parameter set through the association graph and binding them to the extended parameter set, the context parameter set can be obtained.
8. A system for improving the understanding of user input completion parameters, characterized in that, It is implemented based on the method for improving the understanding of user input completion parameters as described in any one of claims 1 to 7, the system comprising: The basic framework module takes an industrial standardization document as input, generates semantic tags, and performs inference using the industrial standardization document and semantic tags to derive an optimized industrial standardization document. It also obtains the industrial rule set and the text parameter set input by the user, derives the first parameter set from the text parameter set input by the user, and constructs an association graph based on the first parameter set and historical sessions. The initial parameter module extracts multiple candidate parameters from the first parameter set, combines them with historical click data for filtering and weighted calculation to obtain the weight values of the candidate parameters, and sorts them based on the weight values of the candidate parameters to obtain the sorted candidate parameter table; The text storage module receives and stores the user's dialogue content, parses the user's input text parameter set to obtain a second parameter set, and generates a context parameter set based on the second parameter set and the association graph. The parameter completion module matches the user-input text parameter set with the industry rule set to obtain the matching result. Based on the matching result and the historical session, it identifies the core parameters and completes and provides feedback based on the sorted candidate parameter table, context parameter set, and core parameters.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the method for improving the understanding of user input completion parameters as described in claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the method for improving the understanding of user input completion parameters as described in claims 1-7.