Large Model-Based Key Point Generation Method, Device, Apparatus and Medium for Railway Design

Abstract

This application relates to the field of computer technology, and particularly to a method, device, apparatus and medium for generating key points of railway design based on large models. The method of the present invention uses computer programming technology to systematically collect multi-source data sets, and based on this, trains a deep learning model, and then constructs a professional large model that can screen review opinions and preliminarily extract keywords from them, significantly simplifying the complexity of data governance compared with the prior art; calculates the weights of the preliminarily extracted keywords to obtain the design key points that need attention, so as to be able to intelligently extract valuable review data. The method of this application not only helps to deeply summarize the current railway engineering projects, but also provides accurate data support and experience reference for the smooth development and design of future projects.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular, to a method, device, apparatus and medium for generating key points of railway design based on a large model. Background Art

[0002] In the current collaborative design process of railway projects, design units use various advanced digital systems and platforms to conduct detailed review and verification of design outputs, and a large number of opinions will be generated in this process. The opinions generated in the full professional fields of a single railway project can reach several thousand at least and up to tens of thousands at most. These opinion data obtained after strict review and verification are valuable knowledge and experience accumulations for subsequent projects. They can effectively assist future review and verification work, and thus significantly improve the quality of design results. However, the governance of these review opinions is extremely challenging, especially the difficulty of accurately classifying and utilizing them. Therefore, how to efficiently utilize the experience of previous projects to provide guidance for subsequent projects has become an urgent problem to be solved. Summary of the Invention

[0003] The purpose of the present invention is to provide a method, device, apparatus and medium for generating key points of railway design based on a large model to improve the above problems. To achieve the above purpose, the technical solutions adopted by the present invention are as follows:

[0004] In a first aspect, the present application provides a method for generating key points of railway design based on a large model, including:

[0005] Obtain review opinions, and based on a large model, perform validity screening on the review opinions to obtain valid opinions;

[0006] Based on a large model, extract keywords from the valid opinions to obtain a keyword data set;

[0007] Count the number of times each keyword appears in the valid opinions, and calculate the weight of each keyword according to the data attributes of each valid opinion and the number of times the keyword appears;

[0008] Sort the keywords according to the weight of each keyword;

[0009] Obtain the target number of opinions, and according to the target number of opinions, extract keywords from the keyword data set in the sorted order to obtain target keywords;

[0010] Retrieve in the valid opinions according to the target keywords to obtain design key points.

[0011] In a second aspect, the present application further provides a device for generating key points of railway design based on a large model, which includes:

[0012] A screening module, configured to obtain review opinions and perform validity screening on the review opinions based on a large model to obtain valid opinions;

[0013] A keyword extraction module, configured to extract keywords from the valid opinions based on a large model to obtain a keyword data set;

[0014] A weight calculation module, configured to count the number of occurrences of each keyword in the valid opinions and calculate the weight of each keyword according to the data attributes of each valid opinion and the number of occurrences of the keyword;

[0015] A sorting module, configured to sort the keywords according to the weights of each keyword;

[0016] A target keyword determination module, configured to obtain the number of target opinions and extract keywords from the keyword data set in the sorting order according to the number of target opinions to obtain target keywords;

[0017] A design focus generation module, configured to retrieve in the valid opinions according to the target keywords to obtain design focuses.

[0018] In a third aspect, the present application further provides a railway design focus generation device based on a large model, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the railway design focus generation method based on a large model according to the present application are implemented.

[0019] In a fourth aspect, the present application further provides a readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the railway design focus generation method based on a large model according to the present application are implemented.

[0020] The beneficial effects of the present invention are as follows:

[0021] The present invention adopts computer programming technology to systematically collect multi-source data sets and, based on this, train a deep learning model, and then constructs a professional large model that can screen review opinions and initially extract keywords from them, significantly simplifying the complexity of data governance compared with the prior art; calculate the weights of the initially extracted keywords to obtain the design key points and difficulties that need attention, so as to be able to intelligently extract valuable review data. The method of the present application not only helps to deeply summarize the current railway engineering projects, but also provides accurate data support and experience reference for the smooth development and design of future projects.

[0022] Other features and advantages of the present invention will be described in the subsequent description, and some of them will become obvious from the description or be understood by implementing the embodiments of the present invention. Brief Description of the Drawings

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0024] Figure 1 It is a flowchart of a method for generating key points of railway design based on a large model in an embodiment of the present application;

[0025] Figure 2 It is a schematic structural diagram of a device for generating key points of railway design based on a large model in an embodiment of the present application;

[0026] Figure 3 It is a schematic structural diagram of a device for generating key points of railway design based on a large model in an embodiment of the present application.

[0027] Reference Numerals: 100, device for generating key points of railway design based on a large model; 200, screening module; 300, keyword extraction module; 400, weight calculation module; 410, first calculation unit; 420, second calculation unit; 430, third calculation unit; 440, fourth calculation unit; 450, fifth calculation unit; 460, sixth calculation unit; 431, H calculation subunit; 432, N1 calculation subunit; 433, N2 calculation subunit; 434, N3 calculation subunit; 435, weighted calculation subunit; 500, sorting module; 600, target keyword determination module; 610, target word quantity calculation module; 620, association table construction unit; 630, priority association table acquisition unit; 640, extraction unit; 700, design key point generation module; 800, device for generating key points of railway design based on a large model; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component. Detailed Embodiments

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. The components of the embodiments of the present invention usually described and illustrated in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0029] It should be noted that similar reference numerals and letters indicate similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present invention, terms such as "first" and "second" are only used for descriptive distinction and cannot be understood as indicating or implying relative importance.

[0030] Embodiment 1

[0031] As Figure 1 shown, the present application provides a method for generating key points of railway design based on a large model, including steps S100, S200, S300, S400, S500, and S600.

[0032] Step S100: Obtain review opinions, and based on the large model, perform validity screening on the review opinions to obtain valid opinions;

[0033] The review opinions are a large number of modification suggestions generated during the collaborative design process of railway engineering and have been reviewed by experts. Each review opinion corresponds to its data attributes. When obtaining the review opinions from the railway engineering collaborative design platform, their data attributes are also obtained. In this embodiment, the data attributes include project type (ProjectType) attribute, area (Area) attribute, railway level (Standard) attribute, length (Length) attribute, stage level (Stage) attribute, major (Majors) attribute, and importance level (Rank) attribute.

[0034] Among these review opinion data, some data are invalid data. For example, "See the review form for review opinions", "Agree", "No review opinions", etc. These are meaningless opinion data that need to be eliminated. Since the number of review opinions is huge and it is impossible to screen them one by one manually, the present application uses a large model for screening.

[0035] Specifically, based on the existing open source large models, such as Qwen2-72B, large model training data is written to conduct validity screening training. The data writing content adopts the form of Prompt+Completion, and the data format uses JSON format.

[0036] Here is an example:

[0037] ①Prompt: Please judge whether the following opinions are valid opinions, answer yes or no: In the case of current uncertain geological conditions, combined with the review experience and internal control principles of the appraisal center, the control values of the clear distances of various types of adjacent, parallel, underpass and overpass risk source structures are temporarily determined by the corresponding multiples of the maximum span D of tunnel excavation.

[0038] Completion: Yes.

[0039] ②Prompt: Please judge whether the following opinions are valid opinions. Answer yes or no: Agree to send it to the hospital for review.

[0040] Completion: No.

[0041] ③Prompt: Please judge whether the following opinions are valid opinions, and answer yes or no: "Control elevation" should be clearly defined as "Track surface control elevation".

[0042] Completion: Yes.

[0043] ④Prompt: Please judge whether the following opinions are valid opinions, answer yes or no: Please refer to the audit form for details of the audit opinions.

[0044] Completion: No.

[0045] After effectiveness screening training, the resulting large screening model can be used to judge and screen the effectiveness of all review opinions to obtain effective opinions.

[0046] Step S200: extract keywords from the valid opinions based on the large model to obtain a keyword data set;

[0047] In order to make a refined statistics of the effective opinions and further refine the high-frequency design difficulties that need to be paid attention to in the effective opinions to assist in design review and output results report, it is first necessary to extract keywords for each effective opinion.

[0048] This application is based on the existing open source big model to perform keyword extraction training, so that the trained big model can accurately give the key contents of railway engineering collaborative design. The training data writing type adopts the form of Prompt+Completion, and the data format uses JSON format. The example is as follows:

[0049] ①Prompt: Please list the keywords in the following opinions: Supplementary water level, overpass, etc. for controlling elevation.

[0050] Completion: Water level, overpass, controlling elevation.

[0051] ②Prompt: Please list the keywords in the following opinions: The span of the catenary in the subgrade section is generally 55m, and the maximum exceeds 60m, which is incorrect. It is recommended to change it to that the span of the catenary in the subgrade section is generally 55m, and the maximum does not exceed 60m.

[0052] Completion: Subgrade section, catenary span.

[0053] ③Prompt: Please list the keywords in the following opinions: Verify whether the mileage position of the culvert at C1K97 + 629 conflicts with the bridge table

[0054] Completion: Culvert, mileage, bridge table.

[0055] After training, the obtained analysis large model can extract keywords from all valid opinions to obtain a keyword dataset.

[0056] The training process of the large model in step S200 further includes steps S210, S220, and S230.

[0057] Step S210, eliminate Tokenization misunderstandings;

[0058] In railway engineering design review data, there will be situations where chapter serial numbers are like 1.11 and design values are like 1.11. After the large model tokenizes, it will form a result of 1 / . / 11, and it cannot correctly understand the actual magnitude of the design value. Through the optimization module of the large model, the positions of different types of numbers in the opinions and their collocations with different words are provided, enabling the large model to determine in what context the value is a floating-point number and eliminating misunderstandings.

[0059] Step S220, eliminate errors in confusing professional term concepts;

[0060] In the process of railway engineering design review, since each review expert's expression of the same thing cannot be unified, the large model cannot make a judgment. The following is an example: "Gradient difference" and "algebraic difference of gradients" are two ways of saying the same word. Without training, the large model will extract "algebraic difference" in the latter as the design focus, and at this time, the output result will analyze "gradient difference" and "algebraic difference" as two results. Therefore, it is necessary to optimize by providing more training data containing these two terms and their contexts, enabling the dedicated large model to fully learn the relationship between them and perform combined statistics.

[0061] Step S230, eliminate errors in the ambiguity of railway project specific nouns

[0062] In the railway engineering design review data, there are a large number of cases where abbreviations are used instead of project names. For example, "Wenfu" represents Wenzhou to Fuzhou in railway engineering. The large model cannot determine the meaning of words, and combining with other words will cause ambiguity. Therefore, it is necessary to optimize and correct the errors of such words in the model, and through word combination, make the large model understand the meaning of proper nouns in sentences.

[0063] Step S300: Count the number of occurrences of each keyword in the valid opinions, and calculate the weight of each keyword according to the data attributes of each valid opinion and the number of occurrences of the keyword.

[0064] After analyzing the keywords extracted by the large model for each valid opinion, it is necessary to analyze the value (importance level) of each keyword. The specific methods include:

[0065] Step S310: Calculate the importance weight coefficient W according to the importance level (Rank) attribute of the valid opinion and the number of occurrences of the keyword. i ;

[0066] The Rank attribute is divided into three categories: MI (approved level by the general engineer of the group), I (approved level by the professional general engineer), N (approved level by the special register and compliance).

[0067] Count the number of occurrences of the keyword in the valid opinions of these three Rank attributes. For example, the total number of occurrences of a certain keyword in all valid opinions is x times. Among them, it appears x MI times in the valid opinions with the Rank attribute of MI, x I times in the valid opinions with the Rank attribute of I, and x N times in the valid opinions with the Rank attribute of N, and x = x MI + x I + x N .

[0068] Under the MI attribute, W i (MI) = 3x MI 2 ;

[0069] Under the I attribute, W i (I) = 2x I + 1;

[0070] Under the N attribute, W i (N) = x N + 0.5;

[0071] Then

[0072] Step S320: Obtain the occurrence frequency weight coefficient W based on the occurrence times of the said keyword t ;

[0073] Occurrence frequency weight coefficient W t =α×log2(1 + x), where x is the total occurrence times of the keyword; α is an adjustment parameter. The value of the adjustment parameter α can be set according to the total number of valid opinions to ensure that W t will not be too large or too small.

[0074] Step S330: Obtain the railway level weight coefficient W according to the railway level (Standard) attribute of the valid opinion and the occurrence times of the said keyword sd ;

[0075] The Standard attribute is divided into H, N1, N2, N3 (representing the railway level). Count the occurrence times of the keyword in the valid opinions of these four types of Standard attributes, which are x H , x N1 , x N2 , x N3 ; x is the total occurrence times of the keyword, x = x H +x N1 +x N2 +x N3 ;

[0076] According to the occurrence times x H of the keyword in the valid opinions with the railway level attribute type of H, calculate the first weight coefficient W sd (H):

[0077]

[0078] According to the occurrence times x N1 of the keyword in the valid opinions with the railway level attribute type of N1, calculate the second weight coefficient W sd (N1):

[0079] W sd (N1)=1.3x N1 ·(1 + 0.08ln(1 + 0.5x N1 ));

[0080] According to the occurrence times x N2 of the keyword in the valid opinions with the railway level attribute type of N2, calculate the third weight coefficient W sd (N2):

[0081] W sd (N2)=1.1x N2 ·(1 + 0.06x N2+0.003x N2 2 );

[0082] According to the number of occurrences x of the keyword in the valid opinions with the railway level attribute type of N3 N3 , calculate the fourth weight coefficient W sd (N3):

[0083]

[0084] According to the first weight coefficient, the second weight coefficient, the third weight coefficient and the fourth weight coefficient, calculate the railway level weight coefficient W by weighted calculation sd :

[0085]

[0086] Step S340. According to the number of occurrences of the keyword and the total number of valid opinions, obtain the data density weight coefficient W ρ ;

[0087] Data density weight coefficient where x is the total number of occurrences of the keyword, n is the total number of valid opinions, and β is a regulation parameter. The value of the regulation parameter β can be set according to the total number of valid opinions to ensure that W ρ will not be too large or too small.

[0088] Step S350. According to the stage level (Stage) attribute of the valid opinion and the number of occurrences of the keyword, obtain the stage level weight coefficient W sg ;

[0089] The Stage attribute is divided into Pre (pre-feasibility study), FS (feasibility study), PD (preliminary design), CD (construction drawing design); count the number of occurrences of the keyword in the valid opinions of these four types of Stage attributes, which are x Pre , x FS , x PD , x CD , x is the total number of occurrences of the keyword, x = x Pre +x FS +x PD +x CD ;

[0090] The formula for calculating W sg under each attribute:

[0091] Under the Pre attribute, W sg (Pre) = 1.1x Pre ×e 0.06xPre ;

[0092] Under the FS attribute, Wsg (FS) = 1.4x FS × e 0.09xFS ;

[0093] Under the PD attribute, W sg (PD) = 1.7x PD × e 0.12xPD ;

[0094] Under the CD attribute, W sg (CD) = 2x CD × e 0.15xCD ;

[0095] Then

[0096] Step S360, according to W i 、W t 、W sd 、W ρ and W sg Get the weight W of the keyword; The calculation formula of the weight W is:

[0097] W = W i × W t × W sd × W ρ × (1 + W sg ).

[0098] Step S400, sort the keywords according to the weight of each keyword;

[0099] After the calculation in step S300, each keyword can get a weight W, and all keywords are sorted in descending order of the weight W.

[0100] Step S500, obtain the number of target opinions, and extract keywords from the keyword dataset in the sorted order according to the number of target opinions to obtain the target keywords;

[0101] The number of target opinions refers to the number of opinions that need to be output for display. In this embodiment, the number of target opinions is determined by the number y of engineering projects, and the number of target opinions R = 10 + 40 × {1 - math.exp(-0.1y)} × {math.log(1 + 0.1y) / math.log(2)}.

[0102] The number y of engineering projects can be obtained through the engineering project (ProjectType) attribute of the valid opinions. By counting how many ProjectType attributes the valid opinions correspond to, the value of y can be obtained.

[0103] According to the calculated number of target opinions R, calculate the number of target keywords K, where K = R / J, and J is the preset number of opinions that each target keyword needs to display;

[0104] After obtaining K, the first K keywords can be directly extracted from the keyword dataset in the sorted order, and used as the target keywords.

[0105] In the field of railway engineering design, the key points concerned by different specialties are likely to be different, that is, for different specialties, the required keywords will be different. In order to improve the relevance between the target keywords and the specialties in this embodiment, the following method is preferably used to extract the target keywords:

[0106] The data attributes of the valid opinions include the specialty (Majors) attribute. Classify the valid opinions according to the specialty attribute; for the valid opinions of each specialty category, judge whether the number of times any two keywords appear in the same valid opinion exceeds the preset threshold. If so, these two keywords are related to each other; a keyword may have one or more related keywords, or may not have any; construct an association table according to the related keywords under each specialty attribute;

[0107] For example, according to the valid opinions of specialty M1, an association table 1 is constructed, where keyword A and keyword B are related keywords;

[0108] According to the valid opinions of specialty M2, an association table 2 is constructed, where keyword A and keyword C are related keywords;

[0109] Before extracting the keywords, obtain the preferred specialty type, and obtain the corresponding preferred association table according to the preferred specialty type;

[0110] The preferred specialty type can be the query condition input by the user. For example, the preferred specialty type is specialty M2, that is, obtain association table 2 as the preferred association table;

[0111] Extract keywords from the keyword dataset in the sorted order, and extract their related keywords in the preferred association table one by one according to the extracted keywords. When the sum of the extracted keywords and related keywords reaches the number of target keywords, stop extraction to obtain the target keywords.

[0112] For example, in the keyword dataset, the first keyword is keyword A. Extract keyword A, and at the same time retrieve the related keyword of keyword A according to association table 2, which is keyword C, and extract keyword C; then perform the same operation on the subsequent keywords in the keyword dataset until the sum of the extracted keywords and related keywords reaches the number of target keywords K, and use these extracted keywords as the target keywords.

[0113] It should be noted that the target keywords are not extracted repeatedly. For example, after extracting keyword A and keyword C according to the above example, if it is found that the tenth keyword in the keyword dataset is keyword C, then the tenth keyword is skipped and the next one is extracted.

[0114] S600. Retrieve in the valid opinions according to the target keywords to obtain the design focus.

[0115] Perform a reverse search in the analysis large model according to K target keywords, retrieve the valid opinions containing the target keywords. The same valid opinion may appear multiple times. Arrange the valid opinions in descending order according to the number of occurrences, and select the top J valid opinions as the design focus for display.

[0116] For example, it is preset that the number of opinions J to be displayed for each target keyword is 3, and the target keyword "safety line" is obtained. Among the valid opinions containing "safety line", the three opinions "Verify whether the safety line adopts the form of setting curves away from the main line and arrival / departure lines", "Whether to add a safety line when connecting the maintenance work area", and "The safety line should not enter the bridge range" have the most occurrences. These three valid opinions are used as the design focus for display.

[0117] As an optional implementation manner of the present application, the railway engineering design focus generation method further includes step S700:

[0118] Classify the design focus according to the professional attributes corresponding to the obtained design focus;

[0119] Fill the design focus into a preset template according to the classification to obtain an analysis report.

[0120] Embodiment 2

[0121] As Figure 2 shown, this embodiment provides a railway design focus generation device based on a large model, including:

[0122] A screening module for obtaining review opinions and performing validity screening on the review opinions based on a large model to obtain valid opinions;

[0123] A keyword extraction module for extracting keywords from the valid opinions based on a large model to obtain a keyword dataset;

[0124] A weight calculation module for counting the number of occurrences of each keyword in the valid opinions and calculating the weight of each keyword according to the data attributes of each valid opinion and the number of occurrences of the keyword;

[0125] A sorting module for sorting the keywords according to the weight of each keyword;

[0126] A target keyword determination module, configured to obtain the number of target opinions, and extract keywords from a keyword dataset in a sorted order according to the number of target opinions to obtain target keywords;

[0127] A design focus generation module, configured to retrieve in the valid opinions according to the target keywords to obtain design focuses.

[0128] In a specific implementation manner disclosed in the present application, the data attributes of the valid opinions include an importance level attribute, a railway level attribute, and a stage level attribute; the weight calculation module includes:

[0129] A first calculation unit, configured to obtain an importance weight coefficient according to the importance level attribute of the valid opinion and the occurrence times of the keyword;

[0130] A second calculation unit, configured to obtain an occurrence times weight coefficient according to the occurrence times of the keyword;

[0131] A third calculation unit, configured to obtain a railway level weight coefficient according to the railway level attribute of the valid opinion and the occurrence times of the keyword;

[0132] A fourth calculation unit, configured to obtain a data density weight coefficient according to the occurrence times of the keyword and the total number of valid opinions;

[0133] A fifth calculation unit, configured to obtain a stage level weight coefficient according to the stage level attribute of the valid opinion and the occurrence times of the keyword;

[0134] A sixth calculation unit, configured to obtain the weight of the keyword according to the importance weight coefficient, the occurrence times weight coefficient, the railway level weight coefficient, the data density weight coefficient, and the stage level weight coefficient.

[0135] In a specific implementation manner disclosed in the present application, the types of the railway level attribute include H, N1, N2, and N3; the third calculation unit includes:

[0136] An H calculation subunit, configured to calculate a first weight coefficient according to the occurrence times of the keyword in the valid opinions with the railway level attribute type of H;

[0137] An N1 calculation subunit, configured to calculate a second weight coefficient according to the occurrence times of the keyword in the valid opinions with the railway level attribute type of N1;

[0138] An N2 calculation subunit, configured to calculate a third weight coefficient according to the occurrence times of the keyword in the valid opinions with the railway level attribute type of N2;

[0139] An N3 calculation subunit, configured to calculate a fourth weight coefficient according to the number of occurrences of the keyword in the valid opinions with the railway level attribute type of N3;

[0140] A weighted calculation subunit, configured to perform weighted calculation to obtain a railway level weight coefficient according to the first weight coefficient, the second weight coefficient, the third weight coefficient, and the fourth weight coefficient.

[0141] In a specific embodiment disclosed in the present application, the data attribute of the valid opinion includes a professional attribute, and the target keyword determination module includes:

[0142] A target word quantity calculation unit, configured to calculate the number of target keywords according to the number of target opinions;

[0143] An association table construction unit, configured to classify the valid opinions according to the professional attribute; for each category of the valid opinions, determine whether the number of times any two of the keywords appear in the same valid opinion exceeds a preset threshold, and if so, the two keywords are associated with each other; construct an association table according to the associated keywords under each category of professional attributes;

[0144] A priority association table acquisition unit, configured to acquire a priority professional type, and acquire a corresponding priority association table according to the priority professional type;

[0145] An extraction unit, configured to extract keywords in the keyword dataset in the sorting order, and extract their associated keywords of the extracted keywords in the priority association table one by one. When the sum of the extracted keywords and the associated keywords reaches the number of target keywords, stop extraction to obtain the target keywords.

[0146] Embodiment 3

[0147] Corresponding to the above method embodiment, in this embodiment, a railway design key point generation device based on a large model is further provided. The railway design key point generation device based on a large model described below can be correspondingly referred to the railway design key point generation method based on a large model described above.

[0148] Figure 3 It is a block diagram of a railway design key point generation device 800 based on a large model shown according to an exemplary embodiment. As Figure 3As shown, the large model-based key point generation device 800 for railway design includes a processor 801 and a memory 802. The large model-based key point generation device 800 for railway design may also include one or more of a multimedia component 803, an input / output (I / O) interface 804, and a communication component 805. Among them, the processor 801 is used to control the overall operation of the large model-based key point generation device 800 to complete all or part of the steps in the above-mentioned large model-based key point generation method for railway design. The memory 802 is used to store various types of data to support the operation of the large model-based key point generation device 800. These data may include, for example, commands for any application or method operating on the large model-based key point generation device 800, as well as application-related data, such as contact data, sent and received messages, pictures, audio, video, and so on. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0149] The multimedia component 803 may include a screen and an audio component. Among them, the screen may be a touch screen, for example, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone, and the microphone is used to receive external audio signals.

[0150] The received audio signal can be further stored in the memory 802 or sent through the communication component 805. The audio component also includes at least one speaker for outputting the audio signal. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the other interface modules can be a keyboard, a mouse, buttons, etc. These buttons can be virtual buttons or physical buttons. The communication component 805 is used for the large model-based railway design key point generation device 800 to communicate with other devices in a wired or wireless manner. The wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them. Accordingly, the communication component 805 can include: a Wi-Fi module, a Bluetooth module, an NFC module.

[0151] In an exemplary embodiment, the device for mutual signature and verification of digital files can be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, and is used to execute the above-mentioned large model-based railway design key point generation method.

[0152] In another exemplary embodiment, a computer-readable storage medium including program commands is further provided. When the program commands are executed by a processor, the steps of the above-mentioned large model-based railway design key point generation method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program commands, and the program commands can be executed by the processor 801 of the large model-based railway design key point generation device 800 to complete the above-mentioned large model-based railway design key point generation method.

[0153] Embodiment 4

[0154] Corresponding to the above-mentioned embodiment of the large model-based railway design key point generation method, in this embodiment, a readable storage medium is further provided, and a readable storage medium described below can be correspondingly referred to with the above-mentioned large model-based railway design key point generation method.

[0155] A readable storage medium stores a computer program thereon. When the computer program is executed by a processor, the steps of the above-mentioned embodiment of the method for generating key points of railway design based on a large model are implemented.

[0156] Specifically, the readable storage medium can be various readable storage media 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 disc that can store program codes.

[0157] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

[0158] As mentioned above, this is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or replacements, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for generating key points of railway design based on large models, characterized in that, Including: Obtain review opinions, and based on a large model, perform validity screening on the review opinions to obtain valid opinions; Based on a large model, extract keywords from the valid opinions to obtain a keyword data set; Count the number of occurrences of each keyword in the valid opinions, and calculate the weight of each keyword according to the data attributes of each valid opinion and the number of occurrences of the keyword; The data attributes of the valid opinions include importance level attributes, railway level attributes, and stage level attributes; The counting of the number of occurrences of each keyword and the calculation of the keyword weight according to the data attributes of each valid opinion and the number of occurrences of the keyword include: Obtain an importance weight coefficient according to the importance level attribute of the valid opinion and the number of occurrences of the keyword; Obtain an occurrence frequency weight coefficient according to the number of occurrences of the keyword; Obtain a railway level weight coefficient according to the railway level attribute of the valid opinion and the number of occurrences of the keyword, including: The railway level attributes are divided into H, N1, N2, and N3, and the number of occurrences of the keyword in these four types of railway level attributes are x H , x N1 , x N2 , x N3 ; the total number of occurrences of the keyword in all valid opinions is x times; Calculate the first weight coefficient W sd (H): Calculate the second weight coefficient W sd (N1): W sd (N1) = 1.3x N1 ·(1 + 0.08ln(1 + 0.5x N1 )) Calculate the third weight coefficient W sd (N2): W sd (N2) = 1.1x N2 ·(1 + 0.06x N2 + 0.003x N2 2 ); Calculate the fourth weight coefficient W sd (N3): The weighted calculation yields the railway level weight coefficient W sd : Obtain a data density weight coefficient according to the number of occurrences of the keyword and the total number of valid opinions; Obtain a stage level weight coefficient according to the stage level attribute of the valid opinion and the number of occurrences of the keyword; Obtain the weight of the keyword according to the importance weight coefficient, occurrence frequency weight coefficient, railway level weight coefficient, data density weight coefficient, and stage level weight coefficient; Sort the keywords according to the weight of each keyword; Obtain the number of target opinions, and according to the number of target opinions, extract keywords from the keyword data set in the sorted order to obtain target keywords; Retrieve in the valid opinions according to the target keywords to obtain the design focus.

2. The method for generating key points of railway design based on large models according to claim 1, wherein The extracting of keywords from the keyword data set in the sorted order according to the number of target opinions to obtain target keywords includes: Calculate the number of target keywords according to the number of target opinions; The data attributes of the valid opinions include professional attributes. Classify the valid opinions according to the professional attributes; for each category of valid opinions, determine whether the number of times any two keywords appear in the same valid opinion exceeds a preset threshold. If so, the two keywords are associated with each other; construct an association table according to the associated keywords under each professional attribute; Obtain the preferred professional type, and obtain the corresponding preferred association table according to the preferred professional type; Extract keywords from the keyword data set in the sorted order, and extract their associated keywords in the preferred association table one by one according to the extracted keywords. When the sum of the extracted keywords and the associated keywords reaches the number of target keywords, stop extraction to obtain the target keywords.

3. A device for generating key points of railway design based on large models, characterized in that, Including: A screening module for obtaining review opinions and performing validity screening on the review opinions based on a large model to obtain valid opinions; A keyword extraction module for extracting keywords from the valid opinions based on a large model to obtain a keyword data set; A weight calculation module for counting the number of occurrences of each keyword in the valid opinions and calculating the weight of each keyword according to the data attributes of each valid opinion and the number of occurrences of the keyword; The weight calculation module includes: A first calculation unit, configured to obtain an importance weight coefficient according to the importance level attribute of the valid opinions and the occurrence times of the keywords; A second calculation unit, configured to obtain an occurrence times weight coefficient according to the occurrence times of the keywords; A third computing unit, configured to obtain a railway level weight coefficient according to the railway level attribute of the valid opinion and the occurrence times of the keyword; the railway level attribute is divided into H, N1, N2, and N3, and the occurrence times of the keyword in these four types of railway level attributes are x H , x N1 , x N2 , x N3 respectively; the total occurrence times of the keyword in all valid opinions is x times; The third calculation unit includes: An H calculation subunit, configured to calculate a first weight coefficient W sd (H): N1 calculation subunit, configured to calculate the second weight coefficient W sd (N1): W sd (N1) = 1.3x N1 ·(1 + 0.08ln(1 + 0.5x N1 )) N2 calculation subunit, used to calculate the third weight coefficient W sd (N2): W sd (N2) = 1.1x N2 ·(1 + 0.06x N2 + 0.003x N2 2 ); N3 calculation subunit, configured to calculate a fourth weight coefficient W sd (N3): A weighted calculation subunit, configured to calculate a railway level weight coefficient W through weighted calculation sd : A fourth calculation unit, configured to obtain a data density weight coefficient according to the occurrence times of the keywords and the total number of valid opinions; A fifth calculation unit, configured to obtain a stage level weight coefficient according to the stage level attribute of the valid opinions and the occurrence times of the keywords; A sixth calculation unit, configured to obtain the weight of the keywords according to the importance weight coefficient, the occurrence times weight coefficient, the railway level weight coefficient, the data density weight coefficient and the stage level weight coefficient; A sorting module, configured to sort the keywords according to the weight of each keyword; A target keyword determination module, configured to obtain the number of target opinions, and extract keywords from the keyword dataset in the sorted order according to the number of target opinions to obtain target keywords; A design focus generation module, configured to retrieve in the valid opinions according to the target keywords to obtain design focuses; 4. The device for generating key points of railway design based on a large model according to claim 3, wherein, The data attribute of the valid opinions includes a professional attribute, and the target keyword determination module includes: A target word quantity calculation unit, configured to calculate the number of target keywords according to the number of target opinions; An association table construction unit, configured to classify the valid opinions according to the professional attribute; for each category of valid opinions, determine whether the number of times that any two keywords appear in the same valid opinion exceeds a preset threshold, and if so, the two keywords are mutually associated keywords; construct an association table according to the associated keywords under each category of professional attributes; A priority association table acquisition unit, configured to obtain a priority professional type, and obtain the corresponding priority association table according to the priority professional type; An extraction unit, configured to extract keywords from the keyword dataset in the sorted order, and extract their associated keywords in the priority association table one by one according to the extracted keywords. When the sum of the extracted keywords and the associated keywords reaches the number of target keywords, stop extraction to obtain target keywords.

5. A key point generation device for railway design based on a large model, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, the steps of the method for generating railway design focuses based on a large model according to any one of claims 1 to 2 are implemented.

6. A readable storage medium, characterized in that, A computer program is stored on the readable storage medium. When the computer program is executed by the processor, the steps of the method for generating railway design focuses based on a large model according to any one of claims 1 to 2 are implemented.

Images