System and method for planning standard system based on business scene model

Abstract

The invention relates to the technical field of computers, and provides a system and a method for planning a standard system based on a business scene model, and the system comprises a database, a standard demand document creation module, a standard and demand association matching module, a matching result confirmation module, a standard analysis module and a standard system planning module. Wherein the database is used for storing various documents and tables; the standard demand document creating module is used for processing a standard demand document; a standard and demand association matching module queries a matched standard; the matching result confirmation module confirms the matching result; the standard analysis module performs standard planning according to the confirmed matching result; and the standard system planning module is used for displaying the existing standard system. The invention further provides a using method of the system. According to the invention, the process of a business scene model planning standard system based on artificial experience is realized by using a standardized system, and the accuracy of matching with the existing standard and subsequent standard planning is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a system and method for planning a standard system based on a business scenario model. Background Art

[0002] The standard system is a scientific and organic whole formed by the internal connections of standards within a certain range. Planning and building a standard system is a method of using systems theory to guide standardization work. It is the foundation and prerequisite for developing a standard system, and it is also the basis for formulating standards and revising plans and programs.

[0003] The traditional planning and construction of the standard system is based on the analysis of the current situation and the decision-making of expert collective wisdom. The relevant standards for achieving specific standardization goals are summarized and classified in a certain logical sequence, which has certain limitations. First, in the planning process of the existing standard system construction, the description of the standardization goals is relatively general, resulting in unclear standard requirements. Second, the correspondence between the standard and the structure of the standard system is determined by empirical judgment, which is easily affected by personal subjectivity and limitations and may lead to deviations. Third, the standard data that standard system planning experts have mastered is limited, and it is difficult to fully identify the existing relevant standards.

[0004] Therefore, a system is needed to plan the standard system based on the business scenario model, unify the standard system planning process, achieve matching between the itemized business scenario requirements and the clause and indicator levels of the standards, and improve the support efficiency of the standards for business scenarios. Summary of the invention

[0005] In view of the limitations of traditional standard system planning and construction, the present invention proposes a system and method for planning a standard system based on a business scenario model. The system uses matching technology to achieve clause-level and indicator-level matching of business scenario requirements with standards based on standard requirement documents of business scenarios and itemized descriptions of requirements, thereby scientifically and efficiently planning and constructing a standard system.

[0006] The present invention discloses a system for planning a standard system based on a business scenario model, which comprises the following modules: a database, a standard requirement document creation module, a standard and requirement association matching module, a matching result confirmation module, a standard analysis module and a standard system planning module; wherein:

[0007] The database is used to store various documents and tables, including at least a business scenario model sub-database, a standard requirement document sub-database, a structured standard sub-database, and a standard system sub-database;

[0008] The standard requirement document creation module is used to create, modify and save standard requirement documents. The standard requirement document creation module enters a new standard requirement document in a table form or directly imports an existing standard requirement document. The standard requirement document includes at least three fields: number, requirement item, and requirement content. Each requirement is a requirement item and has a unique number. Only the bottom-level requirement has corresponding requirement content. If the requirement item can be decomposed into multiple sub-requirement items, this requirement item has no corresponding requirement content. When the standard requirement document is completed, the standard requirement document creation module saves the completed standard requirement document into the standard requirement document sub-database in the database;

[0009] The standard and requirement association matching module uses a multi-dimensional similarity matching algorithm to query the standards that match the requirement content in the structured standard sub-database one by one according to the standard requirement document, and returns the standard information. The standard information at least includes the standard number, standard name, matching content and applicability. The standard and requirement association matching module includes the following sub-modules: word segmentation sub-module, keyword extraction sub-module, entity recognition sub-module, semantic vector generation sub-module, semantic vector similarity sub-module, keyword matching sub-module, comprehensive matching merging module and relevance sorting sub-module; the word segmentation sub-module is used to obtain a word set after word segmentation of the requirement items and requirement content, the keyword extraction sub-module is used to select keywords from the word set output by the word segmentation sub-module, the entity recognition sub-module is used to identify entity words with specific meanings from the requirement character string, and the semantic vector generation sub-module is used to A vector is generated according to the requirement string. The semantic vector similarity submodule is used to retrieve the generated semantic vector of the requirement string through the vector index of the structured standard sub-database to obtain standard clauses that match the requirement string. The keyword matching submodule is used to compare the keywords obtained by the keyword extraction submodule and the entity words obtained by the entity recognition submodule with the keyword index in the structured standard sub-database to obtain standard clauses that match the requirement string. The comprehensive matching merging module will merge the standard clauses obtained by the keyword matching submodule and the standard clauses obtained by the semantic vector similarity submodule for the same requirement string to obtain a set of standard clauses that match the requirement string. The relevance sorting submodule sorts the standard clauses that match the requirement string by similarity score and returns the standard clauses that meet the threshold requirements.

[0010] The matching result confirmation module displays the matching results of the association between the standard and the requirement, and manually modifies, confirms and supplements the matching results of the association between the standard and the requirement, and saves the matching results after confirmation;

[0011] The standard analysis module plans the standards according to the confirmed matching results and saves the planning results as a new standard system;

[0012] The standard system planning module is used to display the existing standard system. After the standard system name is entered in the query interface and query is clicked, four page tabs will appear: business scenario modeling, standard system structure diagram generation, standard detailed table generation, and standard system statistical table generation. By clicking on the four tabs respectively, the business scenario model, standard system structure diagram, standard detailed table, and standard statistical results will be displayed respectively. The four tabs can be displayed and closed in combination at will.

[0013] Preferably, the word segmentation submodule in the standard and requirement association matching module is implemented by a dual matching algorithm based on a dictionary and probability, specifically: a requirement item and the requirement content in the standard requirement document are combined into a requirement string as the input of the word segmentation submodule. When there is a matching term in the dictionary, a forward maximum matching algorithm (MM, Maximum Matching) is used to scan the text from left to right to find the longest matching word:

[0014] Word length = max(length of words matched in dictionary)

[0015] When there is no matching term in the dictionary, the statistical word segmentation algorithm Hidden Markov Model (HMM) is used to calculate the transition probability and emission probability between words through training and annotating the corpus, and split the required string into possible word sequences;

[0016]

[0017] Among them, P(ω i |ω i-1 ) represents the word ω i In the word ω i-1 The probability of appearing later.

[0018] Preferably, the keyword extraction submodule in the standard and requirement association matching module uses the TF-IDF algorithm to automatically extract the words or phrases that best reflect the theme or core content of the requirement string from the text, specifically:

[0019] The term frequency TF(t,d) calculates the frequency of a certain term in the requirement in the standard clause string. The formula is as follows:

[0020]

[0021] Inverse document frequency IDF(t) calculates the rarity of a word in the entire structured standard sub-database. The formula is as follows:

[0022]

[0023] Where N is the total number of standard clauses in the structured standard sub-database, t is a word in the word set, d is a standard clause string, D is a clause set, |{d∈D:t∈d}| is the number of standard clauses containing word t;

[0024] TF-IDF value: Calculates the importance of a word in the required string. The formula is as follows:

[0025] TF-IDF(t,d)=TF(t,d)×IDF(t).

[0026] Preferably, the semantic vector generation submodule in the standard and requirement association matching module is implemented using the RoBERT model;

[0027] The semantic vector similarity submodule in the standard and requirement association matching module uses cosine similarity as the basis for judging similarity;

[0028]

[0029] Among them, q is the semantic vector of the input demand string, v i It is a vector index into the structured standard sub-database.

[0030] Preferably, the relevance sorting submodule in the standard and requirement association matching module is sorted by similarity scoring, and the similarity scoring is implemented using a machine learning model based on historical matching records, specifically: using {requirement string, standard clause, similarity score} as a training set of the machine learning model based on historical matching records, training the machine learning model based on historical matching records, and the training target is the optimized similarity score. The machine learning model architecture based on historical matching records includes: an input layer for receiving requirement strings and standard clauses; a feature processing layer for extracting features from requirement strings and standard clauses, and a fully connected layer for obtaining a similarity score between the requirement strings and standard clauses based on the features; and an output layer for outputting the optimized similarity score S final The goal of training the machine learning model based on historical matching records is to minimize the difference between the similarity score between the requirement string and the standard terms in the actual historical records and the model prediction score. The loss function is the mean square error (MSE):

[0031]

[0032] in, is the similarity score between the model-predicted demand string and the standard terms, y (i) is the actual similarity score, and θ is the model parameter; the machine learning model based on historical matching records is trained until the loss function converges.

[0033] Preferably, the matching result confirmation module displays the matching result of the association between the standard and the requirement, manually modifies, confirms and supplements the matching result of the association between the standard and the requirement, and saves the matching result after confirmation; specifically:

[0034] For each requirement item and requirement content, the matching results of the standard terms and requirement content are divided into three situations. If the requirement content matches the standard terms correctly, no operation is required or direct confirmation is required; if the requirement content does not match the standard terms appropriately, the standard terms are modified, and the modified standard terms are selected from existing standards or newly defined standard names; if no matching standard terms are found for the requirement item, a newly defined standard name is given. After confirmation, click Confirm to save the matching results of the modified standard terms and requirement content.

[0035] Preferably, the standard analysis module plans the standard according to the confirmed matching results, specifically: if the demand is already met by existing standard clauses, the matched standard clauses are formed into an existing standard configuration list; if the demand is not met by relevant standard clauses, they are merged according to the given new standard name, and similar or related demands are aggregated to form a list of missing standard development items.

[0036] Preferably, the database is used to store various documents and tables, including at least a business scenario model sub-database, a standard requirement document sub-database, a structured standard sub-database and a standard system sub-database; specifically:

[0037] The business scenario model sub-database is used to store business scenario models in the form of graphs or documents; the standard requirement document sub-database uses structured storage of standard requirement documents obtained from the standard requirement document creation module; the structured standard sub-database uses structured storage of standard files, wherein the index part of the structured standard sub-database includes at least a keyword inverted index and a vector index of the standard clause content; the standard system sub-database is used to store tables, matching results and the final standard system in the standard system planning process.

[0038] The present invention also discloses a method for planning a standard system based on a business scenario model, which specifically includes the following steps:

[0039] S1, use the standard requirement document creation module to create or import a standard requirement document corresponding to the business scenario model; fill in or modify the displayed standard requirement document, and save the standard requirement document to the standard requirement document sub-database of the database after selecting Save;

[0040] S2, select the "Standard Matching" option on the display page, call the standard and requirement association matching module, and use the multi-dimensional similarity matching algorithm to query the standards that match the requirement content in the structured standard sub-database one by one for the displayed standard requirement document; the specific matching process is as follows:

[0041] S21, select a demand item with demand content, combine the demand item and demand content into a demand string, call the word segmentation submodule to use a double matching algorithm based on dictionary and probability to segment the demand string to obtain a word set, and send it to the keyword extraction submodule;

[0042] S22, the keyword extraction submodule selects keywords from the word set using the TF-IDF algorithm;

[0043] S23, using the entity recognition submodule to identify entity words from the requirement string using the Transformer-based pre-trained language model BERT;

[0044] S24, using the semantic vector generation submodule to obtain the semantic vector of the required string using the RoBERT model;

[0045] S25, according to the semantic vector of the required character string, using the semantic vector similarity submodule to use cosine similarity as a basis for judging similarity to retrieve matching standard terms from the structured standard sub-database;

[0046] S26, using the keywords obtained in step S22 and the entity words obtained in step S23 as inputs of the keyword matching submodule, and using the TF-IDF algorithm to retrieve matching standard terms from the structured standard subdatabase;

[0047] S27, merging the standard terms obtained in step S25 and the standard terms obtained in step S26 using a comprehensive matching and merging module to obtain a set of standard terms that match the requirement string;

[0048] S28, inputting each standard clause and requirement string in the standard clause set into a relevance ranking submodule, the relevance ranking submodule gives a similarity score to each standard clause and requirement string, and selects the standard clause to be returned according to the similarity score;

[0049] S29, determining whether all demand items with demand content have been processed, if not, moving to the next unprocessed demand item with demand content, returning to step S21 to continue processing, otherwise exiting the loop and executing step S3;

[0050] S3, sending the matching results to the matching result confirmation module, which displays the standard information of the requirement items, requirement contents and the matched standard terms in a table form, and allows the user to modify, confirm and supplement the standard terms;

[0051] S4, according to all the confirmed matching results, use the standard analysis module to plan the standards and save the planning results as a new standard system; specifically: if the business scenario needs are already met by relevant standard clauses, the matched standard clauses will be formed into an existing standard configuration list; if the business scenario needs are not met by relevant standard clauses, they will be merged according to the given new standard name, and similar or related needs will be aggregated to form a list of missing standard development;

[0052] S5. When the name of an existing standard system is entered or selected in the query interface and query is clicked, the standard system planning module will be used to display the existing standard system. Four page tabs will appear on the display interface, namely, business scenario modeling, standard system architecture diagram generation, standard detailed table generation, and standard system statistical table generation. By clicking on the four tabs respectively, the business scenario model, standard system architecture diagram, standard detailed table, and standard statistical results will be displayed respectively. The four tabs can be displayed and closed in combination at will.

[0053] Preferably, when the matching result confirmation module displays the requirement items, requirement contents and standard information of the matched standard clauses in a table form, it can display only the requirement contents and requirement items that are not matched with the standard clauses, and the standard information related contents are all empty.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] 1. Use standard requirement documents to make the description of business scenario requirements more systematic, complete and clear;

[0056] 2. The standard and demand association matching module uses a multi-dimensional similarity matching algorithm to make the matching of business scenario standard requirements and existing standard requirements more accurate and efficient;

[0057] 3. Using the structured standard sub-database as the demand matching object can effectively alleviate the situation where existing standard requirements are missed. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 is a flow chart of a standard system planning method driven by business scenario requirements according to the present invention;

[0059] Figure 2 This is a schematic diagram of the electric power low-carbon business scenario model of the present invention;

[0060] Figure 3It is a schematic diagram of the structure of the standard requirement document of the present invention;

[0061] Figure 4 It is a schematic diagram for confirming the matching between the requirements of the present invention and the existing standard content;

[0062] Figure 5 It is a schematic diagram of the list of missing standards for the present invention;

[0063] Figure 6 It is a schematic diagram of the standard system structure diagram of the present invention;

[0064] Figure 7 This is the standard detailed representation of the present invention;

[0065] Figure 8 It is a statistical representation of the standard system of the present invention. DETAILED DESCRIPTION

[0066] This embodiment takes the low-carbon electricity business as an example to illustrate the specific implementation of the system based on the business scenario model planning standard system. Figure 1 As shown, it includes: a database 1, a standard requirement document creation module 2, a standard and requirement association matching module 3, a matching result confirmation module 4, a standard analysis module 5 and a standard system planning module 6.

[0067] The database 1 is used to store various documents and tables, including at least a business scenario model sub-database 11, a standard requirement document sub-database 12, a structured standard sub-database 13 and a standard system sub-database 14. The business scenario model sub-database 11 is used to store business scenario models, which can be stored in the form of graphs or documents in this application. The standard requirement document sub-database 12 uses structured storage to store the standard requirement document obtained from the standard requirement document creation module. The structured standard sub-database 13 uses structured storage of standard files, that is, the standard files are structured according to the content of standard clauses, wherein the index part of the structured standard sub-database includes a keyword inverted index and a vector index of the content of standard clauses. The keyword inverted index of the structured standard sub-database 13 is an efficient search data structure for quickly finding terms and retrieval keywords in documents. Its working principle is to segment the text in the document into several terms, and create an inverted list for each term to record which documents the term appears in. Through this structure, the query no longer needs to scan the full text of each document, but directly finds the document position of the corresponding term in the inverted list, thereby greatly improving the search efficiency. The vector index of the structured standard sub-database 13 is generated by slicing the standard file into text, converting the content of the clauses into high-dimensional vectors for storage through a text vector encoder (BERT). By dividing, quantizing or approximating the high-dimensional vector data, it is possible to quickly find the vector that is most similar to the query vector. The standard system sub-database 14 is used to store the tables, matching results and final standard system in the standard system planning process.

[0068] The standard requirement document creation module 2 is used to create, modify and save standard requirement documents. The standard requirement document is based on the business scenario model, and is a structured and standardized requirement formed by decomposing, refining and expanding the business scenario model layer by layer. The requirements in the standard requirement document are usually determined by the expert group through discussion based on the business scenario model. The standard requirement document creation module enters a new standard requirement document in a tabular form or directly imports an existing standard requirement document. The standard requirement document includes at least three fields: number, requirement item, and requirement content. Each requirement is a requirement item and has a unique number. Only the bottom-level requirements have corresponding requirement content. If the requirement item can be decomposed into multiple sub-requirement items, then this requirement item has no corresponding requirement content. The numbering adopts a hierarchical structure, and the hierarchy is preferably 1-5 layers. The requirements of the same parent node are numbered sequentially. When the standard requirement document creation module creates a new initial standard requirement document, it is displayed as a blank form with three fields as column headers. Users can fill in the form according to the determined standard requirement document, or they can choose to directly import an existing standard requirement document in the standard requirement document creation module and then modify it. When the standard requirement document is completed, the standard requirement document creation module saves the completed standard requirement document to the standard requirement document sub-database in the database. In order to ensure the uniformity and universality of standard requirement documents, office software (such as Excel, etc.) or requirement management software (such as DOORS) can be used to compile standard requirement documents.

[0069] Prepare standard requirements documents based on the electricity low-carbon business scenario model. Figure 2 The following is a schematic diagram of the low-carbon business scenario model for electricity. According to the schematic diagram of the low-carbon business scenario model for electricity, the components related to the low-carbon business scenario for electricity are obtained, including carbon emission quantification, carbon emission reduction assessment, low-carbon evaluation, carbon market, and other greenhouse gases. Each business scenario section is then decomposed, refined, and expanded layer by layer to determine the demand and the demand content of the underlying demand. Use the standard demand document creation module to create a standard demand document for the low-carbon business of electricity, and fill in the blank document with the demand determined according to the low-carbon business scenario model for electricity, such as Figure 3 As shown, after completing the filling, save the standard requirement document for the low-carbon electricity business scenario to the standard requirement document sub-database.

[0070] The standard and requirement association matching module 3 uses a multi-dimensional similarity matching algorithm to query the standards that match the requirements in the structured standard sub-database one by one according to the standard requirement document, and returns the standard information, which at least includes the standard number, standard name, matching content and applicability, and can also include relevant information such as standard nature, standard level, standard category, centralized management unit, standard status, release date, implementation date, matching chapter number, etc. In order to realize the multi-dimensional similarity matching algorithm, the standard and requirement association matching module includes the following sub-modules: word segmentation sub-module 31, keyword extraction sub-module 32, entity recognition sub-module 33, semantic vector generation sub-module 34, semantic vector similarity sub-module 35, keyword matching sub-module 36, comprehensive matching merging module 37 and relevance sorting sub-module 38. The standard and requirement association matching module uses the above sub-modules to not only realize the method of multi-dimensional recall of keyword matching and semantic vector similarity, increase the recall dimension and matching range, but also use historical records to train the machine learning model to complete the sorting of the relevance between requirements and standards, thereby improving the accuracy of matching standard information with application scenario requirements. The standard and requirement association matching module only matches requirement items with requirement content and directly ignores requirement items without requirement content.

[0071] The word segmentation submodule 31 is used to segment the requirement items and requirement contents to obtain a word set. The word segmentation submodule is implemented using a dual matching algorithm based on a dictionary and probability, which can be simply described as follows: a requirement item and requirement content in the standard requirement document are combined into a requirement string as the input of the word segmentation submodule. When there is a matching term in the dictionary, the forward maximum matching algorithm (MM, Maximum Matching) is used to scan the text from left to right to find the longest matching word:

[0072] Word length = max(length of words matched in dictionary)

[0073] When there is no matching term in the dictionary, the statistical word segmentation algorithm Hidden Markov Model (HMM) is used to calculate the transition probability and emission probability P(ω) between words by training and annotating the corpus, and split the required string into possible word sequences.

[0074]

[0075] Among them, P(ω i |ω i-1 ) represents the word ω i In the word ω i-1 The probability of occurrence after that, n is a positive integer.

[0076] The keyword extraction submodule 32 is used to select keywords from the word set output by the word segmentation submodule. The keyword extraction submodule uses the TF-IDF (Term Frequency-Inverse Document Frequency) algorithm to automatically extract the words or phrases that best reflect the theme or core content of the demand string from the text. These phrases should appear frequently in the standard terms and have a high weight.

[0077] Term frequency TF(t,d): Calculates the frequency of a certain word in the requirement in the standard clause string. The formula is as follows:

[0078]

[0079] Inverse Document Frequency (IDF) (t): Calculates the rarity of a word in the entire structured standard sub-database. The formula is as follows:

[0080]

[0081] Where N is the total number of standard clauses in the structured standard sub-database, t is a word in the word set, d is a standard clause string, D is a clause set, and |{d∈D:t∈d}| is the number of standard clauses containing word t.

[0082] TF-IDF value: Calculates the importance of a word in the required string. The formula is as follows:

[0083] TF-IDF(t,d)=TF(t,d)×IDF(t).

[0084] The entity recognition submodule 33 is used to identify entity words with specific meanings from the requirement string. It uses the Transformer-based pre-trained language model BERT to capture the bidirectional dependency of words in the context, complete the entity word recognition task, and identify the entity words in the requirement.

[0085] The semantic vector generation submodule 34 is used to generate a vector according to the required string, using the RoBERT model, and mapping the required string into a high-dimensional semantic space through a pre-trained deep learning model to obtain a semantic vector of the required string.

[0086] The semantic vector similarity submodule 35 is used to search through the vector index of the structured standard sub-database according to the generated semantic vector of the demand string, so as to quickly match potential standard terms and obtain standard terms that match the demand string. The semantic vector similarity submodule uses cosine similarity as the basis for judging similarity.

[0087]

[0088] Among them, q is the semantic vector of the input demand string, v i is the vector index of the structured standard sub-database, |||| is the norm calculation symbol, cosin_similarity() is the cosine similarity function, which can be implemented using the cosine similarity function already available in the prior art.

[0089] The keyword matching submodule 36 is used to compare the keywords obtained by the keyword extraction submodule and the entity words obtained by the entity recognition submodule with the keyword index in the structured standard subdatabase to obtain standard terms that match the requirement string. The keyword matching submodule is implemented using the TF-IDF algorithm.

[0090] The comprehensive matching and merging module 37 merges the standard terms obtained by the keyword matching submodule and the standard terms obtained by the semantic vector similarity submodule for the same requirement string to obtain a set of standard terms that matches the requirement string.

[0091] The relevance sorting submodule 38 sorts the standard clauses that match the requirement string to ensure that the standard clauses returned to the user best match the requirement string. The relevance sorting submodule determines the standard clause that is most relevant to the requirement string by scoring the similarity between the requirement string and each standard clause in the set of standard clauses. The similarity score is implemented using a machine learning model based on historical matching records, specifically: using {requirement string, standard clause, similarity score} as the training set of the machine learning model based on historical matching records, the machine learning model based on historical matching records is trained, and the training target is the optimized similarity score. The machine learning model architecture based on historical matching records includes: an input layer for receiving requirement strings and standard clauses; a feature processing layer for extracting features from requirement strings and standard clauses, and the feature extraction can extract feature vectors through the text encoder BERT, and the feature processing layer can also include further vectorization or nonlinear transformation; a fully connected layer is used to obtain the similarity score of the requirement string and the standard clause based on the features; and an output layer outputs the optimized similarity score S final The goal of training a machine learning model based on historical matching records is to minimize the difference between the similarity score between the requirement string and the standard terms in the actual historical records and the model prediction score. The commonly used loss function is the mean square error (MSE):

[0092]

[0093] in, is the similarity score between the model-predicted demand string and the standard terms, y (i)is the actual similarity score, and θ is the model parameter. The machine learning model based on historical matching records is trained until the loss function converges.

[0094] The input of the relevance sorting submodule 38 is the requirement string and the set of standard clauses matched with it obtained by the comprehensive matching merging module. The requirement string and each standard clause in the set of standard clauses are combined as a machine learning model based on historical matching records after training to obtain a similarity score, which is the applicability in the standard information. The standard clauses are sorted according to the similarity score, and the standard clauses corresponding to the requirement string are selected according to the threshold as the standard clauses matching the requirement item. The standard clauses with a similarity score greater than the specified threshold can be used as the matched standard clauses to be returned. If the threshold is 70%, the standard clauses with a similarity score greater than 70% are considered to match the requirement item. Alternatively, the standard clauses before the specified threshold after sorting by the similarity score can be returned as the matched standard clauses. If the threshold is set to 1, the standard clause with the largest similarity score after sorting is returned as the matched standard clause, and the matching result is saved. Therefore, it can be seen that for a requirement item, there may be only one standard clause matching it, or there may be multiple standard clauses matching it.

[0095] The matching result confirmation module 4 displays the matching results of the association between standards and requirements, and improves the matching results of the association between standards and requirements through manual modification, confirmation and supplementation. The standard information of the requirement items, requirement contents and matched standard clauses is displayed in a table. For each requirement item and requirement content, the matching results of the association between standard clauses and requirement content are divided into three situations. If the requirement content matches the standard clause correctly, no operation is required or direct confirmation is sufficient; if the requirement content does not match the standard clause appropriately, the standard clause is modified, and the modified standard clause can be selected from the existing standards or a newly defined standard name; if no matching standard clause is found for the requirement item, a newly defined standard name is given. After confirmation, click Confirm to save the matching results of the modified standard clause and requirement content. In this embodiment, the schematic diagram for confirming the matching of requirements with existing standard content is shown in the figure below. Figure 4As shown, the requirement items and requirement contents are obtained from the standard requirement documents. The requirement contents of the requirement items cannot be empty. The matching contents are the standard clause contents that are matched. The standard number and standard name are the standard number and standard name of the standard where the matched standard clause is located. If there is no match, the contents related to the standard clause are all displayed as empty. The applicability is the similarity score of the standard clause. In order to make it easier to review and supplement, the matching result confirmation module can count the matching results on the basis of displaying the matching results, and only display the requirement contents that are not matched to the standard clauses, so as to facilitate the user to supplement. The matching result confirmation module saves the confirmed matching results.

[0096] The standard analysis module 5 plans the standard according to the confirmed matching results and saves the planning results as a new standard system. The standard analysis module analyzes the confirmed matching results. If the business scenario requirements are already met by relevant standard clauses, the matching standard clauses will be formed into an existing standard configuration list; if the business scenario requirements are not met by relevant standard clauses, they will be merged according to the given new standard name, and similar or related requirements will be aggregated to form a list of missing standard development. Figure 5 As shown in the figure, the missing standard development list information includes the name of the missing standard, main content, standard nature, standard level, standard category, and urgency. The missing standard names in the missing standard development list are newly defined standard names. The contents in the missing standard development list are modified, supplemented, and improved by experts in the subsequent standard formulation process. The standard analysis module only provides these fields as suggestions. In the aggregation process of missing standards, highly similar or repeated standard requirements should be normalized to ensure the completeness, accuracy, and uniqueness of the missing standards.

[0097] The standard system planning module 6 is used to display the existing standard system. After the standard system planning module enters the name of the standard system in the query interface and clicks on the query, four page tabs will appear: business scenario modeling, standard system structure diagram generation, standard detail table generation, and standard system statistical table generation. By clicking on the four tabs respectively, the business scenario model, standard system structure diagram, standard detail table, and standard statistical results will be displayed respectively. The four tabs can be displayed and closed in combination at will. The business scenario model is obtained directly from the business scenario model sub-database, the standard system structure diagram is obtained based on the standard requirements document, and the standard detail table and standard system statistical table are obtained based on the corresponding standard system. The standard system for the low-carbon power business is named Power Low-Carbon Standard System 3.0 when it is saved using the standard analysis module. Therefore, the standard system structure diagram of the low-carbon power standard system 3.0 is as follows Figure 6 As shown in the standard table Figure 7 As shown in the table below, the standard system statistics are as follows Figure 8 shown.

[0098] The present invention also discloses a method for planning a standard system based on a business scenario model, which specifically includes the following steps:

[0099] S1, use the standard requirement document creation module to create or import a standard requirement document corresponding to the business scenario model; fill in or modify the displayed standard requirement document, and save the standard requirement document to the standard requirement document sub-database 12 of database 1 after selecting Save.

[0100] S2, select the "Standard Matching" option on the display page, call the standard and requirement association matching module 3, use the multi-dimensional similarity matching algorithm to query the standards that match the requirement content in the structured standard sub-database one by one for the displayed standard requirement document; the specific matching process is as follows:

[0101] S21, select a demand item with demand content, combine the demand item and demand content into a demand string, call the word segmentation submodule 31 to use a double matching algorithm based on dictionary and probability to segment the demand string to obtain a word set, and send it to the keyword extraction submodule 32.

[0102] S22, the keyword extraction submodule 32 selects keywords from the word set using the TF-IDF algorithm.

[0103] S23, using the entity recognition submodule 33 to identify entity words from the required character string using the Transformer-based pre-trained language model BERT.

[0104] S24, using the semantic vector generation submodule 34 to obtain the semantic vector of the required character string using the RoBERT model.

[0105] S25, according to the semantic vector of the required character string, the semantic vector similarity submodule 35 uses cosine similarity as a basis for judging similarity to retrieve matching standard terms from the structured standard sub-database.

[0106] S26, the keywords obtained from step S22 and the entity words obtained from step S23 are used as inputs of the keyword matching submodule 36, and the TF-IDF algorithm is used to retrieve matching standard terms from the structured standard sub-database.

[0107] S27, the standard terms obtained in step S25 and the standard terms obtained in step S26 are merged using the comprehensive matching and merging module 37 to obtain a set of standard terms that match the requirement character string.

[0108] S28, each standard clause and requirement string in the standard clause set is input into the relevance sorting submodule 38, the relevance sorting submodule 38 gives a similarity score to each standard clause and requirement string, and selects the standard clause to be returned based on the similarity score.

[0109] S29, determine whether all demand items with demand content have been processed. If not, move to the next unprocessed demand item with demand content and return to step S21 to continue processing. Otherwise, exit the loop and execute step S3.

[0110] S3, sending the matching results to the matching result confirmation module 4, the matching result confirmation module 4 displays the standard information of the requirement items, requirement contents and the matched standard clauses in a table form, and allows the user to modify, confirm and supplement the standard clauses; further, for more convenient confirmation, only the requirement contents and requirement contents that have not matched the standard clauses can be displayed, and the standard information related contents are all empty at this time, so as to facilitate the user to supplement; the confirmed matching results are saved in the standard system sub-database 14 of the database.

[0111] S4, use the standard analysis module 5 to plan the standards according to all the confirmed matching results and save the planning results as a new standard system; specifically: if the needs of the business scenario are already met by relevant standard clauses, the matched standard clauses will form an existing standard configuration list; if the needs of the business scenario are not met by relevant standard clauses, they will be merged according to the given new standard name, and similar or related needs will be aggregated to form a list of missing standard development.

[0112] S5. When the name of an existing standard system is entered or selected in the query interface and query is clicked, the standard system planning module 6 will be used to display the existing standard system. Four page tabs will appear on the display interface, namely, business scenario modeling, standard system structure diagram generation, standard detailed table generation, and standard system statistical table generation. By clicking on the four tabs respectively, the business scenario model, standard system structure diagram, standard detailed table, and standard statistical results will be displayed respectively. The four tabs can be displayed and closed in combination at will.

[0113] Finally, it should be noted that the above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions or parameters described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A system for planning a standard system based on a business scenario model, characterized in that: It includes the following modules: Database, standard requirement document creation module, standard and requirement association matching module, matching result confirmation module, standard analysis module and standard system planning module; among which: The database is used to store various documents and tables, including at least a business scenario model sub-database, a standard requirement document sub-database, a structured standard sub-database, and a standard system sub-database; The standard requirement document creation module is used to create, modify and save standard requirement documents. The standard requirement document creation module enters a new standard requirement document in a table form or directly imports an existing standard requirement document. The standard requirement document includes at least three fields: number, requirement item, and requirement content. Each requirement is a requirement item and has a unique number. Only the bottom-level requirement has corresponding requirement content. If the requirement item can be decomposed into multiple sub-requirement items, this requirement item has no corresponding requirement content. When the standard requirement document is completed, the standard requirement document creation module saves the completed standard requirement document into the standard requirement document sub-database in the database; The standard and requirement association matching module uses a multi-dimensional similarity matching algorithm to query the standards that match the requirement content in the structured standard sub-database one by one according to the standard requirement document, and returns the standard information. The standard information at least includes the standard number, standard name, matching content and applicability. The standard and requirement association matching module includes the following sub-modules: word segmentation sub-module, keyword extraction sub-module, entity recognition sub-module, semantic vector generation sub-module, semantic vector similarity sub-module, keyword matching sub-module, comprehensive matching merging module and relevance sorting sub-module; the word segmentation sub-module is used to obtain a word set after word segmentation of the requirement items and requirement content, the keyword extraction sub-module is used to select keywords from the word set output by the word segmentation sub-module, the entity recognition sub-module is used to identify entity words with specific meanings from the requirement character string, and the semantic vector generation sub-module is used to A vector is generated according to the requirement string. The semantic vector similarity submodule is used to retrieve the generated semantic vector of the requirement string through the vector index of the structured standard sub-database to obtain standard clauses that match the requirement string. The keyword matching submodule is used to compare the keywords obtained by the keyword extraction submodule and the entity words obtained by the entity recognition submodule with the keyword index in the structured standard sub-database to obtain standard clauses that match the requirement string. The comprehensive matching merging module will merge the standard clauses obtained by the keyword matching submodule and the standard clauses obtained by the semantic vector similarity submodule for the same requirement string to obtain a set of standard clauses that match the requirement string. The relevance sorting submodule sorts the standard clauses that match the requirement string by similarity score and returns the standard clauses that meet the threshold requirements. The matching result confirmation module displays the matching results of the association between the standard and the requirement, and manually modifies, confirms and supplements the matching results of the association between the standard and the requirement, and saves the matching results after confirmation; The standard analysis module plans the standards according to the confirmed matching results and saves the planning results as a new standard system; The standard system planning module is used to display the existing standard system. After the standard system name is entered in the query interface and query is clicked, four page tabs will appear: business scenario modeling, standard system structure diagram generation, standard detailed table generation, and standard system statistical table generation. By clicking on the four tabs respectively, the business scenario model, standard system structure diagram, standard detailed table, and standard statistical results will be displayed respectively. The four tabs can be displayed and closed in combination at will.

2. The system for planning a standard system based on a business scenario model according to claim 1, characterized in that: The word segmentation submodule in the standard and requirement association matching module is implemented by a dual matching algorithm based on a dictionary and probability, specifically: a requirement item and the requirement content in the standard requirement document are combined into a requirement string as the input of the word segmentation submodule. When there is a matching term in the dictionary, the forward maximum matching algorithm is used to scan the text from left to right to find the longest matching word: Word length = max(length of words matched in dictionary) When there is no matching term in the dictionary, the statistical word segmentation algorithm Hidden Markov Model (HMM) is used to calculate the transition probability and emission probability between words through training and annotating the corpus, and split the required string into possible word sequences; Among them, P(ω i |ω i-1 ) represents the word ω i In the word ω i-1 The probability of appearing later.

3. The system for planning a standard system based on a business scenario model according to claim 1, characterized in that: The keyword extraction submodule in the standard and demand association matching module uses the TF-IDF algorithm to automatically extract the words or phrases that best reflect the theme or core content of the demand string from the text, specifically: The term frequency TF(t,d) calculates the frequency of a certain term in the requirement in the standard clause string. The formula is as follows: Inverse document frequency IDF(t) calculates the rarity of a word in the entire structured standard sub-database. The formula is as follows: Where N is the total number of standard clauses in the structured standard sub-database, t is a word in the word set, d is a standard clause string, D is a clause set, |{d∈D:t∈d}| is the number of standard clauses containing word t; TF-IDF calculates the importance of a word in the required string. The formula is as follows: TF-IDF(t,d)×TF(t,d)×IDF(t).

4. The system for planning a standard system based on a business scenario model according to claim 1, characterized in that: The semantic vector generation submodule in the standard and requirement association matching module is implemented using the RoBERT model; The semantic vector similarity submodule in the standard and requirement association matching module uses cosine similarity as a basis for judging similarity.

5. The system for planning a standard system based on a business scenario model according to claim 1, characterized in that: The relevance sorting submodule in the standard and requirement association matching module is sorted by similarity scoring, and the similarity scoring is implemented using a machine learning model based on historical matching records, specifically: using {requirement string, standard clause, similarity score} as a training set of the machine learning model based on historical matching records, training the machine learning model based on historical matching records, and the training target is the optimized similarity score. The machine learning model architecture based on historical matching records includes: an input layer for receiving requirement strings and standard clauses; a feature processing layer for extracting features from requirement strings and standard clauses, and a fully connected layer for obtaining a similarity score between the requirement strings and standard clauses based on the features; and an output layer for outputting an optimized similarity score S final The goal of training the machine learning model based on historical matching records is to minimize the difference between the similarity score between the requirement string and the standard terms in the actual historical records and the model prediction score. The loss function is the mean square error (MSE): in, is the similarity score between the model-predicted demand string and the standard terms, y (i) is the actual similarity score, and θ is the model parameter; the machine learning model based on historical matching records is trained until the loss function converges.

6. The system for planning a standard system based on a business scenario model according to claim 1, characterized in that: The matching result confirmation module displays the matching results of the association between the standard and the requirement, manually modifies, confirms and supplements the matching results of the association between the standard and the requirement, and saves the matching results after confirmation; specifically: For each requirement item and requirement content, the matching results of the standard terms and requirement content are divided into three situations. If the requirement content matches the standard terms correctly, no operation is required or direct confirmation is required; if the requirement content does not match the standard terms appropriately, the standard terms are modified, and the modified standard terms are selected from existing standards or newly defined standard names; if no matching standard terms are found for the requirement item, a newly defined standard name is given. After confirmation, click Confirm to save the matching results of the modified standard terms and requirement content.

7. The system for planning a standard system based on a business scenario model according to claim 1, characterized in that: The standard analysis module performs standard planning according to the confirmed matching results, specifically: If the demand is already met by existing standard clauses, the matching standard clauses will be used to form an existing standard configuration list; if the demand is not met by relevant standard clauses, they will be merged according to the given new standard name, and similar or related demands will be aggregated to form a list of missing standard development.

8. The system for planning a standard system based on a business scenario model according to claim 1, characterized in that: The database is used to store various documents and tables, including at least a business scenario model sub-database, a standard requirement document sub-database, a structured standard sub-database and a standard system sub-database; specifically: The business scenario model sub-database is used to store business scenario models in the form of graphs or documents; the standard requirement document sub-database uses structured storage of standard requirement documents obtained from the standard requirement document creation module; the structured standard sub-database uses structured storage of standard files, wherein the index part of the structured standard sub-database includes at least a keyword inverted index and a vector index of the standard clause content; the standard system sub-database is used to store tables, matching results and the final standard system in the standard system planning process.

9. A method for planning a standard system based on a business scenario model using the system described in claims 1-8, characterized in that: The specific steps include: S1, use the standard requirement document creation module to create or import a standard requirement document corresponding to the business scenario model; fill in or modify the displayed standard requirement document, and save the standard requirement document to the standard requirement document sub-database of the database after selecting Save; S2, select the standard matching option on the display page, call the standard and requirement association matching module, use the multi-dimensional similarity matching algorithm to query the standards that match the requirement content in the structured standard sub-database one by one for the displayed standard requirement document; the specific matching process is as follows: S21, select a demand item with demand content, combine the demand item and demand content into a demand string, call the word segmentation submodule to use a double matching algorithm based on dictionary and probability to segment the demand string to obtain a word set, and send it to the keyword extraction submodule; S22, the keyword extraction submodule selects keywords from the word set using the TF-IDF algorithm; S23, using the entity recognition submodule to identify entity words from the requirement string using the Transformer-based pre-trained language model BERT; S24, using the semantic vector generation submodule to obtain the semantic vector of the required string using the RoBERT model; S25, according to the semantic vector of the required character string, the semantic vector similarity submodule uses cosine similarity as a basis for judging similarity to retrieve matching standard terms from the structured standard sub-database; S26, using the keywords obtained in step S22 and the entity words obtained in step S23 as inputs of the keyword matching submodule, and using the TF-IDF algorithm to retrieve matching standard terms from the structured standard subdatabase; S27, merging the standard terms obtained in step S25 and the standard terms obtained in step S26 using a comprehensive matching and merging module to obtain a set of standard terms that match the requirement string; S28, inputting each standard clause and requirement string in the standard clause set into a relevance ranking submodule, the relevance ranking submodule gives a similarity score to each standard clause and requirement string, and selects the standard clause to be returned according to the similarity score; S29, determining whether all demand items with demand content have been processed, if not, moving to the next unprocessed demand item with demand content, returning to step S21 to continue processing, otherwise exiting the loop and executing step S3; S3, sending the matching results to the matching result confirmation module, which displays the standard information of the requirement items, requirement contents and the matched standard clauses in a table form, and allows the user to modify, confirm and supplement the standard clauses; S4, using the standard analysis module to perform standard planning according to all confirmed matching results and save the planning results as a new standard system; Specifically: if the requirements of the business scenario are already met by relevant standard clauses, the matching standard clauses will be used to form an existing standard configuration list; if the requirements of the business scenario are not met by relevant standard clauses, they will be merged according to the given new standard name, and similar or related requirements will be aggregated to form a list of missing standard development; S5. When the name of an existing standard system is entered or selected in the query interface and query is clicked, the standard system planning module will be used to display the existing standard system. Four page tabs will appear on the display interface, namely, business scenario modeling, standard system architecture diagram generation, standard detailed table generation, and standard system statistical table generation. By clicking on the four tabs respectively, the business scenario model, standard system architecture diagram, standard detailed table, and standard statistical results will be displayed respectively. The four tabs can be displayed and closed in combination at will.

10. The method for planning a standard system based on a business scenario model according to claim 9, characterized in that: The matching result confirmation module displays the standard information of the requirement items, requirement contents and matched standard clauses in a table format; specifically: It is possible to display only the requirement contents and requirement items that do not match the standard terms. In this case, the standard information related contents are empty.

Images