Database extension method and device, equipment, storage medium and product

Abstract

The invention discloses a database extension method and device, equipment, a storage medium and a product. The method comprises the following steps: acquiring a plurality of candidate attributes extracted from an engineering quality safety data source; performing compliance evaluation on the plurality of candidate attributes to generate corresponding attribute evaluation coefficients; wherein the compliance evaluation at least comprises the following dimensions: structural integrity, time continuity and similarity with stored attributes; screening out a target attribute from the plurality of candidate attributes based on the attribute evaluation coefficient; and based on the target attribute, creating a corresponding data table in a project quality safety database. According to the scheme, the candidate attributes extracted from the engineering quality safety data source are subjected to multi-dimensional quantitative compliance evaluation, and the target attributes meeting the requirements are automatically screened based on the evaluation to create the data table, so that the problem of low efficiency caused by dependence on manual screening in the prior art is solved; and the efficiency of database expansion and the consistency, normalization and availability of the stored attributes are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a database expansion method, apparatus, device, storage medium, and product. Background Technology

[0002] In the field of engineering quality and safety management, building and maintaining a high-quality database is crucial. Currently, methods using large language models to automatically extract attributes from engineering quality and safety texts to expand the database have been applied. However, the attributes extracted from existing large language models often fail to meet the requirements of engineering quality and safety database design principles, paradigm theory, and database management system usage; furthermore, the accuracy of attributes extracted from large language models is low, still requiring significant manual screening of the extracted attributes, making it difficult to adapt to the rapid expansion needs of massive amounts of data in engineering scenarios. Summary of the Invention

[0003] This invention provides a database expansion method, apparatus, device, storage medium, and product to solve the problem of low database expansion efficiency caused by the inconsistent quality of attributes directly extracted from large language models and their non-compliance with database specifications in the prior art.

[0004] According to one aspect of the present invention, a database expansion method is provided, the method comprising:

[0005] Retrieve multiple candidate attributes extracted from engineering quality and safety data sources;

[0006] Compliance assessments are performed on multiple candidate attributes to generate corresponding attribute assessment coefficients. The compliance assessment includes at least the following dimensions: structural integrity, temporal continuity, and similarity to attributes already in the database.

[0007] The target attribute is selected from multiple candidate attributes based on the attribute evaluation coefficient.

[0008] Based on the target attributes, create corresponding data tables in the engineering quality and safety database.

[0009] According to another aspect of the present invention, a database expansion device is provided, the device comprising:

[0010] The attribute retrieval module is used to retrieve multiple candidate attributes extracted from the engineering quality and safety data source;

[0011] The attribute evaluation module is used to evaluate the compliance of multiple candidate attributes and generate corresponding attribute evaluation coefficients. The compliance evaluation includes at least the following dimensions: structural integrity, temporal continuity, and similarity to attributes already in the database.

[0012] The attribute filtering module is used to filter out target attributes from multiple candidate attributes based on attribute evaluation coefficients;

[0013] The data table creation module creates corresponding data tables in the engineering quality and safety database based on the target attributes.

[0014] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0015] At least one processor; and

[0016] A memory communicatively connected to the at least one processor; wherein,

[0017] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the database expansion method according to any embodiment of the present invention.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the database expansion method according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the database expansion method described in any embodiment of the present invention.

[0020] The technical solution of this invention involves acquiring multiple candidate attributes extracted from engineering quality and safety data sources; conducting compliance assessments on each candidate attribute to generate corresponding attribute assessment coefficients; wherein the compliance assessment includes at least the following dimensions: structural integrity, temporal continuity, and similarity to existing attributes; selecting target attributes from the candidate attributes based on the attribute assessment coefficients; and creating a corresponding data table in the engineering quality and safety database based on the target attributes. This technical solution effectively solves the inefficiency problem caused by manual screening in existing technologies by conducting multi-dimensional quantitative compliance assessments of candidate attributes extracted from engineering quality and safety data sources, covering structural integrity, temporal continuity, and similarity to existing data, and automatically selecting qualified target attributes to create data tables. This significantly improves the efficiency of database expansion and the consistency, standardization, and usability of the attributes entered into the database.

[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a database expansion method provided according to Embodiment 1 of the present invention;

[0024] Figure 2 This is a flowchart of a database expansion method provided according to Embodiment 2 of the present invention;

[0025] Figure 3 This is a flowchart of a database expansion method provided according to Embodiment 3 of the present invention;

[0026] Figure 4 This is a schematic diagram of the structure of a database expansion device according to Embodiment 4 of the present invention;

[0027] Figure 5 This is a schematic diagram of the structure of an electronic device that implements the database expansion method of this invention. Detailed Implementation

[0028] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] Example 1

[0031] Figure 1 This is a flowchart of a database expansion method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where attributes are intelligently extracted from data sources related to engineering quality and safety, and the database is automatically expanded. This method can be executed by a database expansion device, which can be implemented in hardware and / or software, and can be configured in an electronic device. Figure 1 As shown in the figure, the database expansion method provided in this embodiment includes the following steps:

[0032] S110. Obtain multiple candidate attributes extracted from the engineering quality and safety data source.

[0033] Among them, the data source of engineering quality and safety can refer to various information carriers that carry information related to engineering quality and safety. It is the original data source for attribute extraction, and may include, but is not limited to, quality inspection report texts, engineering testing forms, safety hazard rectification records, sensor data logs, engineering drawing annotation information, construction process ledgers, etc.

[0034] Candidate attributes can refer to attributes that are initially extracted from engineering quality and safety data sources, are considered to be valuable for building an engineering quality and safety database, but have not yet undergone further compliance verification. They are the candidates for the final database attributes, and may include, but are not limited to, project number, quality inspection records, test index values, safety hazard level, rectification completion status, project location, and test time.

[0035] In this embodiment of the invention, various engineering quality and safety data sources, such as quality inspection reports and engineering testing forms, can be collected and subjected to unified data cleaning and preprocessing. Then, a pre-configured attribute extraction algorithm or model can be called to extract candidate attributes related to engineering quality and safety from the processed data. These candidate attributes will provide high-quality basic data for subsequent compliance assessment and target attribute screening, ensuring the compliance and reliability of the engineering quality and safety database expansion.

[0036] S120. Conduct compliance assessments on multiple candidate attributes and generate corresponding attribute assessment coefficients. The compliance assessment includes at least the following dimensions: structural integrity, temporal continuity, and similarity to attributes already in the database.

[0037] Compliance assessment refers to a multi-dimensional process of checking and judging whether candidate attributes meet the inclusion criteria of the engineering quality and safety database, based on the database's design standards, data storage requirements, and business usage needs. Compliance assessment can include at least the following dimensions: ① Structural integrity, i.e., the degree to which the organization of candidate attributes conforms to database design paradigms (first, second, and third normal forms) and field constraint rules (primary keys, foreign keys, uniqueness constraints, etc.), with the core being ensuring that data storage is free of logical defects and structurally sound; ② Temporal continuity, i.e., the absence of missing or abnormally discontinuous records corresponding to candidate attributes, fully reflecting the characteristics of the temporal evolution of engineering quality and safety-related events; ③ Similarity to already included attributes, i.e., the degree to which candidate attributes are duplicated or similar to attributes already stored in the database in terms of content and characteristics, used to determine whether data redundancy exists.

[0038] Attributes already included in the database refer to attributes in data tables that have passed compliance assessments and been successfully incorporated into the engineering quality and safety database. They serve as a benchmark for judging whether candidate attributes are redundant.

[0039] The attribute evaluation coefficient is a quantitative score determined after evaluating the compliance of candidate attributes. It is a core indicator that represents the overall compliance level of candidate attributes and a direct basis for subsequent attribute screening.

[0040] In this embodiment of the invention, each candidate attribute can be evaluated for compliance in multiple dimensions. For example, structural integrity can be evaluated by checking its key constraints, temporal continuity can be evaluated by analyzing the dispersion of its corresponding time series, and similarity can be evaluated by comparing it with attributes already in the database. Finally, the evaluation results of each dimension are fused to obtain the attribute evaluation coefficient corresponding to each candidate attribute, so as to fully present the comprehensive level of whether each candidate attribute meets the requirements for inclusion in the engineering quality and safety database.

[0041] S130. Select the target attribute from multiple candidate attributes based on the attribute evaluation coefficient.

[0042] Among them, the target attribute can refer to the attribute that is determined to meet the requirements of structural integrity, temporal continuity and no redundancy of the engineering quality and safety database after comparing the attribute evaluation coefficient with the preset threshold. It is the core data unit used to create new data tables in the database.

[0043] In this embodiment of the invention, the attribute evaluation coefficient of each candidate attribute can be compared with a preset threshold, and the candidate attributes that are less than the preset threshold are determined as target attributes. These target attributes are the attributes that will be subsequently added to the database.

[0044] S140. Based on the target attributes, create the corresponding data table in the engineering quality and safety database.

[0045] Among them, the engineering quality and safety database can refer to a database system used to store and manage engineering quality and safety related data. It needs to meet the business needs of engineering quality and safety management and have the characteristics of data consistency, reliability, and correlation. For example, it can include various data tables such as basic project information, quality inspection, safety hazards, and rectification tracking.

[0046] In this embodiment of the invention, the target attributes selected through screening can first be sorted and standardized to clarify their naming, logical relationships, and business meanings. Then, based on the attribute characteristics, field types and constraints are designed, and the data table structure is designed in accordance with the requirements of the three normal forms. Foreign keys are then established to associate with existing tables in the database. Subsequently, data tables are created in the database system, and multi-dimensional verification is used to ensure that the data table structure is compliant and the data is valid. Finally, the verified data tables are incorporated into the engineering quality and safety database, and the metadata is updated, thereby completing the expansion of the database.

[0047] The technical solution of this invention involves acquiring multiple candidate attributes extracted from engineering quality and safety data sources; conducting compliance assessments on each candidate attribute to generate corresponding attribute assessment coefficients; wherein the compliance assessment includes at least the following dimensions: structural integrity, temporal continuity, and similarity to existing attributes; selecting target attributes from the candidate attributes based on the attribute assessment coefficients; and creating a corresponding data table in the engineering quality and safety database based on the target attributes. This technical solution effectively solves the inefficiency problem caused by manual screening in existing technologies by conducting multi-dimensional quantitative compliance assessments of candidate attributes extracted from engineering quality and safety data sources, covering structural integrity, temporal continuity, and similarity to existing data, and automatically selecting qualified target attributes to create data tables. This significantly improves the efficiency of database expansion and the consistency, standardization, and usability of the attributes entered into the database.

[0048] Example 2

[0049] Figure 2 This is a flowchart of a database expansion method provided in Embodiment 2 of the present invention. It is further optimized and expanded based on the above embodiments and can be combined with various optional technical solutions in the above embodiments. For example... Figure 2 As shown in the figure, the database expansion method provided in this embodiment includes the following steps:

[0050] S210. Input the engineering quality and safety data source into the preset attribute extraction model to extract candidate attributes; wherein, the preset attribute extraction model includes at least a large language model.

[0051] The preset attribute extraction model can refer to a model that has been trained and optimized in advance for the attribute requirements of the engineering quality and safety database, and is specifically used to identify and extract the required attributes from various data sources. For example, the preset attribute extraction model can include at least a large language model.

[0052] Large language models can refer to artificial intelligence models trained on large-scale text data that possess powerful natural language understanding, semantic analysis, and information extraction capabilities. In this solution, the model serves as the core engine and is specifically used to accurately identify domain-specific terms, parse engineering-related text logic, and extract core attribute information, such as BERT and GPT models.

[0053] In this embodiment of the invention, multi-source data such as quality inspection reports, testing logs, and rectification records in the field of engineering quality and safety can be collected. After preprocessing such as format standardization and content cleaning, the data is input into a pre-trained preset attribute extraction model. The model uses its deep semantic analysis capabilities to automatically scan and understand the text content, accurately identify key information such as project number, testing indicators, risk level, and responsible entity, and then filter out invalid content to obtain candidate attributes that meet the basic requirements of database attributes, providing basic data for subsequent compliance assessment.

[0054] S220. Determine the structural integrity coefficient of each candidate attribute, and record the fluctuation coefficient and similarity coefficient.

[0055] The structural integrity coefficient is a parameter used to quantitatively characterize the degree to which the structure of candidate attributes conforms to the constraints and specifications of the engineering quality and safety database. It can be determined based on the violation rates of primary key constraints, uniqueness constraints, and foreign key constraints, and is a core indicator for assessing the compliance of attribute structures. The primary key constraint violation rate is the ratio of the number of non-primary key attributes without corresponding primary keys to the total number of candidate attributes, reflecting the integrity of the primary key association of candidate attributes; a smaller ratio indicates better primary key constraint compliance. The uniqueness constraint violation rate is the ratio of the number of attributes violating uniqueness constraints to the total number of candidate attributes, reflecting the uniqueness of candidate attribute records; a smaller ratio indicates no duplicate attributes and better uniqueness compliance. The foreign key constraint violation rate is the ratio of the number of records where the foreign key is inconsistent with the attributes in the related table to the total number of candidate attributes, reflecting the consistency of the association between candidate attributes and the related data table; a smaller ratio indicates better foreign key constraint compliance.

[0056] The record fluctuation coefficient can be a parameter used to quantify the time dispersion of candidate attribute records. It can be determined based on the dispersion of the corresponding time points of the candidate attribute and is a core indicator for assessing the temporal continuity of an attribute. The record time point can refer to the actual time information of the occurrence or recording of the candidate attribute (such as the execution time of quality inspection, the reporting time of safety hazards, the completion time of rectification, etc.), and is the basic data for judging temporal continuity. The dispersion can refer to the degree of deviation between each record time point of the candidate attribute and the average record time. The smaller the dispersion, the more concentrated the time distribution and the stronger the continuity.

[0057] The similarity coefficient can be a parameter used to quantify the redundancy between candidate attributes and attributes already included in the engineering quality and safety database. It can be determined based on the similarity between candidate attributes and attributes already included in the database and is a core indicator for evaluating the uniqueness of data.

[0058] In this embodiment of the invention, for each candidate attribute extracted from the engineering quality and safety data source, the following steps are taken: First, the number of non-primary key attributes without corresponding primary keys, attributes violating uniqueness constraints, and attributes whose foreign keys are inconsistent with the associated tables are counted. The ratio of the three types of constraints is calculated, and a structural integrity coefficient is obtained to quantitatively evaluate the compliance of the attribute structure. Then, all record time points of each candidate attribute are extracted, and the record fluctuation coefficient is obtained by calculating the average record time and the dispersion of each time point to quantify the temporal continuity of the attribute. Subsequently, a candidate attribute set is constructed by associating primary keys with non-primary keys. The database tables that have been entered into the database are traversed, and the similarity between the two is calculated. The average of the maximum similarity is taken as the similarity coefficient to quantify the redundancy of the candidate attribute. Finally, the candidate attributes are comprehensively evaluated in terms of structure, time, and uniqueness through the three coefficients, providing a basis for the generation of subsequent attribute evaluation coefficients.

[0059] S230. Based on the structural integrity coefficient, record fluctuation coefficient, and similarity coefficient, determine the attribute evaluation coefficient corresponding to each candidate attribute.

[0060] In this embodiment of the invention, for each candidate attribute, the structural integrity coefficient, record fluctuation coefficient and similarity coefficient determined above can be fused, for example by weighted summation, so as to transform the multi-dimensional compliance assessment indicators into a single attribute assessment coefficient.

[0061] S240. Compare the evaluation coefficients of each attribute with preset thresholds, and determine the candidate attributes that are less than the preset thresholds as target attributes.

[0062] In this embodiment of the invention, a scientifically reasonable preset threshold for attribute evaluation coefficients can be determined in advance based on historical compliance attribute data and domain risk tolerance of the engineering quality and safety database. Subsequently, for each candidate attribute extracted from the engineering quality and safety data source, its attribute evaluation coefficient generated through compliance assessment is retrieved, and this coefficient is compared one by one with the preset threshold. For candidate attributes with evaluation coefficients less than the preset threshold, they are directly determined as target attributes that meet the database expansion requirements. Furthermore, for candidate attributes with evaluation coefficients greater than or equal to the preset threshold, a warning signal containing attribute identifiers and non-compliant dimensions is generated. After targeted verification by professionals, a decision is made based on the verification results whether to add them as target attributes.

[0063] S250. Based on the target attributes, create the corresponding data table in the engineering quality and safety database.

[0064] The technical solution of this invention significantly improves the coverage and semantic understanding depth of information extraction by accurately extracting candidate attributes from complex engineering texts using a large language model. Secondly, through a comprehensive quantitative evaluation of structural integrity, temporal continuity, and data similarity, an objective and comprehensive attribute quality screening mechanism is constructed to ensure that the selected attributes meet high standards in terms of data structure standardization, temporal reliability, and non-redundancy. Finally, automatic decision-making and data table creation based on preset thresholds not only greatly reduce the cost and subjective error of manual review and database construction, but also enable the engineering quality and safety database to continuously, efficiently, and with high quality absorb new knowledge, realizing the dynamic, reliable, and intelligent growth of the database.

[0065] Furthermore, based on the above embodiments of the invention, the training process of the preset attribute extraction model adopts a supervised learning mechanism. The supervision signal used by the supervised learning mechanism is determined based on the engineering quality and safety score of the training sample attributes and the frequency information of the training sample attributes in the training set.

[0066] Supervised learning mechanism can refer to a machine learning training method that introduces explicit supervision signals to guide model learning, enabling the model to clearly distinguish between effective and ineffective features, thereby optimizing towards a preset goal and improving the accuracy and relevance of task execution.

[0067] Training sample attributes can refer to various attributes related to engineering quality and safety extracted from training data in the field of engineering quality and safety, such as project number, test results, hazard level, rectification status, etc., which are the core data foundation for model training.

[0068] Engineering quality and safety scoring can refer to a score used to measure the risk level and business importance of an attribute in the context of engineering quality and safety management. It can be determined by factors such as, but not limited to, the probability of attribute changes (the probability of an attribute becoming abnormal or malfunctioning) and the impact of changes (the degree of harm caused to the project by the changes).

[0069] Frequency information refers to the data on the occurrence patterns of training sample attributes in the training set. It can include the total number of occurrences, the highest number of occurrences in a single text, the average number of occurrences in different texts, etc., and is used to reflect the frequency and popularity of the attribute in engineering quality and safety scenarios.

[0070] In this embodiment of the invention, the preset attribute extraction mode can be trained using a supervised learning mechanism, and the supervised signal used can be determined by the following two parts of information: ① the engineering quality and safety score of each training sample attribute, which can be determined based on factors such as the probability of attribute change and the impact of the change; ② the frequency information of each training sample attribute in the training set, such as the total number of times the attribute appears in the training set, the highest number of times it appears in a single text, and the average number of times it appears in different texts.

[0071] Furthermore, based on the above embodiments of the invention, the process of generating the monitoring signal includes:

[0072] Obtain the security matrix corresponding to the attributes of the training samples; wherein, the security matrix includes the probability score and the impact score of the attributes of the training samples;

[0073] Based on probability scores and impact scores, determine the engineering quality and safety scores corresponding to the attributes of the training samples;

[0074] Statistical analysis of the frequency of occurrence of training sample attributes in the training set;

[0075] Based on the frequency of occurrence data, the attribute density coefficients corresponding to the attributes of the training samples are determined by a preset density formula.

[0076] The training evaluation coefficients are determined based on the engineering quality and safety score and the attribute density coefficient, and then used as a supervision signal.

[0077] Among them, the safety matrix can refer to a two-dimensional tool used for engineering quality and safety risk assessment. The horizontal axis usually covers various possible changes in attributes in the engineering scenario (such as missing data, abnormal parameters, logical conflicts, etc.), and the vertical axis contains the probability level and impact level of attribute changes. The two-dimensional assessment provides a basis for quantifying attribute risks.

[0078] The probability score can be a quantitative score obtained by experts in the field of engineering quality and safety, based on a safety matrix, by evaluating the probability of changes in the attributes of training samples (such as data errors, missing data, failures, etc.) in combination with factors such as historical data, engineering environment, and project characteristics. The higher the score, the greater the probability of changes in the attributes.

[0079] Impact score can refer to a quantitative score given by experts in the field of engineering quality and safety, based on a safety matrix, to assess the degree of impact of changes in the attributes of training samples on core dimensions such as engineering quality, construction safety, project schedule, and cost control. The higher the score, the wider the scope of impact and the greater the degree of harm caused by the attribute change.

[0080] Frequency data can refer to statistical data describing the patterns of occurrence of training sample attributes in the model training set. Specifically, it can include three core indicators: total occurrence count, highest occurrence count in a single text, and average occurrence count across different texts. These indicators are used to reflect the frequency characteristics of attribute usage in domain texts.

[0081] The attribute density coefficient can be a quantitative indicator calculated by a preset density formula. It is used to intuitively reflect the frequency and prevalence of the use of training sample attributes in training texts in the field of engineering quality and safety. The higher the frequency, the larger the coefficient.

[0082] The training evaluation coefficient can be a comprehensive quantitative indicator obtained by integrating engineering quality and safety scores and attribute density coefficients. It is a specific manifestation of the supervision signal and can simultaneously take into account the risk importance and usage frequency characteristics of attributes, providing precise guidance for model training.

[0083] In this embodiment of the invention, the process of generating the monitoring signal specifically includes:

[0084] (1) Obtain a dedicated security matrix pre-constructed for each training sample attribute. This matrix is ​​a two-dimensional evaluation tool with two key scoring dimensions: the probability score and the impact score of the training sample attribute. These are used to quantify the likelihood (i.e., the probability magnitude) of the attribute causing problems in engineering practice and the severity of the impact once a problem occurs.

[0085] (2) Based on the two core scores in the safety matrix, the probability score and the impact score are comprehensively calculated by pre-set quantitative fusion rules to obtain the engineering quality and safety score of each training sample attribute. This score directly reflects the risk level and importance level of the attribute in engineering quality and safety management.

[0086] (3) A comprehensive statistical analysis of the occurrence of training sample attributes in the entire model training set is conducted, which can specifically cover three key dimensions: the total number of occurrences of the attribute in the training set, the highest number of occurrences of the attribute in a single training text, and the average number of occurrences of the attribute in different training texts, thereby obtaining the frequency data of the occurrence of training sample attributes in the training set.

[0087] (4) Based on the statistically obtained frequency data, the data is fused using a preset density calculation formula to generate the attribute density coefficient corresponding to each training sample attribute. This coefficient is used to characterize the frequency and popularity of the attribute in the text of engineering quality and safety.

[0088] (5) The engineering quality and safety score (reflecting the importance of attribute risk) and the attribute density coefficient (reflecting the frequency of attribute use) are fused according to a preset strategy (such as a fusion model or fusion formula) to obtain the training evaluation coefficients corresponding to each training sample attribute. The training evaluation coefficients are then used directly as the supervision signal of the preset attribute extraction model supervised learning mechanism to guide the model training process.

[0089] This invention, through the fusion of the risk importance (engineering quality and safety score) and usage frequency (attribute density coefficient) of training sample attributes, generates a supervisory signal that can accurately guide the preset attribute extraction model to focus on high-value attributes in the field of engineering quality and safety. This effectively solves the problems of low accuracy and poor alignment with engineering needs when extracting attributes using traditional large language models. During training, the model can prioritize learning and identifying core attributes with high risk levels and high usage frequency, significantly improving the matching degree between the model's extracted attributes and the needs of the engineering quality and safety database. This reduces the workload of subsequent manual screening and provides a high-quality candidate attribute foundation for database expansion, indirectly improving the efficiency and reliability of database expansion and ensuring that the expanded data table meets the core requirements of engineering quality and safety management.

[0090] Example 3

[0091] Figure 3 This is a flowchart of a database expansion method provided in Embodiment 3 of the present invention. Based on the above embodiments, this embodiment provides one implementation of the database expansion method, which can automatically extract, evaluate, and filter high-value attributes conforming to database paradigms from unstructured text data in the field of engineering quality and safety, and intelligently expand the database. For example... Figure 3 As shown, the database expansion method provided in Embodiment 3 of the present invention specifically includes the following steps:

[0092] S310, Training Attribute Extraction Large Language Model.

[0093] In this embodiment of the invention, a large amount of unstructured text in the field of engineering quality and safety (such as engineering reports, inspection logs, accident records, etc.) can be collected as a training set, and the training set can be used to train the attribute extraction big language model so that the big language model can extract the attributes required by the engineering quality and safety database from specific texts.

[0094] In one embodiment, the attributes required by the engineering quality and safety database may include project number, quality inspection records, project location, etc. These attributes are used to record various information related to engineering quality and safety. The engineering quality and safety database includes multiple data tables, each composed of attributes related to engineering quality and safety data. The database satisfies the first normal form, second normal form, and third normal form to ensure the rationality, consistency, and reliability of the data structure. It should be noted that the first normal form requires each attribute in the data table to be atomic and indivisible; the second normal form requires non-primary key attributes to be fully dependent on the primary key, avoiding partial dependencies; and the third normal form requires non-primary key attributes not to depend on other non-primary key attributes, eliminating transitive dependencies.

[0095] Before training the model, the supervision signal, i.e., the training evaluation coefficient, corresponding to each training sample attribute (hereinafter referred to as attribute) can be determined. The supervision signal for the large language model can be determined by comprehensively analyzing the engineering quality and safety score of the attribute and the frequency of its occurrence. The engineering quality and safety score of the attribute is obtained by setting up a safety matrix for the attribute, representing the degree of safety of the attribute in the engineering quality and safety database. Attributes with lower safety levels are more likely to be included in the engineering quality and safety database.

[0096] The logic for obtaining engineering quality and safety scores is as follows: Define the probability of changes in different attributes, and classify the probability of changes in different attributes into probability intervals. These intervals are typically represented as: (0, 10%), (10%, 25%), (25%, 50%), (50%, 75%), (90%, 100%). Each interval represents a classification of the probability of changes in different attributes; the higher the probability, the greater the likelihood of the attribute changing. Quantitative scores are assigned to different probability intervals, ensuring that each interval corresponds to a unique score, with scores increasing progressively from the probability level within the interval. It should be noted that the probability of changes in different attributes can be judged by professionals in the relevant field. The probability assessment may be based on various factors, including the frequency of similar past events, the characteristics of the current project, and changes in the engineering environment.

[0097] The impact of changes in different attributes is defined and categorized into impact levels, typically represented as: small, relatively small, medium, large, and relatively large. Each impact level indicates a degree of influence on the attribute change; the higher the impact level, the greater the impact. A quantitative score is assigned to each impact level to ensure a unique score for each level, with scores increasing progressively from the largest impact level. It should be noted that the impact of changes in different attributes is assessed by professionals in the relevant field. The impact of a safety matrix typically represents the potential impact of a specific event on a project or engineering activity, including personnel safety, project schedule, cost, and reputation.

[0098] The likelihood and impact of changes in different attributes are scored. Following the principle of equal weighting by experts, the arithmetic mean of the scores is calculated to obtain the likelihood score and impact score for each attribute. Among these, the likelihood score... The calculation formula is: In the formula, Let I be the score given by the i-th expert on the probability of an attribute changing, and let I be the number of experts participating in the scoring. Impact Score The calculation formula is: In the formula, This represents the score given by the i-th expert regarding the impact of the attribute change.

[0099] A safety matrix is ​​established based on the probability and impact scores of changes in different attributes. Then, the engineering quality safety score for each attribute is calculated based on these scores. The expression for the safety matrix is ​​as follows: Engineering quality and safety rating The calculation formula is: .

[0100] As can be seen from the formula, the higher the engineering quality and safety score, the higher the risk level of the attribute, indicating that the attribute is more important in the engineering quality and safety database and is more likely to be used as the attribute required by the database; conversely, the lower the engineering quality and safety score, the lower the risk level of the attribute, indicating that the attribute is less important in the engineering quality and safety database.

[0101] The frequency of an attribute can be represented by an attribute density coefficient, which is obtained by counting the total number of occurrences of different specific text attributes in the training set. ; Statistically determine the highest frequency of an attribute in a single, specific text within the training set. ; Statistical analysis of the average number of times an attribute appears in different specific texts within the training set. Furthermore, the attribute density coefficient is calculated. The calculation formula is:

[0102]

[0103] As can be seen from the formula, the larger the attribute density coefficient, the higher the frequency of the attribute, which means that the attribute is likely to be frequently used in the field of engineering quality and safety, and is more likely to be the attribute needed by the engineering quality and safety database.

[0104] By comprehensively analyzing the frequency of engineering quality and safety scores and attribute occurrences, a data analysis model is established using engineering quality and safety scores and attribute density coefficients. This model generates training evaluation coefficients, which are then used as supervisory signals for the large language model. The training evaluation coefficients are... The calculation formula is:

[0105]

[0106] In the formula, and These are the proportional coefficients for the engineering quality and safety score and the attribute density coefficient, respectively. , Greater than 0.

[0107] When training a large language model using supervised learning methods, a higher training evaluation coefficient indicates a stronger supervisory signal. This means the large language model will prioritize attributes with higher training evaluation coefficients during training, and thus, when extracting attributes from specific text, it will be more inclined to extract those attributes. Therefore, a trained large language model can more accurately extract and identify attributes with higher training evaluation coefficients in subsequent applications.

[0108] This embodiment scores the attributes in specific texts related to engineering quality and safety, and combines this with a comprehensive analysis of the frequency of attribute occurrences to determine the supervision signal for the large language model. By determining the supervision signal for the large language model, the training effect of the large language model is achieved, which helps to extract attributes related to engineering quality and safety more accurately, thereby improving the model's performance.

[0109] S320: Use the attribute extraction large language model to extract multiple candidate attributes from engineering quality and safety data sources.

[0110] In this embodiment of the invention, after the large language model is trained, it can be used to extract candidate attributes that meet the requirements from engineering quality and safety data sources.

[0111] S330. Conduct compliance assessments on multiple candidate attributes and generate corresponding attribute assessment coefficients.

[0112] In this embodiment of the invention, after the large language model is trained, the attributes extracted by the large language model should meet the format of the engineering quality and safety database, and it is necessary to ensure that the extracted attribute structure is complete, the attributes are continuous in time, and that the attributes do not duplicate the attributes already stored in the engineering quality and safety database.

[0113] The reason for ensuring the integrity of the attribute structure is as follows:

[0114] (1) Data consistency: A complete attribute structure can ensure data consistency in the database. If the attribute structure is incomplete, it may lead to data storage chaos or inconsistency, making data analysis and application difficult and unreliable.

[0115] (2) Query efficiency: A complete attribute structure can improve the query efficiency of the database. If the attribute structure is incomplete, the query operation may require more time and resources to process, thus reducing the performance of the database.

[0116] (3) Data reliability: A complete attribute structure can improve the reliability and accuracy of data in the database. A missing or incomplete attribute structure may lead to data loss or errors, thereby affecting the accuracy of data analysis and decision-making.

[0117] The structural information of an attribute can be represented by a structural integrity coefficient. The logic for obtaining this structural integrity coefficient is as follows: based on the candidate attributes extracted by the large language model, the extracted attributes are incorporated into the data tables of the engineering quality and safety database according to the format of the engineering quality and safety database; the total number of attributes extracted by the large language model is obtained, and the total number of main attributes is marked as [value missing]. ; Retrieve non-primary key attributes from the attributes extracted by the large language model, and determine the number of non-primary key attributes that do not have a corresponding primary key attribute. ; Obtain the number of attributes that violate uniqueness constraints among the attributes extracted from the large language model. ; Retrieve the number of records where the foreign key and the associated table attributes are inconsistent. .

[0118] It should be noted that attributes typically include primary keys, foreign keys, and non-primary keys. A primary key is a special attribute used to uniquely identify each record in a data table after the attributes are extracted from the large language model. A foreign key is an attribute in one or more tables that forms a relationship with the primary key of another table, used to establish relationships between tables. Non-primary keys are attributes other than primary keys and foreign keys. Non-primary key attributes include all other attributes in a table, which are used to describe other information about the record, but are not used to uniquely identify the record.

[0119] The formula for calculating the primary key constraint violation ratio ZJ is: The value of ZJ ranges from 0 to 1. The smaller the ZJ, the higher the integrity of the primary key constraint, meaning that the attributes extracted by the large language model all have unique primary key values, and the data integrity is higher.

[0120] The formula for calculating the uniqueness constraint violation rate WY is: The value of WY ranges from 0 to 1. The smaller the value of WY, the better the integrity of the uniqueness constraint, that is, the attributes extracted by the large language model are not duplicated and the data integrity is higher.

[0121] The formula for calculating the foreign key constraint violation ratio EJ is: WJ represents the foreign key constraint violation rate of attributes extracted by the large language model. The value of WJ ranges from 0 to 1. The smaller the WJ, the higher the integrity of the foreign key constraint, meaning that the foreign key value extracted by the large language model is consistent with the primary key in the associated table, and the data consistency is higher.

[0122] Furthermore, the structural integrity coefficient The calculation formula is: As can be seen from the formula, the larger the structural integrity coefficient, the more the attribute structure extracted by the large language model conforms to the specifications of the engineering quality and safety database, and the more likely it is to be directly included in the engineering quality and safety database.

[0123] The reason for ensuring the temporal continuity of the attributes is as follows:

[0124] (1) Data integrity: Ensuring the temporal continuity of attributes helps ensure data integrity, that is, ensuring that no data is missing or lost. This is very important for some applications that require continuous monitoring or recording, such as sensor data, log records, etc.

[0125] (2) Data consistency: The temporal continuity of attributes also helps to maintain data consistency. For example, in some cases, it is necessary to ensure that data changes between adjacent time points do not jump or break, so as to maintain the logical continuity of the data.

[0126] Time information can be represented by a recorded fluctuation coefficient. The logic for obtaining this coefficient is as follows: obtain the attributes extracted from the large language model, and then obtain the corresponding time points. These time points represent the recording time of the attribute, such as the time point of each engineering quality inspection. The time points corresponding to the attributes are then marked as... (n=1,2,…,N, where n represents the index of the time point corresponding to the attribute, and N is a positive integer), and calculate the average time of the attribute records. .

[0127] The formula for calculating the volatility coefficient is as follows: As can be seen from this formula, the larger the fluctuation coefficient, the greater the potential time interval between the attributes acquired by the large language model, and the more likely attribute loss will occur.

[0128] The reason for ensuring the difference in attribute information between the database and the engineering quality and safety database is as follows:

[0129] (1) Save storage space: If there is a lot of redundant data in the database, it will occupy a lot of storage space. By ensuring the difference of attributes, the situation of repeatedly storing the same information can be reduced to the greatest extent, thereby saving storage resources.

[0130] (2) Improve data update efficiency: When there is redundancy in the data in the database, updating the data may require modifying multiple copies, which increases the complexity and cost of data management. By maintaining the differences in attributes, the amount of redundant data can be reduced, the data update process can be simplified, and the efficiency of data update can be improved.

[0131] (3) Improve query performance: A large amount of redundant data in the database may affect query performance because the query needs to process more data. By reducing redundant data, the complexity of the query can be reduced and the speed and efficiency of the query can be improved.

[0132] The difference information of attributes can be represented by a similarity coefficient. The logic for obtaining this similarity coefficient is as follows: obtain the attributes extracted by the large language model, and determine the similarity between the attributes extracted by the large language model and the attributes in the engineering quality and safety database by using the Jaccard similarity calculation method.

[0133] Retrieve the primary key attribute extracted from the large language model, construct a set from the non-primary key attributes associated with the primary key attribute, and mark the set of non-primary key attributes associated with the primary key attribute as... Where r = 1, 2, ..., R, and R represents the number of primary key attributes. This represents the value of the m-th non-primary key attribute associated with the r-th primary key, where m = 1, 2, ..., M, M is the number of non-primary key attributes associated with the primary key attribute, and m represents the index of the non-primary key attribute associated with the primary key attribute.

[0134] Iterate through the tables in the engineering quality and safety database to find the non-primary key attributes associated with the primary key attributes. Calculate the similarity between these non-primary key attributes and each table in the engineering quality and safety database. The similarity calculation formula is as follows: ;in, This is the similarity between the r-th primary key extracted from the large language model and the r-th primary key in the k-th data table of the engineering quality and safety database, where k = 1, 2, ..., K, K is a positive integer, and k is the table number in the engineering quality and safety database. and These represent the intersection symbol and the union symbol, respectively.

[0135] The similarity between different primary keys extracted from the large language model and different data tables in the engineering quality and safety database is obtained. The similarity scores of the different data tables are then ranked to determine the highest similarity score. .

[0136] Furthermore, the formula for calculating the similarity coefficient is: As can be seen from the formula, the larger the similarity coefficient, the more similar the attributes extracted by the large language model are to the attributes already existing in the engineering quality and safety database. If the attributes extracted by the large language model are included in the engineering quality and safety database, redundant data may occur.

[0137] By comprehensively evaluating the structural, temporal, and difference information of attributes, a data analysis model is established using the structural integrity coefficient, record fluctuation coefficient, and similarity coefficient to generate attribute evaluation coefficients. The calculation formula is: ;in, , and These are the proportional coefficients corresponding to the structural integrity coefficient, record fluctuation coefficient, and similarity coefficient, respectively. , and All are greater than 0.

[0138] S340. Based on the attribute evaluation coefficient, select the target attribute from multiple candidate attributes, and create a corresponding data table in the engineering quality and safety database based on the target attribute.

[0139] In this embodiment of the invention, an attribute evaluation coefficient threshold can be set, and then the attribute evaluation coefficients of the candidate attributes extracted by the large language model are compared with the attribute evaluation coefficient threshold. If the attribute evaluation coefficient is greater than the attribute evaluation coefficient threshold, an early warning signal is generated, indicating that the attributes extracted by the large language model do not conform to the format or requirements of the engineering quality and safety database, and that the attributes extracted by the large language model need to be further processed by professionals in the field. The large language model can also be adjusted to ensure that the attributes extracted by the large language model conform to the format or requirements of the engineering quality and safety database. If the attribute evaluation coefficient is less than the attribute evaluation coefficient threshold, no early warning signal is generated, and a separate data table is constructed for the attributes extracted by the large language model and included in the engineering quality and safety database.

[0140] This invention extracts candidate attributes from text after training using a large language model. Based on the structural, temporal, and differential information of these candidate attributes, a comprehensive analysis is performed. While meeting the format and requirements of an engineering quality and safety database, suitable attributes are selected and incorporated into the database. This helps construct a database that conforms to the format and requirements of an engineering quality and safety database, ensuring its integrity, consistency, and standardization, improving its quality and usability, and ultimately enhancing the efficiency and reliability of engineering quality and safety management.

[0141] The technical solution of this invention extracts the required attributes for the engineering quality and safety database from text using a large language model, improving accuracy and reducing the workload of manual screening. Through comprehensive analysis, it can better ensure the rationality, consistency, and reliability of the database. At the same time, it incorporates dynamic attribute adjustment into decision-making, and through automated attribute evaluation and supervised learning mechanisms, it significantly reduces the time and manpower costs required to build the engineering quality and safety database, and significantly improves the efficiency of database construction.

[0142] Example 4

[0143] Figure 4 This is a schematic diagram of the structure of a database expansion device provided in Embodiment 4 of the present invention. Figure 4 As shown, the device includes:

[0144] The attribute acquisition module 41 is used to acquire multiple candidate attributes extracted from the engineering quality and safety data source;

[0145] The attribute evaluation module 42 is used to evaluate the compliance of multiple candidate attributes and generate corresponding attribute evaluation coefficients. The compliance evaluation includes at least the following dimensions: structural integrity, temporal continuity, and similarity to attributes already in the database.

[0146] The attribute filtering module 43 is used to filter the target attribute from multiple candidate attributes based on the attribute evaluation coefficient;

[0147] The data table creation module 44 creates corresponding data tables in the engineering quality and safety database based on the target attributes.

[0148] Furthermore, based on the above embodiments of the invention, the attribute acquisition module 41 is specifically used for:

[0149] The engineering quality and safety data source is input into the preset attribute extraction model to extract candidate attributes; the preset attribute extraction model includes at least a large language model.

[0150] Furthermore, based on the above embodiments of the invention, the attribute evaluation module 42 is specifically used for:

[0151] Determine the structural integrity coefficient, record fluctuation coefficient, and similarity coefficient for each candidate attribute;

[0152] Based on the structural integrity coefficient, record fluctuation coefficient, and similarity coefficient, the attribute evaluation coefficient corresponding to each candidate attribute is determined.

[0153] Furthermore, based on the above embodiments of the invention, the structural integrity coefficient is determined based on the violation rate of primary key constraint, the violation rate of uniqueness constraint, and the violation rate of foreign key constraint of candidate attributes;

[0154] The volatility coefficient is determined based on the degree of dispersion of the candidate attribute at the corresponding time point.

[0155] The similarity coefficient is determined based on the similarity between the candidate attribute and the attribute already in the database.

[0156] Furthermore, based on the above embodiments of the invention, the attribute filtering module 43 is specifically used for:

[0157] The evaluation coefficients of each attribute are compared with preset thresholds, and candidate attributes that are less than the preset thresholds are identified as target attributes.

[0158] Furthermore, based on the above embodiments of the invention, the training process of the preset attribute extraction model adopts a supervised learning mechanism. The supervision signal used by the supervised learning mechanism is determined based on the engineering quality and safety score of the training sample attributes and the frequency information of the training sample attributes in the training set.

[0159] Furthermore, based on the above embodiments of the invention, the process of generating the monitoring signal includes:

[0160] Obtain the security matrix corresponding to the attributes of the training samples; wherein, the security matrix includes the probability score and the impact score of the attributes of the training samples;

[0161] Based on probability scores and impact scores, determine the engineering quality and safety scores corresponding to the attributes of the training samples;

[0162] Statistical analysis of the frequency of occurrence of training sample attributes in the training set;

[0163] Based on the frequency of occurrence data, the attribute density coefficients corresponding to the attributes of the training samples are determined by a preset density formula.

[0164] The training evaluation coefficients are determined based on the engineering quality and safety score and the attribute density coefficient, and then used as a supervision signal.

[0165] The database expansion device provided in the embodiments of the present invention can execute the database expansion method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0166] Example 5

[0167] Figure 5 A schematic diagram of an electronic device 50 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0168] like Figure 5 As shown, the electronic device 50 includes at least one processor 51 and a memory, such as a read-only memory (ROM) 52 and a random access memory (RAM) 53, communicatively connected to the at least one processor 51. The memory stores computer programs executable by the at least one processor. The processor 51 can perform various appropriate actions and processes based on the computer program stored in the ROM 52 or loaded from storage unit 58 into the RAM 53. The RAM 53 can also store various programs and data required for the operation of the electronic device 50. The processor 51, ROM 52, and RAM 53 are interconnected via a bus 54. An input / output (I / O) interface 55 is also connected to the bus 54.

[0169] Multiple components in electronic device 50 are connected to I / O interface 55, including: input unit 56, such as keyboard, mouse, etc.; output unit 57, such as various types of monitors, speakers, etc.; storage unit 58, such as disk, optical disk, etc.; and communication unit 59, such as network card, modem, wireless transceiver, etc. Communication unit 59 allows electronic device 50 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0170] Processor 51 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 51 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 51 performs the various methods and processes described above, such as the database expansion method.

[0171] In some embodiments, the database expansion method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 58. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 50 via ROM 52 and / or communication unit 59. When the computer program is loaded into RAM 53 and executed by processor 51, one or more steps of the database expansion method described above may be performed. Alternatively, in other embodiments, processor 51 may be configured to perform the database expansion method by any other suitable means (e.g., by means of firmware).

[0172] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0173] In some embodiments, the database extension method may be implemented as a computer program, which is implicitly included in a computer program product. When executed by a processor, the computer program implements the database extension method of the present invention. The computer program product can be understood as a software product that primarily implements its solution through a computer program. The computer program used to implement the method of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer program causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer program may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0174] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0175] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0176] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0177] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0178] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0179] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A database expansion method, characterized in that, The method includes: Retrieve multiple candidate attributes extracted from engineering quality and safety data sources; Each of the candidate attributes is evaluated for compliance, and corresponding attribute evaluation coefficients are generated. The compliance evaluation includes at least the following dimensions: structural integrity, temporal continuity, and similarity to attributes already in the database. The target attribute is selected from the plurality of candidate attributes based on the attribute evaluation coefficient; Based on the target attributes, a corresponding data table is created in the engineering quality and safety database.

2. The method according to claim 1, characterized in that, The process of obtaining multiple candidate attributes extracted from engineering quality and safety data sources includes: The engineering quality and safety data source is input into a preset attribute extraction model to extract the candidate attributes; wherein, the preset attribute extraction model includes at least a large language model.

3. The method according to claim 1, characterized in that, The step of performing compliance assessments on the multiple candidate attributes and generating corresponding attribute assessment coefficients includes: Determine the structural integrity coefficient, record fluctuation coefficient, and similarity coefficient for each of the candidate attributes; Based on the structural integrity coefficient, the record fluctuation coefficient, and the similarity coefficient, the attribute evaluation coefficient corresponding to each candidate attribute is determined.

4. The method according to claim 3, characterized in that, The structural integrity coefficient is determined based on the violation rates of primary key constraints, uniqueness constraints, and foreign key constraints of the candidate attributes. The recorded volatility coefficient is determined based on the degree of dispersion of the candidate attribute at the corresponding time point; The similarity coefficient is determined based on the similarity between the candidate attribute and the attribute already in the database.

5. The method according to claim 1, characterized in that, The process of selecting the target attribute from the plurality of candidate attributes based on the attribute evaluation coefficient includes: The evaluation coefficients of each attribute are compared with a preset threshold, and the candidate attributes that are less than the preset threshold are determined as the target attributes.

6. The method according to claim 2, characterized in that, The training process of the preset attribute extraction model adopts a supervised learning mechanism. The supervision signal used by the supervised learning mechanism is determined based on the engineering quality and safety score of the training sample attributes and the frequency information of the training sample attributes in the training set.

7. The method according to claim 6, characterized in that, The process of generating the monitoring signal includes: Obtain the security matrix corresponding to the training sample attributes; wherein, the security matrix includes the probability score and the impact score of the training sample attributes; Based on the probability score and the impact score, the engineering quality and safety score corresponding to the training sample attribute is determined; Statistically analyze the frequency of occurrence of the training sample attributes in the training set; Based on the occurrence frequency data, the attribute density coefficients corresponding to the training sample attributes are determined by a preset density formula. The training evaluation coefficients are determined based on the engineering quality and safety score and the attribute density coefficients, and the training evaluation coefficients are used as the supervision signal.

8. A database expansion device, characterized in that, The device includes: The attribute retrieval module is used to retrieve multiple candidate attributes extracted from the engineering quality and safety data source; The attribute evaluation module is used to evaluate the compliance of the multiple candidate attributes and generate corresponding attribute evaluation coefficients; wherein the compliance evaluation includes at least the following dimensions: structural integrity, temporal continuity, and similarity with attributes already in the database; An attribute filtering module is used to filter out target attributes from a plurality of candidate attributes based on the attribute evaluation coefficient. The data table creation module creates a corresponding data table in the engineering quality and safety database based on the target attributes.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the database expansion method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the database expansion method according to any one of claims 1-7.

11. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the database expansion method according to any one of claims 1-7.

Images