Detection table data storage method and device, electronic equipment and computer readable medium

Abstract

Embodiments of the present disclosure disclose a detection table data storage method and device, electronic equipment and computer readable medium. A specific embodiment of the method comprises: obtaining a detection table data set; performing table structure auditing on a detection table structure information set according to preset detection template requirement information to obtain an auditing result set; performing knowledge graph construction on at least one detection table data to obtain a detection knowledge graph; generating a detection table template set; obtaining a comment text information set; performing requirement identification on the comment text information set and dynamically adjusting the detection table template set to obtain an adjusted detection table template set; performing data quality detection on a field attribute value group set to obtain a detection result set; storing at least one detection table data representing a detection pass to a preset database, and setting access permissions of user identity information. The embodiment can reduce redundant and erroneous detection table data, reduce waste of storage resources, improve user experience and data security.

Description

Technical Field

[0001] Embodiments of this disclosure relate to the field of computer technology, and more specifically to a method, apparatus, electronic device, and computer-readable medium for storing data in a detection table. Background Technology

[0002] As the volume of data in the audit tables increases, the cost of auditing and maintaining this data also rises, making the auditing and maintenance of intermediate tables increasingly important. The typical method for storing audit table data is as follows: audit table templates are generated by recommending and modifying audit table datasets based on subjectively defined template rules; then, auditing and permission settings are applied to the audit table datasets corresponding to these templates.

[0003] However, the inventors discovered that when using the above method to store test table data, the following technical problems often arise:

[0004] First, the detection template is determined by manually set detection template rules. Since the manually set detection template rules have certain subjectivity and limitations and cannot be updated in real time, the accuracy of the obtained detection template is low. As a result, the detection template dataset has a lot of redundant data, which leads to a waste of storage resources and a poor user experience.

[0005] Second, the detection table template is obtained by recommending and modifying the detection table dataset. However, since the recommendation and modification are based solely on comment text information, only the semantic extraction of text information is considered, ignoring the impact of user behavior information on comment text information. Furthermore, only word-segmentation level semantic extraction is performed, resulting in over-correction of the extracted contextual semantic information. This leads to low accuracy in anomaly detection. The resulting detection table template also contains redundant or incomplete data, resulting in low accuracy and a poor user experience.

[0006] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0008] Some embodiments of this disclosure provide detection table data storage methods, apparatuses, electronic devices, and computer-readable media to address one or more of the technical problems mentioned in the background section above.

[0009] In a first aspect, some embodiments of this disclosure provide a method for storing detection table data, including: acquiring a detection table dataset from different data sources, wherein the detection table data in the dataset includes: detection table structure information and field attribute value groups corresponding to the detection table structure information; performing table structure audit on the detection table structure information set according to preset detection template requirements to obtain an audit result set; constructing a knowledge graph on at least one detection table data corresponding to at least one audit result in the audit result set that represents the audit passed to obtain a detection knowledge graph; generating a detection table template set based on the detection knowledge graph and the different data sources; and acquiring a detection table template set for the above-mentioned detection... The process involves: 1) setting a set of comment text information for a set of test templates; 2) identifying the requirements of this set of comment text information to obtain a set of requirements information; 3) dynamically adjusting each test template in the set of test templates based on these requirements information to obtain an adjusted set of test templates; 4) performing data quality checks on the set of field attribute values ​​included in each test table data corresponding to the adjusted set of test templates to obtain a set of test results; 5) storing at least one test table data corresponding to at least one test result that passed the test in the set of test results into a preset database, and 6) setting access permissions for the user identity information corresponding to the access request information to the at least one test table data that passed the test.

[0010] Secondly, some embodiments of this disclosure provide a detection table data storage device, comprising: a first acquisition unit configured to acquire detection table datasets from different data sources, wherein the detection table data in the aforementioned detection table datasets includes: detection table structure information and field attribute value groups corresponding to the detection table structure information; a table structure review unit configured to review the detection table structure information set according to preset detection template requirement information to obtain a review result set; a knowledge graph construction unit configured to construct a knowledge graph on at least one detection table data corresponding to at least one review result representing a passed review in the aforementioned review result set to obtain a detection knowledge graph; a generation unit configured to generate a detection table template set based on the aforementioned detection knowledge graph and the aforementioned different data sources; and a second acquisition unit. The system is configured to: acquire a set of comment text information for the aforementioned test table template set; a demand identification unit configured to identify the demand information from the aforementioned comment text information set, and dynamically adjust each test table template in the aforementioned test table template set according to the aforementioned demand information set, to obtain an adjusted test table template set; a data quality detection unit configured to perform data quality detection on the set of field attribute values ​​included in each test table data corresponding to the aforementioned adjusted test table template set, to obtain a detection result set; and a storage unit configured to store at least one test table data corresponding to at least one test result in the aforementioned detection result set that represents a passed test into a preset database, and to set access permissions for the user identity information corresponding to the access request information to the at least one test table data that represents a passed test.

[0011] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, such that when the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any implementation of the first aspect.

[0012] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method as described in any implementation of the first aspect.

[0013] The various embodiments of this disclosure have the following beneficial effects: the detection table data storage method of some embodiments of this disclosure can reduce redundant and erroneous detection table data, reduce the waste of storage resources, improve user experience, and enhance data security. Specifically, the waste of related storage resources is caused by: determining the detection template through subjectively set detection table template rules. Since subjectively set detection table template rules have certain subjectivity and limitations and cannot be updated in real time, the accuracy of the obtained detection table template is low. Consequently, the detection table dataset corresponding to the detection template contains a large amount of redundant data, resulting in wasted storage resources and a poor user experience. Based on this, the detection table data storage method of some embodiments of this disclosure can first obtain detection table datasets from different data sources. The detection table data in the aforementioned detection table datasets includes: detection table structure information and field attribute value groups corresponding to the detection table structure information. Here, the detection table datasets from different data sources are used for subsequent generation of detection knowledge graphs. Secondly, according to preset detection template requirement information, the detection table structure information set is audited to obtain an audit result set. Here, auditing the detection table structure information set according to preset detection template requirement information can reduce the computational load of subsequent knowledge graph generation and reduce the waste of computational resources. Next, a knowledge graph is constructed from the data of at least one detection table corresponding to at least one approved review result in the aforementioned review result set, resulting in a detection knowledge graph. Here, knowledge graph construction can more accurately represent the relationships between detection table datasets and display detection table datasets from different sources, facilitating the improvement of the accuracy of subsequently generated detection table templates. Then, based on the aforementioned detection knowledge graph and the different data sources, a set of detection table templates is generated. This improves the accuracy of the obtained detection table template set, accurately distinguishing detection templates from different data sources, thereby reducing a large amount of redundant data. Subsequently, a set of comment text information for the aforementioned detection table template set is obtained. Here, the comment text information set is used for subsequent dynamic adjustments to the detection table template set. Afterwards, the aforementioned comment text information set is used for demand identification, resulting in a demand information set. Based on this demand information set, each detection table template in the aforementioned detection table template set is dynamically adjusted, resulting in an adjusted detection table template set. Here, by dynamically adjusting the detection table template set through the demand information set, the accuracy of the detection table templates is ensured, and they are more adaptable to user needs, improving the user experience. Then, data quality checks are performed on the set of field attribute values ​​included in each test table corresponding to the adjusted test table template set, resulting in a test result set. Here, data quality checks can improve the data quality of the set of field attribute values, reduce erroneous data, and thus reduce the waste of storage resources.Finally, the data of at least one detection table corresponding to at least one detection result that passed the above-mentioned detection results set is stored in a preset database, and access permissions for the at least one detection table data that passed the detection are set according to the user identity information corresponding to the access request information. This reduces the waste of storage resources and improves the security of the detection table data, thus reducing data leakage to some extent. Therefore, this detection table data storage method, by performing structural detection on the detection table dataset, generating a detection knowledge graph from at least one detection table data that passed the structural detection, constructing a detection table template set, and reviewing and setting user permissions for field attribute value sets, can reduce redundant and erroneous detection table data, reduce the waste of storage resources, improve user experience, and enhance data security. Attached Figure Description

[0014] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0015] Figure 1 This is a flowchart of some embodiments of the test table data storage method according to this disclosure;

[0016] Figure 2 This is a schematic diagram of the structure of some embodiments of the test table data storage device according to the present disclosure;

[0017] Figure 3 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0018] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0019] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0020] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0021] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0022] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0023] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] Figure 1 A flow 100 of some embodiments of a test table data storage method according to the present disclosure is shown. The test table data storage method includes the following steps:

[0025] Step 101: Obtain the detection table dataset from different data sources.

[0026] In some embodiments, the execution entity (e.g., an electronic device) of the above-described detection table data storage method can acquire detection table datasets from different data sources via wired or wireless connections. The detection table data in the dataset includes: detection table structure information and corresponding field attribute value groups. The different data sources can be audit databases from different departments. The detection table data in the dataset can be audit data to be tested displayed in the form of a data table. For example, the detection table data can be audit intermediate table data. The audit intermediate table can be audit data extracted from databases of different data sources in the form of a database table. The detection table structure information can be the structure information of the detection table. The structure information can include, but is not limited to, at least one of the following: table name, table fields, and constraints on the table fields. For example, the constraints on the table fields can be the data type of the fields.

[0027] Step 102: Based on the preset test template requirements, review the test table structure information set to obtain the review result set.

[0028] In some embodiments, the aforementioned execution entity can perform table structure review on the detection table structure information set based on preset detection template requirement information to obtain a review result set. The preset detection template requirement information can be rule information used to detect whether the aforementioned detection table dataset conforms to the detection template. The review results in the review result set include: review results indicating failure and review results indicating success.

[0029] As an example, the aforementioned executing entity may perform the following review steps for each piece of test table structure information in the aforementioned test table structure information set: review each piece of the aforementioned test table structure information against the set of requirements included in the aforementioned preset test template requirement information to obtain a set of review results. The aforementioned set of requirements information may include at least one of the following: naming requirements for the table name and fields of the test table data; uniqueness requirements for the primary key settings of the test table data; integrity requirements for the field attribute values ​​included in the test table data; and permission settings for the test table data. In response to determining that there is a review result indicating a failure in the aforementioned set of review results, the failure is determined as the review result for the test table structure information.

[0030] In some optional implementations of certain embodiments, the above-mentioned process of auditing the structure of the detection table information set based on preset detection template requirements to obtain an audit result set may include the following steps:

[0031] The first step involves intelligent requirement construction based on the aforementioned preset detection template requirement information, resulting in a template contract code information set. The template contract code information in this set can be in the form of code representing the preset detection template requirement information.

[0032] As an example, the aforementioned execution entity can first extract semantic elements and attributes from the preset detection template requirement information to obtain a semantic triple information set. The semantic elements can be the requirement entity information set within the preset detection template requirement information. The requirement entity information in the requirement entity information set can be keyword information related to the requirement. The attributes can be the relationships between the requirement entity information included in the requirement entity information set and the attribute information of the requirement entity information. Then, using a recursive abstract syntax tree (AST), a mapping relationship between the semantic triple information set and the template contract code information set is generated. Finally, through the mapping relationship, the semantic triple information set is constructed into the template contract code information set.

[0033] The second step involves performing vulnerability detection on each template contract code in the aforementioned template contract code information set to generate vulnerability detection results, resulting in a vulnerability detection result set. These results can include both vulnerability detection results indicating that the vulnerability detection failed and results indicating that the vulnerability detection passed.

[0034] As an example, the aforementioned execution entity can utilize a contract vulnerability detection model to perform vulnerability detection on each template contract code piece in the aforementioned template contract code information set, thereby generating vulnerability detection results and obtaining a vulnerability detection result set. The aforementioned contract vulnerability detection model can convert the Solidity (a contract-based programming language) source code corresponding to the template contract code information set into the form of contract snippets to capture semantic information and control flow dependency information within the template contract code information set. The aforementioned contract vulnerability detection model may include: a bidirectional long short-term memory network model and an attention mechanism.

[0035] The third step is to remove the template contract code information corresponding to at least one vulnerability detection result that indicates a failed vulnerability detection from the above-mentioned vulnerability detection result set, thereby obtaining the code information set after removal.

[0036] Fourth, for each test table structure in the above test table structure information set, perform the following table structure review steps:

[0037] The first sub-step involves performing a structure check on the aforementioned detection table structure information based on the removed code information set, resulting in a structure check result set. This set includes structure check results that indicate successful structure check and results that indicate failed structure check. A successful structure check indicates that the aforementioned detection table structure information meets the requirements of the removed code information. A failed structure check indicates that the aforementioned detection table structure information does not meet the requirements of the removed code information.

[0038] As an example, the aforementioned execution entity can input the aforementioned detection table structure information into the aforementioned removed code information set to obtain a structure detection result set.

[0039] The second sub-step is to determine, in response to the determination that there are structural detection results in the above structural detection result set that the structural detection failed, that the structural detection failed is determined as the review result of the above detection table structural information.

[0040] The third sub-step is to determine, in response to the determination that there are no structural detection results in the above structural detection result set that the structural detection has failed, that the structural detection has passed is determined as the review result of the above detection table structural information.

[0041] Step 103: Construct a knowledge graph for at least one detection table data corresponding to at least one audit result that represents the audit passed, and obtain the detection knowledge graph.

[0042] In some embodiments, the aforementioned executing entity may construct a knowledge graph from the at least one detection table data corresponding to at least one approved audit result in the aforementioned audit result set, thereby obtaining a detection knowledge graph. The detection knowledge graph may be a semantic knowledge base representing the set of detection table structure information, the set of field attribute values, and the relationships between the at least one detection table data.

[0043] As an example, the aforementioned executing entity can first convert the data from at least one detection table corresponding to at least one approved audit result in the audit result set into data in a fixed format, obtaining a converted detection dataset. The predetermined format can conform to the format used to construct a knowledge graph. For example, the predetermined format could be RDF (Resource Description Framework). Then, the converted detection dataset is stored in a graph database to obtain the detection knowledge graph.

[0044] In some optional implementations of certain embodiments, constructing a knowledge graph from at least one detection table data corresponding to at least one approved audit result in the audit result set to obtain a detection knowledge graph may include the following steps:

[0045] The first step is to preprocess the data from at least one of the above-mentioned test tables to obtain a preprocessed test table dataset. This preprocessing may include, but is not limited to, at least one of the following: filling in missing values, deduplication, and smoothing noisy data.

[0046] The second step is to determine the set of relationships between the preprocessed detection table data included in the preprocessed detection table dataset. These relationships can be primary and foreign key constraints between the preprocessed detection table data. In practice, the executing entity can determine this set of relationships by querying primary and foreign key constraints using a query language, or by querying a database table that records the relationships between the detection table data included in the detection table dataset.

[0047] The third step involves generating ontology mapping text information based on the preprocessed detection table dataset and the set of associations. This ontology mapping text information can represent the mapping relationship between the preprocessed detection table dataset and the ontology layer in the detection knowledge graph. It may include: a set of preprocessed detection table names, a set of fields, a set of attribute values ​​corresponding to the field sets, and a set of associations.

[0048] As an example, the aforementioned execution entity can utilize a pattern matching algorithm to generate ontology mapping text information based on the preprocessed detection table dataset and the set of association relationships.

[0049] The fourth step is to generate an initial detection knowledge graph based on the aforementioned ontology-mapped text information. This initial detection knowledge graph can be obtained by mapping at least one detection table data to the ontology layer of the knowledge graph.

[0050] As an example, the aforementioned execution entity can use the mapping relationship in the aforementioned ontology mapping file information to map at least one detection table data to the ontology layer in the knowledge graph, thereby obtaining an initial detection knowledge graph.

[0051] The fifth step involves entity alignment of the initial detection knowledge graph to obtain the fused knowledge graph. This fused knowledge graph can be obtained by fusing the entity information sets included in the initial detection knowledge graph.

[0052] As an example, the aforementioned execution entity can first determine the similarity value set of the entity information set included in the initial detection knowledge graph. The similarity values ​​in this set can be the similarity values ​​of any two entity information pieces in the entity information set. Then, in response to determining that there are similarity values ​​in the set greater than or equal to a preset similarity threshold, the entity information corresponding to similarity values ​​greater than or equal to the preset similarity threshold is fused to obtain the fused knowledge graph. The preset similarity threshold can be a minimum value used to determine whether to fuse entity information. For example, the preset similarity threshold could be 0.75.

[0053] The sixth step involves performing knowledge reasoning on the fused knowledge graph to obtain a reasoned knowledge graph, which serves as the detection knowledge graph. This reasoned knowledge graph can be derived by reasoning new entity relationship information from existing entity relationship information in the fused knowledge graph and then adding this new information to the fused knowledge graph.

[0054] As an example, the aforementioned execution entity can utilize PRA (Path Ranking Algorithm) to perform knowledge reasoning on the fused knowledge graph, thereby expanding the fused knowledge graph to obtain an expanded entity relationship information set. Then, this expanded entity relationship information set is added to the fused knowledge graph to obtain the detection knowledge graph.

[0055] Step 104: Generate a set of detection table templates based on the detection knowledge graph and different data sources.

[0056] In some embodiments, the execution entity can generate a set of detection table templates based on the detection knowledge graph and the different data sources. The detection table templates in this set can be templates specific to a single data source that meet preset detection template requirements.

[0057] As an example, the aforementioned execution entity can first determine the portions of the detection knowledge graph corresponding to different data sources, obtaining a local detection knowledge graph set. Then, it can determine the entity information groups and relation groups included in each local detection knowledge graph within the aforementioned local detection knowledge graph set, obtaining an entity information group set and a relation group set. Finally, it can construct each entity information group in the entity information group set and the relation group set corresponding to the entity information group in the relation group set in the form of a data table to generate a detection table template set.

[0058] Step 105: Obtain the set of comment text information for the detection table template set.

[0059] In some embodiments, the executing entity may obtain a set of comment text information for the aforementioned test table template set. The comment text information in this set may be text information posted by users regarding the aforementioned test table template.

[0060] Step 106: Identify the requirements of the comment text information set to obtain the requirement information set, and dynamically adjust each test table template in the test table template set according to the requirement information set to obtain the adjusted test table template set.

[0061] In some embodiments, the execution entity may perform requirement identification on the aforementioned comment text information set to obtain a requirement information set, and dynamically adjust each detection table template in the aforementioned detection table template set according to the aforementioned requirement information set to obtain an adjusted detection table template set. The requirement information in the aforementioned requirement information set can characterize the text information in which users comment on the structural information of the detection templates. The aforementioned dynamic adjustment may include, but is not limited to, at least one of the following: adding, deleting, or modifying.

[0062] As an example, the aforementioned execution entity can first perform data cleaning on the aforementioned comment text information set to obtain a cleaned comment text information set. Secondly, using a sentiment segmentation dictionary, sentiment analysis is performed on each cleaned comment text information in the aforementioned cleaned comment text information set to generate a comment sentiment tendency set. This comment sentiment tendency can represent the user's sentiment information towards the detection table template. The comment sentiment tendency can include: sentiment tendencies representing positive emotions and sentiment tendencies representing negative emotions. The aforementioned sentiment segmentation dictionary can be a dictionary storing word sets corresponding to different sentiment tendencies. Subsequently, using the aforementioned comment sentiment tendency set, the aforementioned comment text information set is classified to obtain a comment category text information set. Then, using the LDA (Latent Dirichlet Allocation) model, topic extraction is performed on each comment category text information set in the aforementioned comment category text information set to obtain a demand information set. Finally, using the aforementioned demand information set, the field set included in each detection table template in the aforementioned detection table template set is adjusted to obtain an adjusted detection table template set.

[0063] In some optional implementations of certain embodiments, the above-mentioned requirement identification of the comment text information set to obtain a requirement information set, and the dynamic adjustment of each detection table template in the detection table template set according to the requirement information set to obtain an adjusted detection table template set, may include the following steps:

[0064] The first step is to perform anomaly detection on the aforementioned set of comment text information to obtain an abnormal comment text information set. The abnormal comment text information in this set can be text information containing elements that do not conform to the quality of the detection table template. For example, the abnormal comment text information may include, but is not limited to, at least one of the following: abnormal comment content and abnormal user behavior.

[0065] As an example, the aforementioned execution entity can first input the aforementioned set of comment text information into a comment recognition model to obtain a set of text semantic feature vectors. This comment recognition model can be a model that identifies abnormal comment information within the comment text information set. Such models can include TextCNN (Convolutional Neural Networks for Sentence Classification) and BERT (Bidirectional Encoder Representation from Transformers). Next, the user information set included in the comment text information set is one-hot encoded to obtain a user feature vector set. Then, the aforementioned text semantic feature vector set and the aforementioned user feature vector set are concatenated to obtain a concatenated feature vector set. Next, the image set included in the aforementioned comment text information set is input into a residual network to obtain a visual feature vector set. Then, the concatenated feature vector set and the aforementioned visual feature vector set are fused using multimodal methods to obtain a fused feature vector set. Finally, the fused feature vector set is input into a fully connected layer to obtain the abnormal comment text information set.

[0066] The second step is to remove the abnormal comment text information from the above comment text information set to obtain the removed comment text information set, which is used as the target comment text information set.

[0067] Third, for each target comment text in the above target comment text information set, perform the following feature word generation steps:

[0068] The first sub-step involves segmenting the target comment text information to obtain a segmented character sequence. The first segmented character in this sequence can be a marker character indicating the beginning of a sentence, and the remaining segmented character sequences can be character sequences corresponding to the target comment text information.

[0069] The second sub-step involves performing word embedding processing on the segmented character sequence to obtain a word embedding feature vector sequence. The word embedding feature vectors in this sequence represent the character's feature information. This word embedding processing can be performed using the BERT model.

[0070] The third sub-step involves weighted summation of the aforementioned character embedding feature vector sequence using a multi-head attention mechanism layer to obtain character embedding feature vectors with different weight values, which serve as the character embedding weight feature vector sequence. The self-attention mechanism in the multi-head self-attention (MHSA) layer can be a scaled dot product attention (SDA) mechanism. This multi-head attention mechanism layer can be a 12-head attention mechanism layer. The character embedding weight feature vectors in the aforementioned character embedding weight feature vector sequence can be feature vectors that incorporate contextual semantic information and different attention weights.

[0071] The fourth sub-step involves performing text weighting on the aforementioned character embedding weight feature vector sequence to obtain a text weight feature vector sequence. In practice, the execution entity can input the aforementioned character embedding weight feature vector sequence into the self-attention mechanism layer to obtain the text weight feature vector sequence.

[0072] The fifth sub-step involves aggregating the aforementioned text weight feature vector sequence and the aforementioned word embedding weight feature vector sequence to obtain an aggregated feature vector sequence. This aggregation process can be an XOR operation.

[0073] The sixth sub-step involves performing feature compression on the aggregated feature vector sequence to obtain a compressed feature vector sequence.

[0074] The seventh sub-step involves classifying the compressed feature vector sequence to obtain detection feature word groups for the target comment text information. The detection feature words in these groups are terms used to describe the structural information of the detection table template. For example, if the target comment text information could be "table name and field name do not match, fields are disordered," then the detection feature words could be the table name and field names. In practice, the executing entity can input the compressed feature vector sequence into a linear classifier to obtain the detection feature word groups for the target comment text information. The linear classifier can be an LR (Logistic Regression Classifier) ​​classifier.

[0075] The fourth step involves performing sentiment recognition on each detected feature word in the obtained set of detected feature words to generate a sentiment tendency set. This sentiment tendency set can include: expressions representing positive sentiment tendencies, neutral sentiment tendencies, and negative sentiment tendencies. These sentiment tendencies can represent the user's feelings towards the detection table template.

[0076] As an example, the aforementioned execution entity can use VADER (Valence-Aware Dictionary and Threaction Reasoner, a social network text sentiment analysis library) to perform sentiment recognition on the obtained set of detected feature words, thereby obtaining the sentiment tendency of each detected feature word in the set of detected feature words.

[0077] The fifth step involves identifying at least one target comment text information corresponding to a negative sentiment tendency, thus obtaining a set of negative comment text information. Based on this set, each test table template in the aforementioned test table template set is dynamically adjusted to obtain an adjusted set of test table templates. This dynamic adjustment can involve adding, deleting, or modifying fields in the test table templates.

[0078] As an example, the aforementioned execution entity can adjust the field set included in each test table template in the aforementioned test table template set using the aforementioned set of requirement information, thereby obtaining the adjusted test table template set.

[0079] In some optional implementations of certain embodiments, the above-described text weighting process for the word embedding weight feature vector sequence to obtain the text weight feature vector sequence may include the following steps:

[0080] The first step is to perform the following weight determination steps for each character embedding weight feature vector in the above character embedding weight feature vector sequence:

[0081] The first sub-step involves determining the character embedding window corresponding to the aforementioned character embedding weight feature vector. This character embedding window is centered on the aforementioned character embedding weight feature vector and has a window size equal to a preset text distance threshold. This preset text distance threshold can be the text distance to one side of the aforementioned character embedding weight vector. The preset text distance threshold is not limited here; its specific value can be determined based on actual circumstances.

[0082] The second sub-step involves determining the first text distance value between each character embedding weight feature vector included in the aforementioned character embedding window and the aforementioned character embedding weight feature vector, thereby obtaining a sequence of first text distance values. The aforementioned first text distance value can be 1.

[0083] The third sub-step involves determining the second text distance value between each remaining character embedding weight feature vector in the remaining character embedding weight feature vector sequence and the aforementioned character embedding weight feature vector, thus obtaining a second text distance value sequence. The remaining character embedding weight feature vector sequence can be obtained by removing each character embedding weight feature vector included in the character embedding window from the aforementioned character embedding weight feature vector sequence. The second text distance value can be obtained through the following steps: First, determine the absolute value of the difference between the position information of the remaining character embedding weight feature vector and the position information of the character embedding weight feature vector. Second, determine the first value by summing the absolute value with a first preset threshold. The first preset threshold can be 2. Then, determine the second value by taking the logarithm of the first value with a second preset threshold as the base. The second preset threshold can be 2. Finally, determine the second text distance value by the ratio of a third preset threshold to the second value. The third preset threshold can be 1.

[0084] The fourth sub-step involves determining the weight values ​​of the word embedding weight feature vector based on the first text distance value sequence and the second text distance value sequence.

[0085] As an example, the aforementioned execution entity can first determine the sum of exponents, with base e and each first text distance value in the first text distance value sequence as the exponent, as the first exponent sum. Then, it can determine the sum of exponents, with base e and each first text distance value in the first text distance value sequence and each second text distance value in the second text distance value sequence as the exponent, as the second exponent sum. Finally, the ratio of the first exponent sum to the second exponent sum is determined as the weight value of the aforementioned word embedding weight feature vector.

[0086] The fifth sub-step involves weighting the aforementioned weight values ​​and the aforementioned word embedding weight feature vectors to obtain the text weight feature vector.

[0087] In some optional implementations of certain embodiments, the above-mentioned fake text detection of the comment text information set to obtain a fake comment text information set may include the following steps:

[0088] The first step is to perform word segmentation on the above set of comment text information to obtain a set of comment terms.

[0089] The second step involves performing word embedding processing on the aforementioned comment word set to obtain a comment word embedding vector set. This vector set can be a feature vector representing the characteristic information of the comment word set. In practice, the execution entity can use CBOW (Continuous Bag of Words) to perform word embedding processing on the comment word set to obtain the comment word embedding vector set.

[0090] The third step is to input the above-mentioned comment word embedding vector set into the first bidirectional gated loop unit to obtain the comment word embedding vector set containing the semantic information above and below, which serves as the comment word semantic vector set.

[0091] The fourth step involves inputting the aforementioned semantic vector set of comment words into the first self-attention mechanism layer to obtain the comment statement vector set. The comment statement vectors in this set can represent the feature information of the comment statements included in the comment text.

[0092] The fifth step is to input the above comment statement vector set into the second bidirectional gated loop unit to obtain the comment statement embedding vector set containing contextual semantic information, which serves as the comment statement semantic vector set.

[0093] The sixth step involves inputting the aforementioned semantic vector set of comment statements into the second self-attention mechanism layer to obtain a set of comment text vectors, which serves as the comment semantic feature vector set. The comment semantic feature vectors in this set can represent the feature information of the comment text.

[0094] Step 7: Obtain the user information set corresponding to the aforementioned comment text information set. The user information set may contain user-related information. This user information set may include, but is not limited to, at least one of the following: user account information, user account level information, and user account registration time.

[0095] Step 8: Based on the aforementioned user information set, determine the user comment information set. The user comment information in this set can be information related to the comment text. This information may include, but is not limited to, at least one of the following: comment text length, the corresponding detection table template, comment time, the rating value of the comment text, and comment support level. The comment support level can be the number of likes given by other users to a single comment text.

[0096] As an example, the aforementioned executing entity can determine the user comment information group for each user account in the user account information group included in the aforementioned user information set, thereby obtaining the comment user information group set as the user comment information set.

[0097] The ninth step involves extracting features from the aforementioned user information set and user comment information set to obtain a user behavior feature vector set. The user behavior feature vectors in this set can represent the characteristic information of the user comment information and the user information itself. In practice, the executing entity can utilize a convolutional neural network to extract features from the aforementioned user information set and user comment information set to obtain the user behavior feature vector set.

[0098] Step 10: Perform a weighted summation based on meta-paths on the aforementioned comment semantic feature vector set and the aforementioned user behavior feature vector set to obtain a weighted feature vector set. The weighted feature vectors in this set can be weighted feature vectors that include both node-level weight values ​​and semantic-level weight values. The node-level weight values ​​can be the weight values ​​of different comment semantic feature vector sets obtained from the same user behavior association to the same comment semantic feature vector. The semantic-level weights can be the weight values ​​of the association paths connecting comment semantic feature vector sets through different user behavior associations. User behavior associations can include, but are not limited to, at least one of the following: identical user account information, identical comment time, and identical detection table templates for the comments.

[0099] As an example, the aforementioned execution entity can perform the following weighted summation steps for each comment semantic feature vector in the aforementioned comment semantic feature vector set: First, determine the set of comment semantic feature vectors located on the semantic meta-path of the aforementioned comment semantic feature vectors, as the semantic path feature vector set. Here, the aforementioned semantic path can be a path constructed through user behavior relationships. Second, concatenate the aforementioned comment semantic feature vectors with each semantic path feature vector in the semantic path feature vector set and input the concatenation to the self-attention mechanism layer to obtain feature weight values. Third, determine the weight coefficients of the aforementioned comment semantic feature vectors by taking the exponent of the feature weight values ​​as base e and the sum of the exponents of the feature weight values ​​included in the aforementioned feature weight value set as base e, thus obtaining a weight coefficient set. Next, perform a positional weighted summation on the aforementioned weight coefficient set and the aforementioned semantic path feature vector set to obtain the target semantic path feature vector. Subsequently, determine the comment meta-path set corresponding to the aforementioned comment semantic feature vectors. The comment meta-paths in the aforementioned comment meta-path set can be paths that start with the aforementioned comment semantic feature vectors, connect to at least one user behavior feature vector, and end with other comment semantic feature vectors. Next, the aforementioned target semantic path feature vectors and the aforementioned comment meta-path set are input into a semantic attention mechanism layer to obtain a set of meta-path weight values. The aforementioned semantic attention mechanism layer can be a neural network layer that learns the weight values ​​of each comment meta-path in the comment meta-path set. Then, the weight values ​​of each meta-path in the meta-path weight value set are normalized to obtain a set of meta-path weight coefficients. Finally, the aforementioned meta-path weight coefficient set and the aforementioned target semantic path feature vectors are summed bitwise to obtain a weight feature vector.

[0100] The eleventh step involves classifying and predicting the aforementioned weighted feature vector set to obtain a set of virtual comment text information. In practice, the executing entity can input the aforementioned weighted feature vector set into an MLP (Multilayer Perceptron) to obtain a set of fake comment text information.

[0101] The above-mentioned steps one through eleven and related content, combined with step 106, serve as an inventive point of this disclosure's embodiment, solving the second technical problem mentioned in the background art: "By recommending and modifying the detection table dataset to obtain a detection table template, since the recommendation and modification only considers the semantic extraction of text information and ignores the influence of user behavior information on the comment text information, and only performs word-segmentation-level semantic extraction, the extracted contextual semantic information is over-corrected, resulting in low anomaly detection accuracy. The obtained detection table template contains redundant or incomplete data, leading to low accuracy and a poor user experience." The factors leading to low accuracy and a poor user experience in the detection table template are often as follows: By recommending and modifying the detection table dataset to obtain a detection table template, since the recommendation and modification only considers the semantic extraction of text information and ignores the influence of user behavior information on the comment text information, and only performs word-segmentation-level semantic extraction, the extracted contextual semantic information is over-corrected, resulting in low anomaly detection accuracy and redundant or incomplete data in the obtained detection table template. Solving the above factors can improve the accuracy of the detection template and the user experience. To achieve this, this disclosure first extracts sentence-level and text-level feature vectors from the comment text information set. This allows for finer-grained capture of abnormal words and sentences in abnormal comment text information, improving the accuracy of semantic feature extraction. Then, non-semantic features of user behavior information related to the comment text information are extracted. This facilitates the subsequent construction of meta-paths using non-semantic features, determining the node-level and path-level weight values ​​for each comment text information. Finally, meta-paths are constructed using the extracted user behavior feature vectors to connect different comment semantic feature vectors. A dual-layer attention mechanism at the node and semantic levels is then used to obtain node-level and semantic-level weight values. This reduces computational resource waste and allows for more comprehensive extraction of feature information from the comment text information, improving the accuracy of anomaly detection. Consequently, the accuracy of the detection template is improved, resulting in a more precise detection template and ultimately enhancing the user experience.

[0102] Step 107: Perform data quality checks on the set of field attribute values ​​included in each test table data corresponding to the adjusted test table template set to obtain the test result set.

[0103] In some embodiments, the aforementioned execution entity can perform data quality checks on the set of field attribute values ​​included in the data of each test table corresponding to the adjusted test table template set, and obtain a test result set. The aforementioned data quality check can be based on preset detection rules to check the accuracy, completeness, consistency, relevance, and validity of the data. The aforementioned data completeness can be checking whether field attribute values ​​have missing values, duplicate values, or outliers. The aforementioned data accuracy can be checking whether field attribute values ​​meet field constraints. The aforementioned data consistency can be checking whether the same data is consistent across different data sources. The aforementioned data relevance can be the primary and foreign key constraints between fields corresponding to field attribute values.

[0104] Step 108: Store at least one test table data corresponding to at least one test result that has passed the test in a preset database, and set access permissions for the user identity information corresponding to the access request information to the at least one test table data that has passed the test.

[0105] In some embodiments, the executing entity may store at least one detection table data corresponding to at least one detection result that represents a passed detection in the aforementioned detection result set to a preset database, and set access permissions for the user identity information corresponding to the access request information to the at least one detection table data representing a passed detection. The preset database may be a matrix database for storing the at least one detection table data corresponding to the at least one detection result that represents a passed detection. For example, the preset database may be a HANA database. The access request information may be access request information received from a user.

[0106] In some optional implementations of certain embodiments, the access rights of the user identity information corresponding to the access request information to at least one detection table data representing the successful detection may include the following steps:

[0107] The first step, in response to the detected access request information, is to verify the user's identity information corresponding to the access request information, and obtain the identity verification result. The aforementioned user identity information can be information that represents the user's identity. For example, the aforementioned user identity information can be at least one of the following: user account information, username. The aforementioned identity verification result can include: identity verification passed and identity verification failed.

[0108] As an example, the aforementioned executing entity can perform text similarity matching between user identity information and a set of identity information stored in a preset database to obtain a set of matching scores. Upon determining that there is a matching score greater than or equal to a preset matching score threshold in the set of matching scores, the identity verification result is determined to be successful. The preset matching score threshold can be 0.98. Upon determining that there is no matching score greater than or equal to the preset matching score threshold in the set of matching scores, the identity verification result is determined to be unsuccessful.

[0109] The second step, in response to the determination that the aforementioned identity verification result indicates successful verification, is to determine the set of permission role information for the aforementioned user identity information. The permission role information in the aforementioned permission role information set can be a collective term for users with the same data access permissions.

[0110] As an example, the aforementioned execution entity could first determine the user category information set corresponding to the aforementioned user identity information. Then, it could determine the permission role information set corresponding to the aforementioned user category information set as the permission role information set for the aforementioned user identity information.

[0111] Third, in response to determining that the number of permission role information included in the aforementioned permission role information set is a preset number, the data of at least one detection table that has passed the representation detection is logically split according to the aforementioned permission role information set to obtain the split detection table dataset. The aforementioned preset number can be a threshold used to distinguish different access permissions. For example, the aforementioned preset number can be 1.

[0112] As an example, the aforementioned executing entity can first determine the set of access regions for at least one detection table data that has passed the representation detection using the aforementioned set of permission role information. Then, it extracts the detection table data corresponding to the aforementioned access region set from the at least one detection table data that has passed the representation detection, as a regional detection table dataset. Finally, it concatenates the aforementioned regional detection table datasets to obtain the concatenated detection table data, which serves as the split detection table dataset.

[0113] The fourth step is to create a detection view for each of the split detection table data in the split detection table dataset mentioned above, thus obtaining a detection view set.

[0114] Fifth, set the above detection view set to the access permissions of the user's identity information.

[0115] Optionally, after setting the detection view set to the access permissions of the user's identity information, the method may further include the following steps:

[0116] The first step is to determine the number of permission role information included in the above permission role information set that is not a preset number, and to determine the detection table data group corresponding to each permission role information in the above permission role information set, so as to obtain the detection table data group set as the detection permission table data group set.

[0117] The second step is to set the access duration for the aforementioned detection permission table data set, and to set the read permissions for the aforementioned user identity information to the aforementioned detection permission table data set.

[0118] The third step is to set the above-mentioned detection permission table data set, the above-mentioned access duration, and the above-mentioned detection permission table data set as the access permissions of the user's identity information.

[0119] The various embodiments of this disclosure have the following beneficial effects: the detection table data storage method of some embodiments of this disclosure can reduce redundant and erroneous detection table data, reduce the waste of storage resources, improve user experience, and enhance data security. Specifically, the waste of related storage resources is caused by: determining the detection template through subjectively set detection table template rules. Since subjectively set detection table template rules have certain subjectivity and limitations and cannot be updated in real time, the accuracy of the obtained detection table template is low. Consequently, the detection table dataset corresponding to the detection template contains a large amount of redundant data, resulting in wasted storage resources and a poor user experience. Based on this, the detection table data storage method of some embodiments of this disclosure can first obtain detection table datasets from different data sources. The detection table data in the aforementioned detection table datasets includes: detection table structure information and field attribute value groups corresponding to the detection table structure information. Here, the detection table datasets from different data sources are used for subsequent generation of detection knowledge graphs. Secondly, according to preset detection template requirement information, the detection table structure information set is audited to obtain an audit result set. Here, auditing the detection table structure information set according to preset detection template requirement information can reduce the computational load of subsequent knowledge graph generation and reduce the waste of computational resources. Next, a knowledge graph is constructed from the data of at least one detection table corresponding to at least one approved review result in the aforementioned review result set, resulting in a detection knowledge graph. Here, knowledge graph construction can more accurately represent the relationships between detection table datasets and display detection table datasets from different sources, facilitating the improvement of the accuracy of subsequently generated detection table templates. Then, based on the aforementioned detection knowledge graph and the different data sources, a set of detection table templates is generated. This improves the accuracy of the obtained detection table template set, accurately distinguishing detection templates from different data sources, thereby reducing a large amount of redundant data. Subsequently, a set of comment text information for the aforementioned detection table template set is obtained. Here, the comment text information set is used for subsequent dynamic adjustments to the detection table template set. Afterwards, the aforementioned comment text information set is used for demand identification, resulting in a demand information set. Based on this demand information set, each detection table template in the aforementioned detection table template set is dynamically adjusted, resulting in an adjusted detection table template set. Here, by dynamically adjusting the detection table template set through the demand information set, the accuracy of the detection table templates is ensured, and they are more adaptable to user needs, improving the user experience. Then, data quality checks are performed on the set of field attribute values ​​included in each test table corresponding to the adjusted test table template set, resulting in a test result set. Here, data quality checks can improve the data quality of the set of field attribute values, reduce erroneous data, and thus reduce the waste of storage resources.Finally, the data of at least one detection table corresponding to at least one detection result that passed the above-mentioned detection results set is stored in a preset database, and access permissions for the at least one detection table data that passed the detection are set according to the user identity information corresponding to the access request information. This reduces the waste of storage resources and improves the security of the detection table data, thus reducing data leakage to some extent. Therefore, this detection table data storage method, by performing structural detection on the detection table dataset, generating a detection knowledge graph from at least one detection table data that passed the structural detection, constructing a detection table template set, and reviewing and setting user permissions for field attribute value sets, can reduce redundant and erroneous detection table data, reduce the waste of storage resources, improve user experience, and enhance data security.

[0120] Further reference Figure 2 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a detection table data storage device, which are similar to... Figure 1 Corresponding to the method embodiments shown, this detection table data storage device can be specifically applied to various electronic devices.

[0121] like Figure 2As shown, a test table data storage device 200 includes: a first acquisition unit 201, a table structure review unit 202, a knowledge graph construction unit 203, a generation unit 204, a second acquisition unit 205, a requirement identification unit 206, a data quality detection unit 207, and a storage unit 208. The first acquisition unit 201 is configured to acquire test table datasets from different data sources, wherein the test table data in the datasets includes test table structure information and corresponding field attribute value groups. The table structure review unit 202 is configured to review the test table structure information set according to preset test template requirement information, obtaining a review result set. The knowledge graph construction unit 203 is configured to construct a knowledge graph from at least one test table data corresponding to at least one review result in the review result set that represents a passed review, obtaining a test knowledge graph. The generation unit 204 is configured to generate a test table template set based on the test knowledge graph and the different data sources. The second acquisition unit 205 is configured to acquire a set of comment text information for the test table template set. The requirement identification unit 206 is configured to: identify requirements from the aforementioned comment text information set to obtain a requirement information set; and dynamically adjust each test table template in the aforementioned test table template set based on the aforementioned requirement information set to obtain an adjusted test table template set. The data quality detection unit 207 is configured to: perform data quality detection on the set of field attribute values ​​included in each test table data corresponding to the aforementioned adjusted test table template set to obtain a detection result set. The storage unit 208 is configured to: store at least one test table data corresponding to at least one test result that has passed detection in the aforementioned detection result set into a preset database; and set access permissions for the user identity information corresponding to the access request information to the at least one test table data that has passed detection.

[0122] It is understandable that the units recorded in the test table data storage device 200 are related to the reference. Figure 1 The steps in the described method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the test table data storage device 200 and the units contained therein, and will not be repeated here.

[0123] The following is for reference. Figure 3 It shows a schematic diagram of the structure of an electronic device (e.g., an electronic device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0124] like Figure 3As shown, the electronic device 300 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0125] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0126] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from storage device 308, or installed from ROM 302. When the computer program is executed by processing device 301, it performs the functions defined in the methods of some embodiments of this disclosure.

[0127] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0128] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0129] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: acquire detection table datasets from different data sources, wherein the detection table data in the aforementioned detection table datasets includes: detection table structure information and corresponding field attribute value groups; perform table structure audits on the detection table structure information set according to preset detection template requirement information, obtaining an audit result set; construct a knowledge graph on at least one detection table data corresponding to at least one audit result representing a passed audit in the aforementioned audit result set, obtaining a detection knowledge graph; and generate a detection table template set based on the aforementioned detection knowledge graph and the aforementioned different data sources. The process involves: acquiring a set of comment text information for the aforementioned test table template set; identifying the requirements of the comment text information set to obtain a set of requirements information; dynamically adjusting each test table template in the aforementioned test table template set based on the aforementioned requirements information set to obtain an adjusted test table template set; performing data quality checks on the set of field attribute values ​​included in each test table data corresponding to the aforementioned adjusted test table template set to obtain a set of test results; storing at least one test table data corresponding to at least one test result that passed the test in the aforementioned test results set into a preset database; and setting access permissions for the user identity information corresponding to the access request information to the at least one test table data that passed the test.

[0130] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0131] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0132] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including a first acquisition unit, a table structure verification unit, a knowledge graph construction unit, a generation unit, a second acquisition unit, a demand identification unit, a data quality detection unit, and a storage unit. The names of these units do not necessarily limit the specific unit; for example, the first acquisition unit may also be described as "a unit that acquires detection table datasets from different data sources."

[0133] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0134] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for storing test table data, comprising: Obtain detection table datasets from different data sources, wherein the detection table data in the detection table dataset includes: detection table structure information and the field attribute value group corresponding to the detection table structure information; Based on the preset testing template requirements, the test table structure information set is reviewed to obtain the review result set; A knowledge graph is constructed on at least one detection table data corresponding to at least one approved review result in the review result set to obtain a detection knowledge graph. Based on the detection knowledge graph and the different data sources, a set of detection table templates is generated; Obtain the set of comment text information for the detection table template set; The comment text information set is subjected to demand identification to obtain a demand information set, and each detection table template in the detection table template set is dynamically adjusted according to the demand information set to obtain an adjusted detection table template set. Data quality testing is performed on the set of field attribute values ​​included in each test table data corresponding to the adjusted test table template set to obtain a test result set; The detection results are centrally stored in a preset database, and the user identity information corresponding to the at least one detection result that has passed the detection is set to grant access permissions to the at least one detection table data that has passed the detection, as well as the user identity information corresponding to the access request information.

2. The method according to claim 1, wherein, The step involves reviewing the structure of the detection table information set based on the preset detection template requirements to obtain a review result set, including: Intelligent requirement construction is performed on the preset detection template requirement information to obtain a template contract code information set; Vulnerability detection is performed on each template contract code information in the template contract code information set to generate vulnerability detection results and obtain a vulnerability detection result set; Remove each template contract code information corresponding to at least one vulnerability detection result that represents a failed vulnerability detection from the vulnerability detection result set in the above template contract code information set to obtain the code information set after removal. For each piece of test table structure information in the aforementioned test table structure information set, perform the following table structure review steps: Based on the removed code information set, structural detection is performed on the detection table structure information to obtain a structure detection result set; In response to determining that there are structural detection results in the set of structural detection results that indicate that the structural detection has failed, the structural detection failure is determined as the review result of the structural information in the detection table. In response to determining that there are no structural detection results in the set of structural detection results that indicate that the structural detection has failed, the structural detection results that indicate that the structural detection has passed are determined as the review result of the structural information in the detection table.

3. The method according to claim 1, wherein, The step of constructing a knowledge graph from at least one detection table data corresponding to at least one approved review result in the review result set to obtain a detection knowledge graph includes: The data of the at least one detection table is preprocessed to obtain a preprocessed detection table dataset; Determine the set of associations between the preprocessed detection table data included in the preprocessed detection table dataset; Based on the preprocessed detection table dataset and the association set, ontology mapping text information is generated, wherein the ontology mapping text information includes: a preprocessed detection table name set, a field set, an attribute value set corresponding to the field set, and an association set; Based on the ontology mapping text information, an initial detection knowledge graph is generated; The initial detection knowledge graph is aligned with entities to obtain the fused knowledge graph; Knowledge reasoning is performed on the fused knowledge graph to obtain a reasoned knowledge graph, which serves as the detection knowledge graph.

4. The method according to claim 1, wherein, The step of identifying requirements from the comment text information set to obtain a requirement information set, and dynamically adjusting each detection table template in the detection table template set based on the requirement information set to obtain an adjusted detection table template set, includes: Anomaly detection is performed on the comment text information set to obtain an abnormal comment text information set; Remove the abnormal comment text information from the comment text information set to obtain the removed comment text information set, which is used as the target comment text information set; For each target comment text in the target comment text information set, perform the following feature word generation steps: The target comment text information is segmented to obtain a segmented character sequence; The segmented character sequence is subjected to word embedding processing to obtain a character embedding feature vector sequence; The word embedding feature vector sequence is weighted and summed through a multi-head attention mechanism layer to obtain word embedding feature vectors with different weight values, which are used as word embedding weight feature vector sequences. The word embedding weight feature vector sequence is subjected to text weighting to obtain a text weight feature vector sequence; The text weight feature vector sequence and the word embedding weight feature vector sequence are aggregated to obtain an aggregated feature vector sequence. The aggregated feature vector sequence is subjected to feature compression processing to obtain a compressed feature vector sequence; The compressed feature vector sequence is classified to obtain the detection feature word group for the target comment text information, wherein the detection feature words in the detection feature word group are words used to describe the structural information of the detection table template; Sentiment recognition is performed on each detected feature word in the obtained set of detected feature words to generate sentiment tendency, resulting in a sentiment tendency set, wherein the sentiment tendency set includes: a sentiment tendency representing positive sentiment tendency, a sentiment tendency representing neutral sentiment tendency, and a sentiment tendency representing negative sentiment tendency. Determine at least one target comment text information corresponding to a negative sentiment tendency to obtain a negative comment text information set, and dynamically adjust each detection table template in the detection table template set according to the negative comment text information set to obtain an adjusted detection table template set.

5. The method according to claim 4, wherein, The step of performing text weighting processing on the word embedding weight feature vector sequence to obtain a text weight feature vector sequence includes: For each character embedding weight feature vector in the sequence of character embedding weight feature vectors, perform the following weight determination steps: Determine the character embedding window corresponding to the character embedding weight feature vector, wherein the character embedding window is a window centered on the character embedding weight feature vector and with a preset text distance threshold as the window size; Determine the first text distance value between each character embedding weight feature vector included in the character embedding window and the character embedding weight feature vector, and obtain the first text distance value sequence; Determine the second text distance value between each remaining character embedding weight feature vector in the remaining character embedding weight feature vector sequence and the character embedding weight feature vector, to obtain the second text distance value sequence, wherein the remaining character embedding weight feature vector sequence is obtained by removing each character embedding weight feature vector included in the character embedding window from the character embedding weight feature vector sequence; The weight values ​​of the word embedding weight feature vector are determined based on the first text distance value sequence and the second text distance value sequence. The weight values ​​and the word embedding weight feature vectors are weighted to obtain the text weight feature vector.

6. The method according to claim 1, wherein, The access rights of the user identity information corresponding to the access request information to at least one detection table data representing the successful detection include: In response to the detection of access request information, the user identity information corresponding to the access request information is verified to obtain the identity verification result. In response to determining that the identity verification result indicates that the verification has been passed, the permission role information set of the user identity information is determined; In response to determining that the number of permission role information included in the permission role information set is a preset number, the data logic splitting process is performed on at least one detection table data that represents the passing of the detection based on the permission role information set to obtain the split detection table dataset. Create a detection view for each of the split detection table data in the split detection table dataset to obtain a detection view set; Set the detection view set to the access permissions of the user's identity information.

7. The method according to claim 6, wherein, After setting the detection view set to the access permissions of the user's identity information, the method further includes: In response to determining that the number of permission role information included in the permission role information set is not a preset number, the detection table data group corresponding to each permission role information in the permission role information set is determined to obtain the detection table data group set, which is used as the detection permission table data group set; Set the access duration of the detection permission table data set, and set the read permission of the user identity information for the detection permission table data set; Set the detection permission table data set, the access duration, and the read permission of the detection permission table data set as the access permissions of the user's identity information.

8. A test table data storage device, comprising: The first acquisition unit is configured to acquire detection table datasets from different data sources, wherein the detection table data in the detection table dataset includes: detection table structure information and field attribute value groups corresponding to the detection table structure information; The table structure auditing unit is configured to audit the table structure information set of the test table according to the preset test template requirements information, and obtain the audit result set. The knowledge graph construction unit is configured to construct a knowledge graph from at least one detection table data corresponding to at least one audit result that represents the audit passed in the audit result set, thereby obtaining a detection knowledge graph. The generation unit is configured to generate a set of detection table templates based on the detection knowledge graph and the different data sources; The second acquisition unit is configured to acquire a set of comment text information for the detection table template set; The demand identification unit is configured to identify the demand in the comment text information set to obtain a demand information set, and to dynamically adjust each detection table template in the detection table template set according to the demand information set to obtain an adjusted detection table template set. The data quality detection unit is configured to perform data quality detection on the set of field attribute values ​​included in the data of each detection table corresponding to the adjusted detection table template set, and obtain a detection result set; The storage unit is configured to store at least one detection table data corresponding to at least one detection result that represents a successful detection in the detection result set into a preset database, and to set access permissions for the at least one detection table data representing a successful detection to the user identity information corresponding to the access request information.

9. An electronic device, comprising: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

10. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-7.

Images