Artificial intelligence-based analysis system using automated feature extraction and method therefor

Abstract

An embodiment of the present invention provides an artificial intelligence-based analysis system for analyzing data using automated feature extraction, the system comprising: a data interpretation unit that classifies data stored in a database and analyzes context; a feature extraction unit that extracts features of the data on the basis of the analyzed context; a feature-based data structure reconstruction unit that reconstructs a database structure on the basis of feature fields using the extracted features of the data; an analysis model provision unit that provides or recommends an analysis model suitable for feature field data of the reconstructed database structure; an analysis model application unit that generates a result by applying the analysis model provided by the analysis model provision unit to the feature field data of the reconstructed database; and an indicator derivation unit that analyzes a relationship between a business objective and the feature field data, infers a causal relationship, and selects goal-oriented features for the analysis model.

Description

Artificial intelligence-based analysis system and method using automated feature extraction

[0001] The present invention relates to a data analysis system, and more specifically, to a data analysis system and method that utilizes automated feature extraction and artificial intelligence technology to systematically analyze data in a database.

[0002] The business environment is becoming increasingly complex due to rapid change and rapid technological innovation. Companies are actively adopting automation technologies to achieve goals such as cost reduction, productivity improvement, and enhanced customer service. Consequently, the importance of data analysis and utilization is becoming more prominent, and companies are employing various technological approaches to extract meaningful insights from vast amounts of data. In particular, technologies that automatically identify and classify the diverse data structures and meanings stored in databases are garnering attention as core technologies capable of maximizing the efficiency of data utilization.

[0003] However, despite collecting and storing diverse data, companies still struggle to effectively integrate this data into their decision-making processes. While most companies aim to achieve their goals through analysis using their data, they reveal limitations in failing to break away from conventional approaches that rely on the subjective judgment of data or corporate analysts.

[0004] Existing analysis systems are limited to analyzing or deriving data characteristics, failing to provide methods for actually applying these derived features to the analyses required by companies. As a result, companies continue to face difficulties in performing necessary analyses and making strategic decisions based on the results.

[0005] Furthermore, if a company has not consistently defined database rules or if there are structural design flaws, it may be difficult to clearly verify the data it possesses. Additionally, since most corporate databases were not originally constructed to derive features for AI-based analysis models, the desired results cannot be obtained if the database structure and data are applied directly to the analysis model.

[0006] To address these issues, the data classification system must be streamlined to enable efficient utilization for corporate analysis. Furthermore, a system is required to extract and leverage key characteristics aligned with the company's industry or business objectives. Such a system can support efficient data utilization and strategic decision-making by systematically organizing the company's data and deriving objective and reliable analysis results based on it.

[0007] To solve the aforementioned problems, the technical objective of the present invention is to provide an analysis system and method using automated feature extraction that can automatically extract features from a corporate database to simplify the analysis preparation process and reduce the time required for corporate analysis.

[0008] Furthermore, the technical problem that the present invention aims to solve is to provide an analysis system and method using automated feature extraction that obtains analysis results suitable for a company by systematically understanding the data held by the company and providing a suitable analysis model.

[0009] Furthermore, the technical problem that the present invention aims to solve is to provide an analysis system and method using automated feature extraction that learns the potential relationships of data based on artificial intelligence to derive analysis results that are more critical and related to performance indicators.

[0010] To achieve the above technical objectives, an artificial intelligence-based analysis system according to an embodiment of the present invention comprises: a data interpretation unit that classifies data stored in a database and analyzes context; a feature extraction unit that extracts features of the data based on the analyzed context; a feature-based data structure reconstruction unit that reconstructs the database structure based on feature fields based on the extracted features of the data; an analysis model providing unit that provides or recommends an analysis model suitable for the feature field data of the reconstructed database structure; an analysis model application unit that applies the analysis model provided by the analysis model providing unit to the feature field data of the reconstructed database to generate results; and an indicator derivation unit that analyzes the relationship between a business goal and the feature field data, infers causal relationships, and selects goal-oriented features of the analysis model.

[0011] An analysis method using an artificial intelligence-based analysis system according to an embodiment of the present invention may include: (a) a step of classifying data stored in a database and analyzing the context; (b) a step of extracting features of the data based on the analyzed context; (c) a step of reconstructing a database structure based on feature fields based on the features of the extracted data; (d) a step of providing or recommending an analysis model suitable for the feature field data of the reconstructed database structure; (e) a step of generating results by applying the analysis model to the feature field data of the reconstructed database; and (f) a step of analyzing the relationship between a business goal and the feature field data and inferring a causal relationship to select a goal-oriented feature of the analysis model.

[0012] According to an embodiment of the present invention, an artificial intelligence-based analysis system can automatically extract features from a corporate database. This simplifies the analysis preparation process, thereby reducing the time required for corporate analysis.

[0013] Furthermore, an AI-based analysis system utilizing automated feature extraction can provide a method to automate the data analysis process using artificial intelligence. Through this, it is possible to enhance analysis efficiency and the accuracy of results, and to derive key analytical outcomes for the company.

[0014] The effects of the present invention are not limited to the effects described above, and should be understood to include all effects that can be inferred from the composition of the invention described in the description or claims of the present invention.

[0015] FIG. 1 is a diagram illustrating an overview of an artificial intelligence-based analysis system using automated feature extraction according to an embodiment of the present invention.

[0016] FIG. 2 is an overall block diagram illustrating the configuration of an artificial intelligence-based analysis system using automated feature extraction according to an embodiment of the present invention.

[0017] FIG. 3 is a block diagram illustrating the configuration of a data interpretation unit (110) of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0018] FIG. 4 is a block diagram illustrating the configuration of an analysis model providing unit (140) of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0019] FIG. 5 is an indicator derivation unit of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0020] This is a block diagram illustrating the configuration of (160).

[0021] FIG. 6 is a flowchart illustrating the overall analysis method of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0022] FIG. 7 is a flowchart illustrating a detailed method for data interpretation of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0023] FIG. 8 is a flowchart illustrating a detailed method for applying an analysis model of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0024] FIG. 9 is a flowchart illustrating a detailed method for selecting important features of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0025] FIG. 10 illustrates a computing device that implements a drug responsiveness prediction system (100) according to an embodiment of the present invention.

[0026] The present invention will be described below with reference to the attached drawings. However, the present invention may be implemented in various different forms and is therefore not limited to the embodiments described herein. Furthermore, in order to clearly explain the present invention in the drawings, parts unrelated to the explanation have been omitted, and similar parts throughout the specification have been given similar reference numerals.

[0027] Throughout the specification, when it is stated that a part is "connected (connected, in contact, combined)" with another part, this includes not only cases where they are "directly connected," but also cases where they are "indirectly connected" with other members interposed between them. Furthermore, when it is stated that a part "includes" a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but rather allows for the inclusion of additional components.

[0028] The terms used herein are merely for describing specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as “comprising” or “having” are intended to indicate the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0029] In this specification, "module" includes a unit composed of hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit. A module may be a component formed as a whole, or a minimum unit or part thereof that performs one or more functions. For example, a module may be composed of an application-specific integrated circuit (ASIC).

[0030] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

[0031] FIG. 1 is a diagram illustrating an overview of an artificial intelligence-based analysis system using automated feature extraction according to an embodiment of the present invention.

[0032] As illustrated in FIG. 1, the artificial intelligence-based analysis system (100) can extract features of the data by interpreting the data stored in the database (10). The artificial intelligence-based analysis system (100) can extract features from the database (10) owned by the company and apply an analysis model to derive indicators required by the company.

[0033] The database (10) is a centralized repository that allows multiple applications to share and access data simultaneously for efficient data management. The data collected and stored by the company can be structured in a predetermined structure within the database (10). However, as previously mentioned, since the database (10) is not constructed to derive features of an artificial intelligence-based analysis model, it is necessary to interpret and reconstruct it so that the analysis model can understand it.

[0034] To this end, the AI-based analysis system (100) collects and interprets attributes structured in a table form from the database (10). Specifically, the AI-based analysis system (100) can automatically collect and interpret the schema, table name, column name, comments, etc. of the database (10). Meanwhile, the database (10) may have various naming rules specified during the design and operation process, so column names and field data may not be consistent. The AI-based analysis system (100) can efficiently extract the characteristics of the data by utilizing Natural Language Processing (NLP) to consistently organize the meaning and role of the data.

[0035] Based on the characteristics of the above data, the database (10) data can be reorganized into a new structure in which the lowest column is a characteristic field.

[0036] In this way, an analysis model is applied to the values ​​of the extracted feature fields, and the indicators provided by the analysis model can be derived.

[0037] To this end, the artificial intelligence-based analysis system (100) provides various analysis models and may present in advance the data requirements for applying the analysis models. Through this, the company can not only utilize the database (10) more efficiently but also easily apply a suitable analysis method.

[0038] Below, the specific configuration of the artificial intelligence-based analysis system (100) will be described.

[0039] FIG. 2 is a block diagram illustrating the configuration of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0040] Referring to FIG. 2, the artificial intelligence-based analysis system (100) may include a data interpretation unit (110), a feature extraction unit (120), a feature-based data structure reconstruction unit (130), an analysis model provision unit (140), an analysis model application unit (150), and an indicator derivation unit (160).

[0041] The data interpretation unit (110) can systematically classify and understand data stored in the database (10). The data interpretation unit (110) can understand the meaning and role defined by the data fields by analyzing the context in the names and comment descriptions of tables and columns. Here, the data interpretation unit (110) can analyze the context using Natural Language Processing (NLP). The specific configuration of this data interpretation unit (110) will be explained later with reference to FIG. 3.

[0042] The feature extraction unit (120) can extract features for modeling field relationships in a database based on the context in which meaning is defined and analyzed in the data interpretation unit (110). To this end, the feature extraction unit (120) can statistically apply the interactions between data relationships, data distributions, and data attributes by analyzing schema information. Additionally, the feature extraction unit (120) can automatically group highly related fields by calculating correlation coefficients between data fields and identify key features by numerically evaluating the interactions between data fields.

[0043] The feature-based data structure reconstruction unit (130) can reconstruct the database structure based on feature fields based on features derived from the feature extraction unit (120). The feature-based data structure reconstruction unit (130) can integrate fields or define new categories for data clarity. To this end, the feature-based data structure reconstruction unit (130) can remove duplicate data and unnecessary information from the existing database structure. That is, for example, if the customer name field in the 'Customer Information' table has both a column defined as 'Customer_Name' and a column defined as 'CustName', the feature-based data structure reconstruction unit (130) can integrate them into a single field and change the name to 'CustomerName'. Additionally, the amount information defined as 'Total_Amount' and 'TotalSales' in the 'Sales Data' can be integrated into a single field, and duplicate data can be removed.

[0044] The analysis model providing unit (140) can provide or recommend an analysis model suitable for the characteristic field data of the reconstructed database structure. The analysis model providing unit (140) analyzes the characteristics of the data stored in the database (e.g., equipment operation logs, sales data, training performance data, etc.) and provides the artificial intelligence analysis algorithm and analysis technique most suitable for the data. In addition, the analysis model providing unit (140) can recommend an applicable analysis model according to the characteristics of the data in the database.

[0045] The analysis model application unit (150) can perform analysis by applying the analysis model and data provided by the analysis model provision unit (140). To do this, the analysis model application unit (150) can call feature field data from the database (10). Additionally, the analysis model application unit (150) can generate analysis results required by the company by executing the analysis model with the called data as input. For example, the analysis model application unit (150) can execute an analysis model regarding maintenance by calling equipment operation logs and error data from the manufacturing company database. In this process, it can generate analysis results that predict the possibility of operational abnormalities of a specific machine and suggest a maintenance schedule before failure.

[0046] The indicator derivation unit (160) can obtain the results analyzed by the analysis model application unit (150) and derive indicators linked to business goals. Here, important features corresponding to corporate insights can be selected to achieve business goals linked to business goals. In addition, the indicator derivation unit (150) can evaluate the performance of the existing analysis model and perform analysis model improvement through the derived indicators. The specific configuration of this indicator derivation unit (160) will be explained later with reference to FIG. 5.

[0047] In this way, the data interpretation unit (110), feature extraction unit (120), and feature-based data structure reconstruction unit (130) of the artificial intelligence-based analysis system (100) can extract data features and reconstruct the data structure by understanding the meaning of the database (10).

[0048] In addition, the analysis model providing unit (140), analysis model application unit (150), and indicator derivation unit (160) of the artificial intelligence-based analysis system (100) can apply the analysis model based on data and derive an analysis result corresponding to the analysis goal.

[0049] Accordingly, the artificial intelligence-based analysis system (100) can perform the process of deriving indicators to achieve the intended goal by applying analysis methods suitable for the characteristics of the data and business goals based on an understanding of the data.

[0050] FIG. 3 is a block diagram illustrating the configuration of a data interpretation unit (110) of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0051] Referring to FIG. 3, the data interpretation unit (110) may include a data type classification unit (111), a metadata analysis unit (112), an NLP-based data analysis unit (113), and a context analysis unit (114). Each component of the data interpretation unit (110) functions in parallel, but it is understood that mutually complementary functions can be performed as needed.

[0052] First, the data type classification unit (111) automatically classifies data types by analyzing data fields of the database (10). The data type classification unit (111) can automatically classify data types using a machine learning algorithm, and the data types classified here can be classified, for example, into numeric, categorical, time-related, spatial-related, financial-related, ratio, etc.

[0053] The metadata analysis unit (112) can identify statistical characteristics of data by analyzing metadata such as the schema, table name, column name, and comments of the database (10). The metadata analysis unit (112) can identify statistical characteristics of data by analyzing the definition of relationships between tables and the data range and validity of columns using the metadata.

[0054] The NLP-based data analysis unit (113) analyzes the meaning of data fields by utilizing natural language processing (NLP) artificial intelligence. Specifically, field names in the database (10) can be tokenized, and the meaning of each field can be calculated using a pre-trained natural language processing (NLP) model. Here, the natural language processing (NLP) model calculates cosine similarity based on vectors to group semantically similar features or define them as new features. For example, if the similarity between the 'sales volume' and 'quantity' fields is high, they can be defined as a combined feature called 'sales quantity'.

[0055] The context analysis unit (114) can analyze the context of data relationships based on the data type classification result, the metadata analysis result, and the NLP-based semantic analysis result. The context analysis unit (114) can derive the characteristics of data relationships by analyzing the causality and mutual context between features with clearly defined meanings.

[0056] To summarize, the data type classification unit (111) and the metadata analysis unit (112) perform an initial process of classifying and interpreting data in the database (10). The NLP-based data analysis unit (113) can define features from text such as field names or comment descriptions. The context analysis unit (114) understands the characteristics of relationships between data within the overall context. In this way, the data interpretation unit (110) provides a foundation for comprehensively understanding the characteristics of the database, and consistency can be ensured in the analysis process of the artificial intelligence-based analysis system (100).

[0057] FIG. 4 is a block diagram illustrating the configuration of an analysis model providing unit (140) of an artificial intelligence-based analysis system according to an embodiment of the present invention.

[0058] Referring to FIG. 4, the analysis model providing unit (140) includes a knowledge graph analysis model (141), a graph neural network model (142), an automatic correlation analysis model (143), and an analysis model derivation unit (144).

[0059] The knowledge graph analysis model (141) is a model that represents data as a structured graph and analyzes the relationships between the data. The knowledge graph analysis model (141) can analyze the relationships between the schema and tables of the database (10) and represent them as graphs. For example, the knowledge graph analysis model (141) creates a knowledge graph structure by representing the entities of each table as nodes and relationships such as foreign keys between tables as edges.

[0060] Additionally, the knowledge graph analysis model (141) can integrate domain knowledge possessed by the company into the knowledge graph structure. This allows the knowledge graph analysis model (141) to be implemented so that the company's potential associations can be expressed by reflecting the company's expertise or business rules as additional nodes and edges.

[0061] The graph neural network model (142) is a model that performs learning in a graph structure to predict complex relationships between data and learn potential patterns. The graph neural network model (142) may use a graph neural network (GNN) that learns relationship prediction and node importance using graph data generated from the knowledge graph analysis model (141) as input. The graph neural network model (142) can analyze the importance and potential characteristics of data relationships in the patterns derived from the learning results.

[0062] The automatic correlation analysis model (143) can analyze the correlation of data through quantitative calculations such as correlation coefficients and statistical significance between data. To explain using marketing ROI analysis as an example, the automatic correlation analysis model (143) can analyze the correlation between cost and sales data by advertising channel to determine which channel is effective.

[0063] The analysis model derivation unit (144) automatically derives an optimal analysis model based on the characteristics of the database (10) and the setting of business goals. The analysis model derivation unit (144) can derive and provide a suitable analysis model and a combination of analysis models among a knowledge graph analysis model (141), a graph neural network model (142), and an automatic correlation analysis model (143). In addition, the analysis model derivation unit (144) can dynamically provide models suitable for various analysis purposes and can also generate customized analysis models according to the specific needs of the user. For example, suitable candidate analysis models may be automatically generated according to analysis goals such as customer churn analysis, sales forecasting, and product recommendation.

[0064] FIG. 5 is a block diagram illustrating the configuration of an indicator derivation unit (160) in an artificial intelligence-based analysis system (100) according to an embodiment of the present invention.

[0065] Referring to FIG. 5, the indicator derivation unit (160) may include a time indicator analysis unit (161), a data refinement unit (162), a business goal linkage unit (163), and an indicator feature selection unit (164).

[0066] The time indicator analysis unit (161) can analyze trends, seasonality, and periodicity by deriving key points and patterns from time-related data. To this end, the time indicator analysis unit (161) can break down the data into various time units, such as days, weeks, months, and quarters, and identify key points and patterns. For example, the time indicator analysis unit (161) can perform trend analysis and prediction by identifying the trend of change in the data. In addition, the time indicator analysis unit (161) can analyze seasonality and business cycles by identifying patterns that occur periodically at specific points in time.

[0067] The data cleaning unit (162) can clean the data to be analyzed so that it becomes valid for specific goals or performance criteria. To this end, the data cleaning unit (162) can automatically process the data to be analyzed, including handling missing values, detecting outliers, and resolving data inconsistencies. For example, the data cleaning unit (162) can improve the data by performing feature engineering. By modifying or creating input data characteristics through feature engineering, model training becomes faster and accurate predictions become possible.

[0068] The business goal linkage unit (163) can specify or set business goals that the company targets. Here, business goals can be set by the company and may be business goals to be performed on a time-by-time basis. Business goals include increasing revenue, retaining customers, and reducing costs, and are closely related to data analysis. For example, if a company possessing a database (10) sets increasing revenue as a business goal, it performs sales data analysis to derive a profitable product.

[0069] The indicator feature selection unit (164) can select important features for achieving business goals linked in the business goal linkage unit (163). The indicator feature selection unit (164) can select important features related to time indicators and other indicators for achieving business goals. Additionally, the indicator feature selection unit (164) can automatically select goal-oriented feature values ​​using an artificial intelligence causal inference model.

[0070] FIG. 6 is a flowchart illustrating the overall analysis process of an artificial intelligence-based analysis system (100) according to an embodiment of the present invention.

[0071] In step (S110), the artificial intelligence-based analysis system (100) can systematically classify and understand the data stored in the database (10). In this step, the artificial intelligence-based analysis system (100) can understand the meaning defined by the data fields by analyzing the context from the names of tables and columns and comment descriptions. Additionally, the artificial intelligence-based analysis system (100) can analyze the context using Natural Language Processing (NLP).

[0072] Here, the detailed method for step (S110), that is, data interpretation, will be described later with reference to FIG. 7.

[0073] In step (S120), the AI-based analysis system (100) can extract features for modeling field relationships in a database based on the context in which meaning is defined and analyzed in step (S110). In this step, the AI-based analysis system (100) can statistically apply the interactions between data relationships, data distributions, and data attributes by analyzing schema information. Additionally, it can automatically group highly related fields by calculating correlation coefficients between data fields and identify key features by numerically evaluating the interactions between data fields.

[0074] In step (S130), the AI-based analysis system (100) can reconstruct the data structure based on feature fields based on the features analyzed in steps (S110–S120). In this step, the AI-based analysis system (100) can integrate fields or define new categories for data clarity. In step (S130), redundant data and unnecessary information in the existing database structure can be removed.

[0075] In step (S140), the artificial intelligence-based analysis system (100) can provide or recommend an analysis model suitable for the feature field data of the reconstructed database structure. In this step, the artificial intelligence-based analysis system (100) can recommend an applicable analysis model according to the data characteristics of the database. Here, the detailed method for step (S140), i.e., the analysis model, will be described later with reference to FIG. 8.

[0076] In step (S150), the artificial intelligence-based analysis system (100) can perform analysis by applying the analysis model and data provided in step (S140). To do this, the artificial intelligence-based analysis system (100) can call feature field data from the database (10), and execute the analysis model with the called data as input to generate an analysis result.

[0077] In step (S160), the AI-based analysis system (100) can link business goals and derive indicators linked to business goals. In this step, business goals are linked, and important features corresponding to corporate insights can be selected to achieve the goals. Additionally, through the indicators derived in step (S160), the AI-based analysis system (100) can evaluate the performance of existing analysis models and perform analysis model improvements. Here, the detailed method for step (S160), that is, deriving analysis indicators, will be described later with reference to FIG. 9.

[0078] FIG. 7 is a flowchart illustrating a detailed method for data interpretation of an artificial intelligence-based analysis system (100) according to an embodiment of the present invention.

[0079] In step (S111), the artificial intelligence-based analysis system (100) can automatically obtain schemas, table names, column names, comments, etc. from the database (10). To do this, the artificial intelligence-based analysis system (100) can access the database (10) and query, for example, the system catalog or information schema table.

[0080] In step (S112), the artificial intelligence-based analysis system (100) can classify data types. The data types classified in this step can be classified, for example, numeric, categorical, time-related, spatial-related, financial-related, ratio, etc.

[0081] In step (S113), the artificial intelligence-based analysis system (100) can identify the statistical characteristics of the data by analyzing metadata such as the schema, table name, column name, and comments of the database (10). In this step, the artificial intelligence-based analysis system (100) can identify the statistical characteristics of the data by using the metadata to analyze the definition of relationships between tables and the data range and validity of the columns.

[0082] In step (S114), the AI-based analysis system (100) analyzes the meaning of the data using natural language processing (NLP) AI. Specifically, the field names of the database (10) can be tokenized, and the meaning of each field can be calculated using a pre-trained natural language processing (NLP) model. At this time, the natural language processing (NLP) model calculates cosine similarity based on vectors to group semantically similar features or define them as new features. For example, if the similarity between the 'sales volume' and 'quantity' fields is high, they can be defined as a combined feature called 'sales quantity'.

[0083] In step (S115), the artificial intelligence-based analysis system (100) can analyze the context of data relationships based on the data type classification result, the metadata analysis result, and the NLP-based semantic analysis result. The artificial intelligence-based analysis system (100) can derive the characteristics of data relationships by analyzing the causality and mutual context between features with clearly defined meanings.

[0084] FIG. 8 is a flowchart illustrating a detailed method for applying an analysis model of an artificial intelligence-based analysis system (100) according to an embodiment of the present invention.

[0085] In step (S141), the AI-based analysis system (100) can model a knowledge graph analysis model (141). To do this, the AI-based analysis system (100) can analyze the relationships between the schema and tables of the database (10) and represent them as a graph. For example, a knowledge graph can be created by representing the entities of each table as nodes and relationships such as foreign keys between tables as edges. Additionally, in this step, the AI-based analysis system (100) can integrate domain knowledge possessed by the company into the knowledge graph. The AI-based analysis system (100) can implement the knowledge graph analysis model (141) by reflecting the company's expertise or business rules as additional nodes and edges, so that even the company's potential associations can be expressed.

[0086] In step (S142), the artificial intelligence-based analysis system (100) can model a graph neural network model (142). In this step, the system can model a graph neural network model (142) using a graph neural network (GNN) by learning relationship prediction and node importance using the knowledge graph data generated in step (S141) as input. The modeled graph neural network model (142) can analyze the importance and latent characteristics of data relationships in the patterns derived from the learning results.

[0087] In step (S143), the artificial intelligence-based analysis system (100) can model an automatic correlation analysis model (143). The automatic correlation analysis model (143) modeled in this step can analyze the correlation of the data through quantitative calculations such as correlation coefficients and statistical significance between the data.

[0088] In step (S144), the AI-based analysis system (100) can automatically derive an optimal analysis model from among the analysis models implemented in steps (S141–S143) according to the characteristics of the database (10) and the setting of business goals. In this step, the AI-based analysis system (100) can derive and provide a suitable analysis model and a combination of analysis models from among the modeled knowledge graph analysis model (141), graph neural network model (142), and automatic correlation analysis model (143). Additionally, in this step, models suitable for various analysis purposes can be dynamically provided, and customized analysis models can be generated according to the specific needs of the user. For example, suitable candidate analysis models can be automatically generated according to analysis goals such as customer churn analysis, sales forecasting, and product recommendation.

[0089] FIG. 9 is a flowchart illustrating a detailed method for selecting important features of an artificial intelligence-based analysis system (100) according to an embodiment of the present invention.

[0090] In step (S161), the artificial intelligence-based analysis system (100) can obtain data containing information related to time. In this step, the artificial intelligence-based analysis system (100) can break down the obtained data into various time units, such as days, weeks, months, and quarters, so that the data can be analyzed in time units.

[0091] In step (S162), the artificial intelligence-based analysis system (100) can derive time series indicators by analyzing data on an hourly basis. In this step, the artificial intelligence-based analysis system (100) can perform trend analysis and prediction by identifying the trend of data change on an hourly basis. Additionally, the artificial intelligence-based analysis system (100) can analyze seasonality and business cycles by identifying patterns that occur periodically at a specific point in time.

[0092] In step (S163), the AI-based analysis system (100) can refine the data to be analyzed so that it becomes valid for specific goals or performance criteria. In this step, the AI-based analysis system (100) can automatically process the data to be analyzed, including handling missing values, detecting outliers, and resolving data inconsistencies. For example, the AI-based analysis system (100) can improve the data by performing feature engineering. By modifying or creating input data characteristics through feature engineering, model training becomes faster and accurate predictions become possible.

[0093] In step (S164), the artificial intelligence-based analysis system (100) can specify or set business goals that the company targets. In this step, the artificial intelligence-based analysis system (100) can set business goals for the company, which may be business goals to be performed on a time-by-time basis.

[0094] In step (S165), the AI-based analysis system (100) analyzes the relationship between data features corresponding to business goals, and in step (S166), it can link the analyzed data features with the business goals. Then, in step (S167), the AI-based analysis system (100) can select important features for achieving business goals. In this step, the AI-based analysis system (100) can select important features related to temporal indicators and other indicators to achieve business goals linked to data features. Additionally, the AI-based analysis system (100) can automatically select goal-oriented features using a causal inference model. Furthermore, the AI-based analysis system (100) can generate an inference model using indicator feature values ​​as variables and improve it through evaluation and verification.

[0095] FIG. 10 illustrates a computing device that implements an artificial intelligence-based analysis system (100) according to an embodiment of the present invention.

[0096] An embodiment of the present invention described by FIGS. 1 to 9 may be implemented as a computing device (900) operated by at least one processor.

[0097] The computing device (900) may include a processor (910), memory (920), storage (930), a communication interface (940), a system interconnect (950), and a display (960). The processor (910) includes a CPU (Central Processing Unit), an MPU (Micro Processor Unit), an MCU (Micro Controller Unit), a GPU (Graphic Processing Unit), and an APU (Application Processing Unit).

[0098] The memory (920) interacts with the processor (910) to perform the function of storing data and enabling rapid access to necessary information so that the program can be executed efficiently. The memory (920) includes at least one of a register, a cache memory, a main memory, a read-only memory, a virtual memory, and a non-volatile memory.

[0099] Storage (930) serves to permanently store and manage data. Storage preserves data even after the computing system is turned off or rebooted and is used to store operating systems, applications, user files, etc. Storage (930) includes at least one of a hard disk drive (HDD), a solid-state drive (SSD), an optical disc, network storage, and cloud storage.

[0100] The communication interface (940) provides a path for exchanging data between various devices inside and outside the computing system. The communication interface (940) can support at least one of the following communication methods: USB (Universal Serial Bus), PCIe (Peripheral Component Interconnect Express), SATA (Serial ATA), Ethernet, Wi-Fi, Thunderbolt, and HDMI (High-Definition Multimedia Interface).

[0101] The system interconnect (950) serves to exchange data and signals between various components within the computing system. The system interconnect (950) can support at least one of a bus, point-to-point, crossbar switch, or network-on-chip (NoC) method.

[0102] The display (960) is an output device of the computing system and performs the function of providing visual information to the user.

[0103] According to the above configuration, the program according to the embodiment of the present invention is executed based on instructions executed by the processor (910) and can be stored in memory (920) or storage (930).

[0104] The method according to the embodiments of the present invention described above may be implemented in the form of program instructions that can be executed through various computer components and recorded on a computer-readable recording medium. The computer-readable recording medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the computer-readable recording medium may be specially designed and configured for the embodiments of the present invention, or may be known and available to a person skilled in the art of computer software. The computer-readable recording medium includes hardware configured to store and execute program instructions, such as magnetic recording media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; ROMs; RAMs; and flash memory. Program instructions include machine code generated by a compiler and high-level language code that can be executed on a computer using an interpreter. The hardware may be configured to operate as one or more software modules to process the method according to the present invention, and vice versa.

[0105] The method according to an embodiment of the present invention can be executed in the form of program instructions on an electronic device. The electronic device includes portable communication devices such as smartphones or smartpads, computer devices, portable multimedia devices, portable medical devices, cameras, wearable devices, and home appliances.

[0106] The method according to an embodiment of the present invention may be provided as included in a computer program product. The computer program product may be traded between a seller and a buyer as a product. The computer program product may be distributed in the form of a device-readable recording medium or online through an application store. In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily created in a storage medium such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0107] Each component, such as a module or a program, according to an embodiment of the present invention may be composed of a single or multiple sub-components, and some of these sub-components may be omitted or additional sub-components may be included. Some components (modules or programs) may be integrated into a single entity and may perform the functions performed by each corresponding component prior to integration in the same or similar manner. Operations performed by a module, program, or other component according to an embodiment of the present invention may be executed sequentially, in parallel, iteratively, or heuristically, or at least some operations may be executed in a different order, omitted, or additional operations may be added.

[0108] The foregoing description of the present invention is for illustrative purposes only, and those skilled in the art will understand that other specific forms can be easily modified without altering the technical spirit or essential features of the present invention. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. For example, each component described as a single unit may be implemented in a distributed manner, and components described as distributed may likewise be implemented in a combined form.

[0109] The scope of the present invention is defined by the claims set forth below, and all modifications or variations derived from the meaning and scope of the claims and equivalent concepts thereof should be interpreted as being included within the scope of the present invention.

[0110] The modes for carrying out the invention are described together in the best mode for carrying out the invention.

[0111] The artificial intelligence-based analysis system and method using automated feature extraction according to an embodiment of the present invention can automatically extract data features from a corporate database to simplify the analysis preparation process and automate the data analysis process based on artificial intelligence. In other words, it is possible to reduce the time required for corporate analysis, increase analysis efficiency and the accuracy of analysis results, and derive key analysis results for the company.

Claims

1. In artificial intelligence-based analysis systems, A data interpretation unit that classifies data stored in a database and analyzes the context; A feature extraction unit that extracts features of the data based on the analyzed context; A feature-based data structure reconstruction unit that reconstructs the database structure based on feature fields based on the features of the extracted data; An analysis model providing unit that provides or recommends an analysis model suitable for the characteristic field data of the above-mentioned reconstructed database structure; An analysis model application unit that generates results by applying the analysis model provided by the analysis model providing unit to the feature field data of the reconstructed database; and Includes an indicator derivation unit that analyzes the relationship between business objectives and the aforementioned feature field data, infers causal relationships, and selects goal-oriented features of the analysis model. AI-based analysis system.

2. In Paragraph 1, The above data interpretation unit is, Automatically collect the schema, table names, column names, comments, etc. of the above database and By analyzing metadata, the data semantics and interactions of the above database Characterized by understanding AI-based analysis system.

3. In Paragraph 2, The above data interpretation unit is, A data classification unit that automatically classifies the data types of the above database; A metadata analysis unit that identifies statistical characteristics by analyzing the data range and validity of the above-described classified data; NLP-based data analysis unit that establishes the meaning of data using natural language processing (NLP); and It includes a context analysis unit that analyzes the context of data relationships based on the above data type classification result, the above metadata analysis result, and the above NLP-based semantic analysis result. AI-based analysis system.

4. In Paragraph 3, The above data classification unit is, Characterized by classifying the above data types as numeric, categorical, time-related, spatial-related, financial-related, or ratio. AI-based analysis system.

5. In Paragraph 1, The above analysis model providing unit is, Knowledge graph analysis model; Graph neural network model; Automatic correlation analysis model; and It includes an analysis model derivation unit, and The above analysis model derivation unit provides an optimal analysis model by combining at least one of the knowledge graph analysis model, graph neural network model, and automatic correlation analysis model. AI-based analysis system.

6. In Paragraph 5, The above knowledge graph analysis model graphs the relationships between data to generate a graph structure that expresses the interactions and dependencies between data, and The above graph neural network model learns the graph structure and automatically discovers important combinations of data relationships and latent variables using a graph neural network (GNN). AI-based analysis system.

7. In Paragraph 6, The above knowledge graph analysis model is to add terms and rules of a specific industry sector to the graph structure in order to integrate the knowledge graph and domain knowledge. AI-based analysis system.

8. In Paragraph 5, The above automatic correlation analysis model is, Deriving correlations between data using quantitative calculation techniques that analyze correlation coefficients and statistical significance AI-based analysis system.

9. In Paragraph 1, The above indicator derivation unit is, Time indicator analysis unit that derives key points and periodic patterns from time-unit data; A data refinement unit that improves validity by refining data according to the characteristics of the data to be analyzed; A business goal linkage unit that sets and links business goals and analyzes the relationship between the said business goals and data characteristics; and It includes an indicator feature selection unit that selects important features for achieving the above business goals using a causal inference model. AI-based analysis system.

10. In the analysis method of an artificial intelligence-based analysis system, (a) A step of classifying data stored in a database and analyzing the context; (b) a step of extracting features of the data based on the analyzed context; (c) A step of reconstructing the database structure based on feature fields based on the characteristics of the extracted data above; (d) a step of providing or recommending an analysis model suitable for the characteristic field data of the reconstructed database structure above; (e) a step of generating results by applying the above analysis model to the feature field data of the above-reconstructed database; and (f) including the step of analyzing the relationship between business objectives and the aforementioned feature field data, inferring causal relationships, and selecting goal-oriented features of the analysis model. Analysis method of an artificial intelligence-based analysis system.

11. In Paragraph 10, The above step (a) is characterized by automatically collecting the schema, table names, column names, comments, etc. of the database and analyzing metadata to understand the data semantics and interactions of the database. Analysis method of an artificial intelligence-based analysis system.

12. In Paragraph 11, The above (a) step A step of automatically classifying the data types of the above database; A step of identifying statistical characteristics by analyzing the data range and validity of the above-described classified data; A step of establishing the meaning of data using natural language processing (NLP); and A step of analyzing the context of data relationships based on the above data type classification results, metadata analysis results, and NLP-based semantic analysis results. Analysis method of an artificial intelligence-based analysis system.

13. In Paragraph 12, The step of automatically classifying the data types of the above database Characterized by classifying the above data types as numeric, categorical, time-related, spatial-related, financial-related, or ratio. Analysis method of an artificial intelligence-based analysis system.

14. In Paragraph 10, The above step (d) is, Steps for modeling a knowledge graph analysis model; Steps for modeling a graph neural network model; Steps for modeling an automatic correlation analysis model; and It includes the step of deriving an analysis model, and The step of deriving the above analysis model is a step of providing an optimal analysis model by combining at least one of the knowledge graph analysis model, graph neural network model, and automatic correlation analysis model. Analysis method of an artificial intelligence-based analysis system.

15. In Paragraph 14, The step of modeling the above knowledge graph analysis model is a step of graphing the relationships between data to create a graph structure that expresses the interactions and dependencies between data, and The step of modeling the graph neural network model is a step of automatically discovering important combinations of data relationships and latent variables using a graph neural network (GNN) by learning the graph structure. Analysis method of an artificial intelligence-based analysis system.

16. In Paragraph 15, The step of modeling the above knowledge graph analysis model is to add terms and rules of a specific industry sector to the graph structure in order to integrate the knowledge graph and domain knowledge. Analysis method of an artificial intelligence-based analysis system.

17. In Paragraph 14, The step of modeling the above automatic correlation analysis model is a step of deriving correlations between data using quantitative calculation techniques that analyze correlation coefficients and statistical significance. Analysis method of an artificial intelligence-based analysis system.

18. In Paragraph 10, The above step (f) is a step of acquiring time-related data and decomposing it into time units; A step of analyzing time indicators by deriving key points and periodic patterns from the above time unit data; A step of improving validity by refining data according to the characteristics of the data to be analyzed; A step of establishing and linking business goals and analyzing the relationship between the said business goals and data characteristics; A step of linking data characteristics related to the above business objectives, and, It includes the step of selecting key features for achieving the above business objectives using a causal inference model. Analysis method of an artificial intelligence-based analysis system.

Images