KR20210042449A - Data scrapping apparatus and method - Google Patents

Abstract

Provided are a data scraping apparatus and a data scraping method, capable of providing any element information required by a financial institution as alternative information by collecting and analyzing data through an artificial intelligence scraping solution without separate scraping structure analysis for an unstructured region as well as structured open API scraping data. The provided apparatus includes: a retrieval unit for retrieving structured data and unstructured data matching input information from an outside based on a machine learning algorithm; a collection unit for scraping the retrieved structured data and the retrieved unstructured data from a scraping source based on the machine learning algorithm; an analysis unit for analyzing whether the structured data and the unstructured data from the collection unit are appropriate data based on the machine learning algorithm; and a matching unit for matching the structured data and the unstructured data from the analysis unit to a mapping table for each evaluation subject and for each financial institution.

Description

Data scrapping apparatus and method TECHNICAL FIELD

The present invention relates to a data scraping apparatus and method, and more particularly, to an artificial intelligence type big data scraping apparatus and method based on a machine learning algorithm.

The household loan market is a huge market used by 1,540 trillion won/2 million people based on the balance at the end of March 2019, but there are insufficient means to compare loans, so unfavorable loan selection due to information asymmetry, excessive profit deprivation by recruiters, incomplete sales, Information leakage has occurred and customer damage is rampant.

Traditionally, the financial licensing business, where entry barriers exist, has created stable profits with a loan-to-deposit structure, but due to the reduction of interest margins, restrictions on loan limits, and the entry of Internet banks and fintech companies, various customer services are strengthened and IT technologies are linked. There is a need to strengthen the non-face-to-face loan service.

However, since the credit rating model is different for each financial institution, there is a difficulty in scraping the rating information and data that are required in a non-face-to-face manner.

In addition, the customer has to visit financial institutions at least two times or more for real estate mortgage evaluation and submission of documents to be set.

Prior Art 1: Korean Patent Registration No. 10-2009336 (Cloud scraping system and method using pre-scraped big data, and computer program therefor) Prior Art 2: Korean Patent Laid-Open No. 10-2018-0093213 (Method of processing a home mortgage loan through a non-face-to-face channel)

The present invention has been proposed to solve the above-described conventional problems, and any element information required by financial institutions is not only formal open API scraping data, but also data through an artificial intelligent scraping solution without a separate scraping structure analysis for the unstructured area. The purpose of this is to provide a data scraping apparatus and method capable of collecting and analyzing data and providing it as alternative information.

In order to achieve the above object, a data scraping apparatus according to a preferred embodiment of the present invention includes: a search unit that searches for structured data and unstructured data corresponding to input information from the outside based on a machine learning algorithm; A collection unit for scraping the retrieved structured data and unstructured data from a scraping source based on a machine learning algorithm; An analysis unit that analyzes whether the formal data and the unstructured data from the collection unit are appropriate data based on a machine learning algorithm; And a matching unit for matching the formal data and the informal data from the analysis unit to the mapping tables for each evaluation subject and each financial institution.

When there is a scraping request from the outside, the search unit analyzes the evaluation item element based on the machine learning algorithm based on the input information from the outside, and uses the machine learning algorithm to determine what keyword corresponds to the analyzed evaluation item element. Based on the extraction, it is possible to identify which scraping source contains the structured or unstructured data that most matches the extracted keyword, and the structured data that most matches the keyword is converted from the identified scraping source to text or images. It is possible to search for existence based on a machine learning algorithm, and search for the existence of unstructured data that most closely matches a keyword as text or an image in the identified scraping source based on the machine learning algorithm.

The collection unit identifies the location, shape, and structure of the data in the scraping source searched for each of the structured data and the unstructured data, respectively, based on machine learning algorithms, and scrapes the identified structured and unstructured data. After collectively collecting from the source, it can be separated and extracted for analysis and processing based on machine learning algorithms.

The analysis unit learns the structured data and the unstructured data collected by the collection unit using a machine learning algorithm, and if there is incorrectly collected data, the analysis unit feeds back the facts to the collection unit so that they can be collected again.

The analysis unit may analyze adequacy, timing, and utility of the formal data and the unstructured data from the collection unit based on a machine learning algorithm.

The analysis unit may send data that satisfies all of the adequacy, timing, and utility to the matching unit.

The matching unit may store the matched result and output the matched result to a financial institution server.

On the other hand, the data scraping method according to the preferred embodiment of the present invention is a data scraping method performed in a data scraping device, which searches for structured data and unstructured data corresponding to input information from the outside based on a machine learning algorithm. step; Scraping the retrieved structured data and unstructured data in a scraping source based on a machine learning algorithm; Analyzing whether the scraped structured data and the unstructured data are appropriate data based on a machine learning algorithm; And matching the analyzed formal data and unstructured data to mapping tables for each evaluation subject and for each financial institution.

According to the present invention of this configuration, any element information required by financial institutions is collected and analyzed through an artificial intelligence scraping solution to collect and analyze the appropriate formal data and unstructured data, and alternative information to the evaluation message for each loan product desired by all financial institutions. Can provide.

In other words, using artificial intelligence (AI)-type machine learning algorithms for analysis of the holding institution, structure analysis, content analysis, and effectiveness analysis of various formal and informal evaluation element data required for loan application from various institutions with the consent of the customer. Therefore, it is possible to drastically reduce business processing time and cost by automatically matching scraping.

1 is a block diagram of a system to which the present invention is applied.
2 is an internal configuration diagram of the scraping server shown in FIG. 1.
3 is a flowchart illustrating a data scraping method according to an embodiment of the present invention.

In the present invention, various modifications may be made and various embodiments may be provided, and specific embodiments will be illustrated in the drawings and described in detail.

However, this is not intended to limit the present invention to a specific embodiment, it should be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the present invention.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In the present application, terms such as "comprise" or "have" are intended to designate the presence of features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, but one or more other features. It is to be understood that the presence or addition of elements or numbers, steps, actions, components, parts, or combinations thereof does not preclude in advance.

Unless otherwise defined, all terms used herein including technical or scientific terms have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and should not be interpreted as an ideal or excessively formal meaning unless explicitly defined in the present application. Does not.

Hereinafter, preferred embodiments of the present invention will be described in more detail with reference to the accompanying drawings. In describing the present invention, in order to facilitate an overall understanding, the same reference numerals are used for the same elements in the drawings, and duplicate descriptions for the same elements are omitted.

1 is a block diagram of a system to which the present invention is applied.

The system of FIG. 1 may include a customer terminal 10, a scraping server 20, a financial institution server 30, a scraping source 40, and a network 50.

The customer terminal 10 may access the financial company server 30 through the network 50 and input a loan request and details of a loan request for a loan application. At this time, the customer terminal 10 may also input detailed information related to the loan application. Here, the detailed information may be, for example, a name, a social security number, an email address, a home address, a home contact, a work address, a work contact, a mobile phone number, and the like.

The above-described loan request details for the loan application and detailed information related to the loan application may be the basis of the evaluation item elements required by the scraping server 20.

For example, the customer terminal 10 may be implemented as a portable terminal or a portable computer. Here, the portable terminal is a wireless communication device that guarantees portability and mobility, and includes a Personal Communication System (PCS), a Global System for Mobile communications (GSM), a Personal Digital Cellular (PDC), a Personal Handyphone System (PHS), a Personal Digital Digital Assistant (PDA). Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet), Wi-Fi, LTE ( All types of handheld-based wireless communication devices, such as a Long Term Evolution terminal, may be included. The portable computer may include a notebook, a laptop, and the like.

In addition, the customer terminal 10 may be various smart devices such as a smart phone, a smart note, a tablet PC, and a wearable computer.

The scraping server 20 artificially and intelligently scrapes a variety of formal data and unstructured data for evaluation of a corresponding customer in response to a scraping request from the financial company server 30 in the scraping source 40, and the corresponding financial company server 30 Alternative information (evaluation information) may be loaded on the full text of the evaluation and provided to the financial institution server 30.

That is, the scraping server 20 is an artificial intelligence (AI) type based on a machine learning algorithm, as well as scraping data using a formal open API for all loan evaluation item element information required by financial institutions. By collecting and analyzing data through data scraping, alternative information (evaluation information) can be provided to the evaluation message for each loan product desired for each financial institution.

In other words, the scraping server 20 is an artificial intelligence (AI) type based on a machine learning algorithm based on the formal evaluation information and the informal evaluation information for the evaluation of the customer in response to the scraping request from the financial institution server 30. By collecting and analyzing data from the scraping source 40 through data scraping, alternative information applicable to the evaluation items for each desired loan product for each financial institution may be generated and provided.

For example, it is assumed that Park, who works for Small and Medium Business A, applied for a non-face-to-face credit loan of 10 million won to raise money for her daughter's marriage. In this case, in the past, only the information on loan repayment capability (formal data; job, property, income status) was scraped through the Health Insurance Corporation and the National Tax Service to evaluate the loan repayment capability of Park. As a result of the evaluation, since the existing credit loan of 30 million won is being used for the 3rd grade of the credit rating, it is determined that it is not eligible for additional non-face-to-face credit loans and will not provide the loan.

However, in the above case, the scraping server 20 not only evaluates the above-described loan repayment ability evaluation information (formal data; property, income status, etc.), but also information (atypical data; workplace, personal interests, etc.) for evaluating the will to repay the loan. ) By scraping and analyzing the repayment ability and repayment will by multi-faceted evaluation, it is possible to determine whether a loan is possible, set a loan limit, and set a loan interest rate. For example, in the above unstructured data, the workplace may include information that the current workplace is classified as a SME, but has passed the preliminary examination for listing on the KOSDAQ due to the development of new medical drugs. In the above unstructured data, the field of personal interest may include information indicating that a person has a reputation as a major technology development manpower in the workplace, has five related qualifications, and is continuously studying at the same time.

In other words, Director Park, an applicant for a non-face-to-face credit loan of 10 million won, was not eligible for a loan due to insufficient credit rating and qualification requirements when evaluating an existing non-face-to-face loan application. And by comprehensively evaluating stable life and propensity to repay loans, it is possible to determine the possibility of an upgrade to the second grade of credit rating. Based on this information, the financial institution server 30 may approve an additional credit loan (eg, a loan limit of 10 million won + a loan increase approval decision of 4%).

As another example, it is assumed that Mr. Kim, who works for a large company B, applied for a 10 million won loan for a non-face-to-face credit due to excessive luxury and waste of entertainment expenses. In this case, in the past, it was subject to non-face-to-face loan approval only by evaluating credit ratings based on employment and income level.

However, in the above case, the scraping server 20 includes not only the above-described loan repayment ability evaluation information (formal data; employment, property, income status), but also information for evaluating the loan repayment intention (atypical data; personal life pattern, personal (Areas of interest, etc.) can be scraped to determine whether or not a loan is possible by multi-faceted evaluation of the repayment ability and repayment will. For example, the personal life pattern may include life information such as whether or not to comply with the Giro payment date, the number of automobile fines paid, and inquiries about court and police cases. The personal interest field may include propensity information such as activity records such as homepage and SNS, number of visits to bars, and usage history.

In other words, the scraping server 20 is a non-face-to-face credit loan applicant, Mr. Kim has excellent credit rating and repayment ability, but has a very strong desire to display luxury goods on social media, etc. As a person accused of fraud in the court, it can be judged as a target who has a high possibility of bad loans for the time being. Accordingly, based on this information, the financial institution server 30 may make a decision to approve a reduction in which the loan limit is reduced to 3 million won and the loan interest rate is 5%.

As another example, it is assumed that a 20-year-old detached house located in Osan, Gyeonggi-do that was recently remodeled (city price: 100 million won/priority 40 million won/two rooms) as collateral, additional subordinated real estate mortgage loan 30 million won non-face-to-face application is assumed. In this case, according to the existing collateral value evaluation method, when deducting the senior (40 million won) and small deposit (30 million won) from the 80% assessed value (80 million won), it will be possible to approve only the additional subordinated loan limit of 10 million won due to insufficient collateral limit.

However, in the above case, the scraping server 20 scrapes and repays the existing loan repayment capability information (formal data; collateral evaluation data) as well as information for evaluating the willingness to repay the loan (atypical data; value fluctuation evaluation data). The ability and willingness to repay can be evaluated and analyzed to determine whether a loan is possible or not. For example, the collateral evaluation data may include information on a nearby market price, a nearby appraisal price, and a resident registration copy. Value change evaluation data may include details of development goods (eg, Samsung employee housing complex) and changes in asset value (eg, remodeling) near collateral.

In other words, the scraping server 20 is an old general detached house with a brick-type slat roof structure, for study purposes, and has low collateral value and has little additional collateral capacity when deducting seniority and small deposits, but the area is inhabited by Samsung employees through artificial intelligence scraping. As a dedicated housing complex, there are recent developments in the vicinity, and the value is expected to change due to the remodeling and renovation of the property, so it can be judged that there is enough room for additional loans (that is, 30 million won).

As another example, a subordinated real estate lease in Seoul with 3 rooms (owner's address) as a shared apartment to 6 college students (deposit 10 million won each, 500,000 won monthly rent) as collateral (market price 200 million won, senior priority 80 million won). It is assumed that a non-face-to-face application for a mortgage loan of 30 million won is made. In this case, in the past, it will be judged as the owner's residence property, and the loan will be approved through a desk appraisal and a branch office conclusion only by evaluation such as LVT and DTI.

However, in the above case, the scraping server 20 scrapes and repays the existing loan repayment capability information (formal data; collateral evaluation data) as well as information for evaluating the willingness to repay the loan (atypical data; personal life pattern data). The ability and willingness to repay can be evaluated and analyzed to determine whether a loan is possible or not. For example, the collateral evaluation data may include information on a nearby market price, a nearby appraisal price, and a resident registration copy. Personal life pattern data may include all shopping mall parcel delivery destination input information and post office receipt information using the corresponding address.

In other words, the scraping server (20) is a residential apartment in which only the owner resides for study purposes, with an appraised value of 160 million won (80%) compared to the market price-80 million won for senior mortgage-a small lease deposit (three rooms in Seoul) of 45 million won. Even if deducted, subordinated security of 30 million won is possible, but it can be judged as an additional reduction in the value of the collateral by requesting a loan applicant to read the tenant list or taking additional measures to appraise the actual visit after identifying that it is a space where six others live in common. Therefore, in the above case, the loan limit can be set to about 10 million won.

The financial institution server 30 may communicate with the customer terminal 10 or the scraping server 20 through the network 50.

When the financial company server 30 receives the loan application from the customer terminal 10, the scraping server 20 sends the input information related to the loan application from the customer terminal 10 (including the details of the loan request and detailed information related to the loan application). ), and request data scraping for loan review evaluation.

The scraping source 40 includes a source capable of providing structured data and a source capable of providing unstructured data.

One or more sources of formal data can be provided, for example, from the registry office, the IRS, the Health Insurance Corporation, the financial institution, the CB/NICE, the Supreme Court, or a public institution.

Sources that can provide atypical data may be, for example, one or more from SNS, real estate brokerage firms, appraisal firms, civil petitions24, telecommunications companies, VANs, loan companies, religious organizations, shopping malls, schools, and workplaces.

Of course, the source that can provide the structured data and the source that can provide the unstructured data are not limited to the ones exemplified above, but can of course be added.

The network 50 facilitates communication between the customer terminal 10, the scraping server 20, the financial institution server 30, and the scraping source 40.

For example, the network 50 may be implemented as a wired network such as a local area network (LAN), a wide area network (WAN), or a value added network (VAN).

In addition, the network 50 is a mobile radio communication network, satellite communication network, Bluetooth, Wibro (Wireless Broadband Internet), HSDPA (High Speed Downlink Packet Access), WiFi (Wi-Fi), LTE (Long Term Evolution) can be implemented in all types of wireless networks.

If necessary, the network 50 may be a network in which wired and wireless are mixed.

FIG. 2 is an internal configuration diagram of the scraping server shown in FIG. 1.

The scraping server 20 may include a search unit 21, a collection unit 22, an analysis unit 23, and a matching unit 24. The scraping server 20 may be referred to as a data scraping device.

The search unit 21 analyzes the evaluation item element based on the input information from the financial company server 30, automatically extracts (analyzes) a keyword corresponding to the evaluation item element, and provides the most data in accordance with the keyword. Text and images including data may be searched from the possible scraping source 40 and the scraping source 40. Here, the input information from the financial company server 30 will be the details of the loan request and detailed information related to the loan application (e.g., name, social security number, email address, home address, home contact, work address, work contact, mobile phone number, etc.). I can.

The search performed by the search unit 21 is performed based on a machine learning algorithm.

In other words, when there is a scraping request from the financial institution server 30, the search unit 21 automatically analyzes (extracts) the evaluation item element based on the machine learning algorithm based on the input information from the financial institution server 30. Here, the elements of evaluation items are job group, name of work, date of employment, annual income, real estate (apartment, house, etc.) address, real estate market price, delinquency propensity, willingness to repay, compliance level, interests, religious orientation, education level, career level, It can be an increase in income, a decrease in income, an increase in collateral, a decrease in collateral, the status of leases, social media tendencies, hobbies, and specialties.

In addition, the search unit 21 automatically extracts (analyzes) what keywords corresponding to the analyzed evaluation item elements are based on a machine learning algorithm.

In addition, the search unit 21 determines which scraping source contains the most formal data corresponding to the extracted keyword, and identifies which scraping source contains the unstructured data most most consistent with the extracted keyword.

The structured data and the unstructured data may be text or images in the corresponding scraping source. Accordingly, the search unit 21 searches for whether the structured data that most closely matches the keyword exists as text or image in the identified scraping source, based on a machine learning algorithm. In addition, the search unit 21 searches whether or not the unstructured data that most closely matches the keyword exists as text or image in the identified scraping source, based on a machine learning algorithm.

The collection unit 22 analyzes the structure of the structured data and the unstructured data searched by the search unit 21 based on a machine learning algorithm, and then scrapes the data in the scraping source 40.

That is, the collection unit 22 scrapes the retrieved structured data and unstructured data in any format, whether in text or images.

In other words, the collection unit 22 grasps (analyzes) the position, shape, and structure in the scraping source searched for each of the structured data and the unstructured data based on a machine learning algorithm.

In addition, the collection unit 22 collectively collects the identified formal data and unstructured data from the corresponding scraping source 40, separates and extracts them, and analyzes and processes them based on a machine learning algorithm.

The analysis unit 23 analyzes whether the data from the collection unit 22 (ie, structured data or unstructured data) is appropriate data based on a machine learning algorithm.

In other words, the analysis unit 23 analyzes the adequacy, timing, and utility of the formal data and the informal data from the collection unit 22 based on a machine learning algorithm. Here, the adequacy may mean that the data must conform to the elements of the evaluation item. Timing may mean that the data should be the current time point or the most recently extracted data. Usability can mean that the data must be data that can be used by several financial institutions.

In particular, the analysis unit 23 learns the data (i.e., formal data, unstructured data) collected by the collection unit 22 with a machine learning algorithm, and if there is incorrectly collected data, the analysis unit 23 informs the collection unit 22 of the fact. Give feedback so you can collect it again.

In addition, the analysis unit 23 sends data that satisfies all appropriateness, timing, and utility to the matching unit 24.

The matching unit 24 matches the formal data and the informal data from the analysis unit 23 to each evaluation subject (loan subject) and a mapping table for each financial institution. Here, the mapping table for each evaluation subject and for each financial institution is, for example, social security number, name, email address, home address, home contact number, work address, work contact number, mobile phone number, occupation group, company name, date of employment, annual income, real estate holding address, Information such as market price, delinquency disposition, repayment will, compliance level, interests, religious disposition, education level, career level, income increase, income decrease, mortgage increase, mortgage decrease, lease status, SNS disposition, hobbies, specialty, etc. Can include.

The matching unit 24 may store the result of matching. In addition, the matching unit 24 may transmit the matching result to the financial institution server 30 through the network 50.

For example, if the lease status is an evaluation item, the scraping server 20 automatically recognizes the keyword for the delivery and mail receipt information for the corresponding address, and data that can identify the actual tenant tenant information in a shopping mall or post office, etc. , Formal data, and unstructured data) can be automatically analyzed, the location, shape, and structure of the corresponding data, extracted and separated, and matched with the evaluation items of financial institutions and provided.

In addition, for example, the scraping server 20 analyzes and utilizes AI scraping data for a person who is currently receiving a loan interest rate corresponding to the second grade of credit, and determines the loan interest rate and the limit corresponding to the first grade according to the improvement of the credit grade. Comparison information can be provided so that it can be applied.

On the other hand, in the case of real estate appraisal price evaluation, the scraping server 20 includes the surrounding market price, the past appraised price of the same kind, the auction price of the nearby similar property, the past appraisal of the appraisal firm, the past of the financial institution itself, the brokerage price of the nearby real estate agency, and the neighborhood. New pre-sale prices, etc. can be analyzed and reflected as formal and informal data for calculating appraised prices, and provided as alternative information.

In addition, in the case of real estate rights analysis verification, the scraping server 20 includes details of the establishment of root mortgage rights, superficies, lease rights, etc., auction commencement decision, seizure decision, delivery order decision details, actual resident information, actual lease status information, double sales contract. Etc. can be provided as alternative information by analyzing and reflecting them as formal and informal data for understanding rights analysis.

In addition, in the case of real estate value fluctuation calculation, the scraping server 20 has the latest extension or renovation status details, recent repair details, repair details, recent fire/inundation/loss/hand damage status details, favorable development in nearby areas, traffic conditions in nearby areas. Etc. can be provided as alternative information by analyzing and reflecting them as formal and unstructured data for understanding future value fluctuation factors.

According to the embodiment of the present invention as described above, it is possible to design a Smart Meta database capable of accommodating and responding to all evaluation items and formats required by credit rating information system items operated differently for each financial company. .

Meanwhile, the optimal loan conditions are searched by collecting and analyzing formal and unstructured data for credit loans and mortgage evaluation of individuals or companies through artificial intelligence (AI) data scraping based on machine learning algorithms. And make it possible to apply for a non-face-to-face loan

In addition, since all information items and formats constructed differently for each data holding institution can be accommodated, optimal alternative information can be provided.

3 is a flowchart illustrating a data scraping method according to an embodiment of the present invention.

First, the customer terminal 10 may access the financial company server 30 through the network 50 to input a loan request and a loan request details for the loan application. At this time, the customer terminal 10 may input detailed information related to the loan application (eg, name, social security number, email address, home address, home contact, work address, work contact, mobile phone number, etc.).

Accordingly, when receiving the loan application from the customer terminal 10, the financial company server 30 sends the scraping server 20 input information related to the loan application from the customer terminal 10 (the details of the loan request and the loan application Provides detailed information) and requests data scraping for loan review evaluation.

As the scraping server 20 receives a data scraping request for loan review evaluation from the financial institution server 30 (S10), the scraping server 20 is an evaluation item element based on the input information from the financial institution server 30 Is automatically analyzed based on machine learning algorithms, and automatically extracts what keywords corresponding to the analyzed evaluation item elements are based on machine learning algorithms, and formal data and unstructured data most consistent with the extracted keywords Figure out which scraping source is in. In addition, the scraping server 20 searches for whether the structured data that most closely matches the keyword exists as text or image in the identified scraping source, based on a machine learning algorithm. In addition, the scraping server 20 searches based on a machine learning algorithm whether the unstructured data that most closely matches the keyword exists as text or an image in the identified scraping source (S12).

Subsequently, the scraping server 20 grasps (analyzes) the location, shape, and structure in the searched scraping source for each of the structured data and the unstructured data based on machine learning algorithms, and identifies the structured data and the unstructured data. After collectively collecting enemy data from the corresponding scraping source 40, it is separated and extracted, and analyzed and processed based on a machine learning algorithm (S14).

Subsequently, the scraping server 20 analyzes whether the data from the collection unit 22 (ie, structured data or unstructured data) is appropriate data based on a machine learning algorithm. That is, the scraping server 20 analyzes the adequacy, timing, and utility of the formal data and the unstructured data from the collection unit 22 based on the machine learning algorithm (S16).

In addition, the scraping server 20 matches data that satisfies all appropriateness, timing, and usability (ie, formal data and unstructured data) with mapping tables for each loan subject and for each financial institution (S18).

Thereafter, the scraping server 20 stores the matched result and transmits the matched result to the financial institution server 30 through the network 50 (S20).

In addition, the data scraping method of the present invention described above can be implemented as a computer-readable code on a computer-readable recording medium. The computer-readable recording medium includes all types of recording devices that store data that can be read by a computer system. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tapes, floppy disks, and optical data storage devices. In addition, the computer-readable recording medium is distributed over a computer system connected through a network, so that computer-readable codes can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the method can be easily inferred by programmers in the art to which the present invention pertains.

As described above, an optimal embodiment has been disclosed in the drawings and specifications. Although specific terms have been used herein, these are only used for the purpose of describing the present invention, and are not used to limit the meaning or the scope of the present invention described in the claims. Therefore, those of ordinary skill in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical scope of the present invention should be determined by the technical spirit of the appended claims.

10: customer terminal 20: scraping server
21: search unit 22: collection unit
23: analysis unit 24: matching unit
30: financial company server 40: scraping source
50: network

Claims (8)

A search unit that searches for structured data and unstructured data corresponding to input information from the outside based on a machine learning algorithm;
A collection unit for scraping the retrieved structured data and unstructured data from a scraping source based on a machine learning algorithm;
An analysis unit that analyzes whether the formal data and the unstructured data from the collection unit are appropriate data based on a machine learning algorithm; And
And a matching unit for matching the structured data and the unstructured data from the analysis unit to a mapping table for each evaluation subject and each financial institution. The method according to claim 1,
The search unit,
When there is a scraping request from the outside, the evaluation item element is analyzed based on the machine learning algorithm based on the input information from the outside, and the keyword corresponding to the analyzed evaluation item element is extracted based on the machine learning algorithm. , Machine learning to determine which scraping source has the most structured or unstructured data matching the extracted keyword, and whether the structured data most matching the keyword exists as text or image in the identified scraping source. A data scraping apparatus comprising: searching based on an algorithm, and searching based on a machine learning algorithm whether or not unstructured data most likely matching a keyword exists as text or image in the identified scraping source. The method according to claim 1,
The collection unit,
For each of the structured and unstructured data, the location, shape, and structure of the data in the searched scraping source are identified based on machine learning algorithms, and the identified structured and unstructured data are collectively collected from the corresponding scraping source. A data scraping device, characterized in that for analyzing and processing based on a machine learning algorithm by separating and extracting the data. The method according to claim 1,
The analysis unit,
The data scraping device, characterized in that by learning the structured data and the unstructured data collected by the collection unit with a machine learning algorithm, and if there is incorrectly collected data, the information is fed back to the collection unit so that the data can be collected again. The method according to claim 1,
The analysis unit,
A data scraping device, characterized in that analyzing the appropriateness, timing, and usability of the structured and unstructured data from the collection unit based on a machine learning algorithm. The method of claim 5,
The analysis unit,
Data scraping apparatus, characterized in that sending data that satisfies all of the adequacy, timing, and utility to the matching unit. The method according to claim 1,
The matching unit,
A data scraping device, characterized in that storing the matched result and outputting the matched result to a financial institution server. As a data scraping method performed in a data scraping device,
Searching for structured data and unstructured data corresponding to input information from the outside based on a machine learning algorithm;
Scraping the retrieved structured data and unstructured data in a scraping source based on a machine learning algorithm;
Analyzing whether the scraped structured data and the unstructured data are appropriate data based on a machine learning algorithm; And
And matching the analyzed formal data and the unstructured data to a mapping table for each evaluation subject and each financial institution.

Images