WO2019021312A1

WO2019021312A1 - An automated system for default probability prediction of loans and method thereof

Info

Publication number: WO2019021312A1
Application number: PCT/IN2018/050486
Authority: WO
Inventors: Aviruk CHAKRABORTY
Original assignee: Capitaworld Platform Private Limited
Priority date: 2017-07-26
Filing date: 2018-07-25
Publication date: 2019-01-31
Also published as: ZA202001132B

Abstract

The present invention discloses an automated system for default probability prediction and method for foreseeing default probability prediction for non-performing assets and derelict properties comprises various system components such as host computing system (13) consists of high-performance processor-based system, computer accelerator having multiple specialized computers (15) and multiple re-configurable computers (16), and memory (14). The loan application is filled by the loan applicant via smart contract; relevant data is collected and validated for safety measures. The Sham Dearth Score then generated to identify the chances of delinquency. Further, Confederate Semantic Knowledge Engineering Model fetches previous data and provides outstanding results for recognizing plausible delinquent. Later on, the smart contract is generated for loan disbursement and the system of the present invention then monitors the disbursed loan for identifying defaulters, insolvency, delinquency, non-performing assets, and derelict properties. Afterwards, the red flag alerts have been sent to such delinquent cases to overcome financial risks.

Description

AN AUTOMATED SYSTEM FOR DEFAULT PROBABILITY PREDICTION OF LOANS AND METHOD THEREOF

Field of the invention

The present invention relates to an automated system for default probability prediction of loans and method and more particularly it relates to an automated system for default probability prediction of loans and method for foreseeing the default probability prediction for non-performing assets and/or derelict properties using amalgamation of hardware and software which helps out banks to identify the cases of delinquency and insolvency.

Background of the invention The traditional or conventional method for identifying and detecting the cases for insolvency, delinquency, non-performing assets, derelict properties, and bankruptcy is very time consuming and taking years to find out the defaults. These traditional methods leading towards debts and the economy of the country or world has been affected by these conventional systems. These methods also require more human resources, more time and loads of paperwork. The Traditional systems are Human oriented/driven and have high probability of bias and/or human errors. This can be highly misdirecting and becomes an incompetent process overall. Hence, a cost efficient system with the compact, automated and speedier solution is extremely needed in order to eradicate and reduce this delinquency problem hindering the global growth.

Various prior arts have been disclosed describing the detection of cases of the insolvency, delinquency, non-performing assets, derelict properties, and bankruptcy. The prior art document "Ensemble Neural Network Strategy for Predicting Credit Default Evaluation" by A. R. Ghatge, & P. P. Halkarnikar discloses the theory of artificial neural networks and business rules to correctly determine whether a customer is a default or not. The Feed- forward back propagation neural network is used to predict the credit default. The results of applying the artificial neural networks methodology to classify credit risk based upon selected parameters show abilities of the network to learn the patterns. Another prior art document "Machine Learning Application in Online

Leading Credit Risk Prediction" by Xiaojiao Yu describes the risk prediction and monitoring for the success of the business model. In this article, data has been collected with various format and size from a public website, third- parties and assembled with a client's loan application information data. Ensemble machine learning models, random forest model, and XGBoost model, have been built and trained with the historical transaction data and subsequently tested with separate data, and all of these factors are important in predicting loan default probability. Similarly, further prior art document "Loan Approval Prediction based on Machine Learning Approach" by Kumar Arun, Garg Ishan, and Kaur Sanmeet describes the system to predict whether the particular applicant is safe or not and the whole process of validation of features is automated by machine learning technique and is exclusively for the managing authority of Bank/finance company, whole process of prediction is done privately and no stakeholders would be able to alter the processing. Result against particular Loan Id can be sent to various departments of banks so that they can take appropriate action on the application. This helps all others department to carry out other formalities.

Moreover, patent document US8073764B2 describes the computer- implemented method of managing the risk associated with a portfolio, where the portfolio comprising a plurality of positions, each of the plurality of positions being associated with at least one product traded on an exchange. And, patent document US20070192241A1 describes the methods and systems for providing a unique and alternative computing platform operable to address the needs of high- performance computing areas, where financial industry is just one example of such a field where there exists such a need for high performance-computing.

Above mentioned prior arts do not apparently describe disruptive advanced technology- based automated & transparent approach by relevant financial data stored in a data lake, with clarity regarding Decentralized Autonomous Organization (DAO) and Blockchain Technology (BCT), especially for predictions of frauds, insolvency, and delinquency in the loan process.

Therefore, it would be highly desirable to have a completely digitalized and automated loan processing system, which provides clarity regarding stressed loans & accounts, before early signals or prior to special mention accounts, hence to identify insolvency, delinquency, non-performing assets, derelict properties and bankruptcy, and the system which reduces documentation, complex paperwork, and tiresome process.

Object of the invention

The main object of the present invention is to provide an automated system for default probability prediction of loans and method for non- performing assets and derelict properties.

Another object of the present invention is to provide an automated system for default probability prediction of loans and method for non- performing assets and derelict properties with the completely digitalized approach.

Yet another object of the present invention is to provide an automated system for default probability prediction of loans and method for non- performing assets and derelict properties which reduce documentation, complex paperwork, and tiresome process. Still another object of the present invention is to provide an automated system for default probability prediction of loans and method for non- performing assets and derelict properties by pinpointing stressed loans & accounts before early signals or prior to special mention accounts.

The further object of the present invention is to provide an automated system for default probability prediction of loans and method for non- performing assets and derelict properties which having data lake generated with the help of data transmission and creates an autonomous approach. Summary of the Invention

The present invention discloses an automated system for default probability prediction of loans and method for foreseeing the default probability prediction for non-performing assets and/or derelict properties using an amalgamation of hardware and software which helps out banks to identify such cases of delinquency and insolvency with the completely digitalized approach to ease out the loan & timely recovery action oriented process. The present invention comprises various system components such as a host computing system which consists of high- performance processor-based system, a computer accelerator having multiple specialized computers and multiple re-configurable computers, and a memory. The loan application is filled by a loan applicant via smart contract; data has been obtained from the generated data lake and data validation is performed for all key data points once scrutinization is completed. On the basis of these data, Sham Dearth Score has been generated for figuring out whether there are chances of the applicant to be delinquent or not, and helps in recognizing provincially paramount to minimal probable cases. Further, the Confederate Semantic Knowledge Engineering Model is built which works upon the specially designed heterogeneous hardware and software would fetch the previous data and provides outstanding results for identifying and recognizing plausible delinquent. Once this process is getting completed, the system of the present invention automates the plausible cases and provides the finest optimal output. Later on, a smart contract is generated for that outputted result with details on the basis of Blockchain Technology (BCT). In the same way, everything is performed by the automation process. The system of the present invention then monitors, tracks, and evaluates the disbursed loan for identifying likely defaulters, insolvency, delinquency, non-performing assets, and derelict properties.

Brief Description of the Drawings Fig. 1 is a flow diagram showing the main operational steps of an automated system for default probability prediction of loans.

Fig. 2 is a block diagram showing the main operational components of an automated system for default probability prediction of loans. Detailed description of the Invention

The nature of the invention and the manner in which it works is clearly described in the complete specification. The invention has various elements and they are clearly described in the following pages of the complete specification. Before explaining the present invention, it is to be understood that the invention is not limited in its application.

An automated system for default probability prediction of loans and method according to the present invention comprise various sequences of steps to accomplish the objective of the present invention. Figure 1 & 2 show the process flow and components of the completely automated loan processing system of the present invention. As shown in figures 1 and 2, at an input step 1 in starting with the process, a loan applicant or loan borrower comes and fills up the details regarding specific loan amount, loan type etc. These data are stored for further processing with a data lake, where the data lake has been generated with the help of a data transmission which is described in the next input step 2. Here, at input step 1, the loan application is filled via Blockchain Technology (BCT).

At the input step 2, the automated loan processing system of the present invention obtains the applicant's data to create data lake from all relevant sources such as financial data lake, e-KYC, digital footprints, network activities, authenticated government databases, virtual networks, credit card/debit card history, bank transactions, bank statements, deep web search and image recognition. Said the data lake comprises all the relevant data or information of a person such as historical data, real-time data, biometric data etc. A host computing system (13) of the present invention collects the data. Hence, with the help of the host computing system (13), parallel processing of the system is working and the process of automation with said data has been done with all relevant data feeds. The data is then computed and scrutinized via Blockchain Technology (BCT) by the information provided by the applicant or borrower as well as the information obtained by the automated loan processing system of the present invention.

At step 3, the data validation is carried out by Blockchain Technology (BCT) after the scrutinization once completed, where said data has been validated for the all key data points. The automated loan processing system of the present invention performs the data validation step for the safety measures. At next step 4, for the purpose of keeping data secured and intact, a matrix or a huge array of the highly computational system has been operated by an automated loan processing system of the present invention. In this particular system, the host computing system (13) consists of a high- performance processor-based system. In order to enhance the speed of computation in terms of parallelism, the system of the present invention has interfaced a computation accelerator (15, 16). Here, the computation accelerators (15, 16) are multiple of a specialized computer (15) and multiple of a re-configurable computer (16). Further, the specialized computer units (15) are used for carrying out parallel processing on tasks from the host unit (13). Similarly, the re-configurable computers units (16) are used for carrying out the serial or parallel processing on the tasks from the host unit (13). The host unit (13) is used for managing resources and allocating functional tasks to the specialized computer units (15) and/or the re-configurable computers units (16). The whole physical structure is like matrices of the multiple specialized computers (15) and the multiple re-configurable computers (16) according to the requirement of the system of the present invention. Here, a memory (14) for said data is shared between the specialized computers (15) and the re-configurable computers (16) and, this memory management has been done by the host computing machine (13).

At step 5, on the basis of said data scrutinization and data validation as described in the abovementioned steps, a uniquely contrived rate or score has been calculated. This rate or score is calculated on the basis of en masse models which comprises of semantic network models and intelligent retrieval breakthrough method. The score which is generated is called Sham Dearth Rate/Score. Later, on the basis of this said score generation, a high and low range is set for the easy classification. This rate or score would be helpful in recognizing provincially paramount to minimal probable cases. Moreover, this is not the only measurable points for the said classification, as the system of the present invention can take into consideration the behavioral analysis, credit card or debit card history, bank statements, bank transactions, digital footprints, deep web search, virtual network activities, e-KYC and image recognition. Hence, aforesaid germane points guide to identify any plausible odds for fraud, malicious activity, and delinquency. At this step, if the Sham Dearth Score has not been generated, then the system of the present invention has to follow the whole process from initial input step 2 to fourth step 4. If the system of the present invention successfully completes the step 5 of Sham Dearth Score generation, then the loan processing system can proceed towards the further steps of the present invention.

After said score or rate has been generated, at further step 6, the heterogeneous computation i.e. the amalgamation of hardware and software would perform the vital functions for the further proceeding of the system of the present invention. It can work individually on any system or process through the model with uniquely designed in-built software. Here in this step, the previous data has been fetched by the said hardware-software amalgamation of the present invention and execute the functional process to update the data for the model. Generally, this process takes a larger amount of time with the conventional computing system. Whereas, in the proposed heterogeneous computation, the system of the present invention exploiting the advancement of integration of the specialized computer (15) and the re- configurable computer (16) with a processor-based system. The capabilities of the specialized computer (15) and the re-configurable computer (16) for parallel computing take lesser time in classification, clustering, forecasting, predicting and identifying the delinquency rate. At step 7, the anchor of the system of the present invention i.e. a model called Confederate Semantic Knowledge Engineering Model has been run by banks or banking organizations which gives the optimal results for identifying and recognizing plausible delinquent. This model has been structured mainly focusing on pinpointing the stressed loans & accounts before early signals or prior to special mention accounts. This model basically works upon the specially designed heterogeneous hardware and software which is described in abovementioned steps. With the help of this model, which is an amalgamation of hardware and software, and performs a necessary function in detecting and identifying non-performing assets and derelict properties, one of the major cause can be eliminated which is hampering the global growth. This model deprecates the risk of default. So, the banks or banking organizations can easily identify and work upon the recovering and repayment of loans. This way, the hard-earned money of people is saved which lessens the inflation rate. The proposed model of the system of the present invention does not only reduce the time, but is also cost effective & relatively more transparent in the sense that human error/bias are minimized as it helps out the banks or banking organization to identify such cases, which is very costly in the existing systems. Furthermore, this model also aids into low human resources and complex paperwork. At further step 8, on the basis of the Confederate Semantic Knowledge Engineering Model as described in above-mentioned step 7, the feasible solution has been identified which having finest optimal output. The process after giving the output reaches for any case in later stages as it stores the data and has already classified them accordingly. At this step 8, if the feasible solution has not been identified, then the system of the present invention has to repeat the previous step 6 of heterogeneous computation, and then the step 7 of Confederate Semantic Knowledge Engineering Model, on the basis of which optimal result has been obtained to identify the feasible solution. If the system of the present invention successfully completes the step 8 of identifying the feasible solution, then the loan processing system can proceed towards the further steps of the present invention. At step 9, after identifying the best solution, a smart contract or a digital contract has been generated for the identified results with details. Here, a smart contract or a digital contract has been generated on the basis of disruptive technology like Blockchain Technology (BCT). Further at step 10, after completing the all previous processes, the loan has been disbursed to the selected applicants, where the applicants have been selected on the basis of output provided by Confederate Semantic Knowledge Engineering Model. The system of present invention is not limited to the loan disbursement as the system also monitors, tracks, and evaluates the disbursed loan for identifying defaulters or any early signals, where the loan monitoring, loan tracking, and loan valuation have been carried out post disbursement with the help of the disruptive technology like Decentralized Autonomous Organization (DAO).

At step 11a, the automated loan process is completed and successful if the loan taken by the person repays the loan amount on time according to the post disbursement loan tracking records. Further at step lib, according to the post disbursement loan tracking records, the despiteful applicants have been identified and early warnings are sent via Blockchain Technology (BCT) and Decentralized Autonomous Organization (DAO).

At final step 12, the system of the present invention check outs the delinquency cases and sends the red flag alert to the respective delinquency cases to overcome the financial risks.

The system of the present invention is advantageous over the existing loan processing system. The basic concept of the automated loan processing system of the present invention is to work upon individually for any loans such as car loan, home loan, personal loan, business loan, working capital loan, education loan, and term loan to identify the feasible cases of delinquency. As the system of the present invention implementing the merging of hardware & software elements, it can work individually on any system through the model with uniquely designed in-built software. Hence, there is no need of any cloud network architecture. The system of the present invention is completely digitalized so it gives early signals which helps in recognizing any deteriorating asset quality and focuses solely on risk management. The present system comprises all uniquely designed semantic intelligent retrieval models for calculation and/or formulation. The system of the present invention is cost- efficient and helps in having a compact solution for the delinquency problem hindering any country's growth and/or world's growth. Hence, the proposed loan processing system is faster and smoother than the current loan processing system. While various elements of the present invention have been described in detail, it is apparent that modification and adaptation of those elements will occur to those skilled in the art. It is expressly understood, however, that such modifications and adaptations are within the spirit and scope of the present invention as set forth in the following claims.

Claims

im:

A method of an automated default probability prediction of loans, comprising the steps of: a) filling of a loan application by a person or an applicant via Blockchain Technology (BCT); b) fetching of an applicant data to create data lake from all relevant sources by host computing system (13); c) scrutinizing and validating the said applicant data by Blockchain Technology (BCT) for safety measures; d) securing the said applicant data by running matrix or huge array of the highly computational system; e) generating a Sham Dearth Score on the basis of said data scrutiny and data validation; f) classifying and clustering the delinquency rate by implementing heterogeneous computation; g) running Confederate Semantic Knowledge Engineering Model generated from the heterogeneous computation to identify feasible solution; h) generating a smart or digital contract for loan disbursement on the basis of Blockchain Technology (BCT); i) monitoring, and tracking the loan after loan disbursement with; and j) sending the red flags and early warnings by Blockchain Technology (BCT) and Decentralized Autonomous Organization (DAO) when delinquent cases have been noticed.

The method of an automated default probability prediction of loans as claimed in claim 1, wherein said applicant data includes historical data, real-time data, and biometric data.

The method of an automated default probability prediction of loans as claimed in claim 1, wherein said relevant sources comprises the financial data lake, e-KYC, digital footprints, network activities, authenticated government databases, virtual networks, credit card/debit card history, bank transactions, bank statements, deep web search and image recognition. The method of an automated default probability prediction of loans as claimed in claim 1, wherein said Sham Dearth Score is generated for figuring out the chances of the applicant to be delinquent or not, and helps in recognizing provincially paramount to minimal probable cases.

The method of an automated default probability prediction of loans as claimed in claim 1, wherein said heterogeneous computation is an amalgamation of hardware and software having integration of the specialized computer (15) and the re-configurable computer (16) with a processor-based system.

The method of an automated default probability prediction of loans as claimed in claim 1, wherein said monitoring, and tracking of loan is done with the Decentralized Autonomous Organization (DAO).

An automated system for default probability prediction of loans, comprising:

a host computing system (13) consists of high- performance processor- based system;

a computer accelerator having multiple specialized computers (15) and multiple re-configurable computers (16); and

a memory (14) to share the said data between the specialized computers (15) and the re-configurable computers (16). A method of an automated default probability prediction of configured to operate the steps as claimed in claims 1 - 7.