A GAMIFIED REVISION AND AI-POWERED DELAY PREDICTION SYSTEM FOR HIGH AND LOW CREDIT RISK CREDIT PROFILES.

Abstract

This invention relates to a system (1) that combines a gamified repayment revision for customers with different credit profiles with an AI-powered prediction mechanism, enabling task-based credit score improvement processes for customers with high credit risk and reward-focused advantages for customers with low credit risk and reliability.

Description

1 TARIFF HIGH CREDIT RISK AND LOW CREDIT RISK CREDIT PROFILES GAMIFIED REVISION AND AI-POWERED FOR DELAY PREDICTION SYSTEM Technical Area This invention provides a gamified return system for customers with different credit profiles. A combination of payment revision and an AI-powered prediction mechanism. by introducing task-based credit score improvement for customers with high credit risk. 10 providing processes and rewarding low-risk and reliable customers. It relates to a system that enables the provision of focused advantages. Previous Technique In the known state of the art, credit risk estimation and performance are available in current financial technologies. Customer management relies primarily on fixed data and traditional scoring. It has remained limited to these models. Credit bureaus and banks assess customers' credit risk. When determining this, factors such as past payment history, current debt status, and income level are considered. It uses statistical models based on static variables. These systems 20 usually logistic regression, linear discriminant analysis or simple decision It is based on traditional methods in the form of trees, customer from instantly and proactively capturing dynamic changes in their behavior It is far away. Thus, risk assessment is often a passive process based on historical data. It remains limited to the forecasting process, adapting to real-time changes 25 It is unable to provide this. In recent years, the financial technology sector has been facing increased credit risk. It has begun integrating artificial intelligence and machine learning techniques into its analytics; especially decision trees, random forest, XGBoost (eXtreme Gradient Boosting- Default prediction using Extreme Gradient Boosting and artificial neural networks Significant improvements in accuracy have been achieved. For example, AI-based programs like "Upstart" have been used for 30 years. Lending platforms are modeling over a hundred variables, surpassing traditional methods. 2 Compared to other methods, they have reported default rate improvements of up to 16%. In this way, companies like "HighRadius" analyze a customer's financial behavior. Detecting changes in real time and providing credit teams with immediate risk information. They have developed systems that send warnings. However, these systems are mostly background checks. The analytical infrastructures working in the plan, and the interaction with the end user (customer) is 5 It does not establish. Credit risk systems developed to date, user It has lacked proactive intervention mechanisms for its behavior. Current practices mostly update the customer's risk score silently, It takes action when the customer defaults. However, early intervention... It is a critical requirement to be able to improve the customer's risk profile in a timely manner. 10 These systems do not offer the customer opportunities to change the risk situation. a missed potential for both financial institutions and borrowers. It represents this. In addition, gamification in financial services. (Gamification) applications have been used in recent years to increase customer interaction. It has started to be used. 15 in the form of points, badges, levels, achievement rewards. Game mechanics elements have been adapted to the use of financial products. For example, some Banks use badges to encourage regular saving in savings accounts. Some mobile applications offer users the opportunity to set their spending goals. It offers virtual rewards if they accomplish these things. In the US, there's a guy named "Lenny". A financial technology startup is helping young users in the credit score building process. integrated gamified tasks; users who pay on time This allowed him to earn "Lenny Points," which in turn increased his credit limits. However, current gamification applications have several serious shortcomings: It is generally associated with financial behavior at a superficial level; for example, saving. It rewards doing so, but it also increases the risk of loan deferral or the dynamics of financial distress. 25 It cannot model it. It is not strongly integrated with artificial intelligence; the user They evaluate behavior solely based on rules (using if-then constructs), Deep learning-based personalization or behavioral trends over time It does not conduct analyses. Real behavioral change in risky customers. It does not offer proactive suggestions or structured interventions to provide this. 30 Some major banks (such as JPMorgan Chase) improve credit scores. 3 They made plans to offer rewards to customers in the form of interest rate reductions; however, The systems are not yet fully developed in the process of helping users improve their financial behavior. It has not yet been established with a structure that will provide support. It is still based on the principle of "reward if credit score improves." Reactive mechanisms such as these are prominent; how to increase the score The customer is not actively being guided. Furthermore, the current AI-powered 5 to encourage "good profile" users in credit risk systems Comprehensive solutions are also lacking. High-profile customers are often only found in the system in limited numbers. They are passively defined as low risk; these clients have specific strategic plans. Incentives are not being offered proactively. As a result of these shortcomings: Bad Active behavioral modification to reduce the risk of profiled customers 10 There is no mechanism in place to ensure the loyalty and long-term success of high-profile customers. Dynamic incentive systems that would increase its value are insufficient; artificial intelligence An effective integration between predictions and gamified user interaction. This cannot be achieved. In this context, other important observations in existing systems... Technical shortcomings include: financial data, behavioral data, and text-based 15. Customer communications are mostly processed in separate systems; this creates a multifaceted risk. This prevents the analysis of the risk score when user behavior changes. It is usually updated periodically (weekly / monthly); instant adaptation. This is not possible. In some systems, gamification elements and financial... This can undermine the seriousness of the responsibility and create a false sense of security in the user. It can create a feeling. Gamified systems have more users. The need to collect behavioral data is governed by the KVKK (Personal Data Protection Law). GDPR (General Data Protection Regulation) This can create a risk of non-compliance with regulations of this kind. So far... Most gamified credit systems have open and transparent data processing policies. 25 It is incomplete. Therefore, considering the studies and shortcomings in the current technique... when present, gamified tasks and rewards-backed, AI-powered credit score rehabilitation 30 the creation of, interest rate reductions for, and interest-free repayment period extensions for users with good profiles, 4 AI-optimized incentive in the form of cash support in exchange for early payment. creating mechanisms to use artificial intelligence to produce both good and bad profiles. implementing dynamic risk management for customers, for each customer offering specific task and reward combinations, natural language processing algorithms sentiment and risk analysis from customer statements, XGBoost (eXtreme Gradient 5 Tabular data risk modeling with Gradient Boosting (Extreme Gradient Boosting), LSTM (Long Terrestrial Transformation) Short-Term Memory (STM) and Time Series Payment Behavior optimal monitoring and RL (Reinforcement Learning) providing intervention options, fully compliant with KVKK and GDPR principles, data Providing transparent data usage supported by anonymization processes, task 10 Completion, leveling up, personalized rewards, and dynamic credit terms. Encouraging genuine behavioral change through revisions, rather than just risky a company that doesn't focus on getting rid of customers, but also rewards good customers. a system is needed that provides a holistic approach that increases loyalty It is understood. 15 United States Regulation US11514517B2, which is included in the known state of the art. The patent document describes a method that increases a user's credit score. The system is being discussed. The invention in question involves a credit associated with a user. 20 a credit change software tool configured to automate The system receives the appropriate accounts and application amounts for improvements. It can determine the application process and its impact on the user's credit score. The invention in question aims to reduce recurring debts in situations where funds are unavailable. 25 ways to improve a user's credit score by applying for a loan These scenarios can include various factors, thus resulting in a clear improvement in credit scores. Loans, negotiable and autonomous depending on the credit score obtained. It is feasible. Which methods are used to retrieve data and benefit a credit score? a transaction model that determines the order and amount in which the calculations will be applied A universal payment system is described to determine this. 30 Another document included in the known state of the art is AU2014233357B2. In an Australian patent document, number [number], a system for customer payment processing is described. to maintain interest in addressing concerns through gamification techniques The text describes a system that enables its use. The invention in question involves a... The processor is configured to store customer profile data in a persistent 5 It includes an intelligent automated tool that contains a storage device and memory. The processor; a customer from past interactions between the contact center and the customer to learn about and apply the learned information to future interactions to run an artificial intelligence engine structured for customer profiling It performs operations such as storing data on a storage device. 10 Maintaining customer profile data allows us to guide the customer at the start of a new interaction. retrieval of profile data, new interactions based on the retrieved customer profile data. to be used to decide how to handle and new interaction Once completed, customer profile data and new interaction history will be displayed. It includes updating it to reflect one of the interactions. 15 Another document included in the known state of the art is US2018276710A1. In the United States patent document number [number], a registered customer is a registered customer. a loyalty program method that encourages transactions with the seller It is mentioned that in the invention in question, the method is performed by a supercomputer on 20... registered sellers and registered customers are identified by an artificial intelligence engine. It collects transaction data between them through data mining. The method is based on a recorded data. the probability of accepting an offer that includes incentives by conducting a transaction with the customer It estimates. The incentive is for a registered customer to conduct a transaction with a registered seller. a donation made by the seller to an organization with which they have established a close relationship, and in return 25 This could be in the form of a registered customer transacting with a registered seller. Brief Description of the Invention The aim of this invention is to create a gamified 30-point system for customers with different credit profiles. A reimbursement revision combined with an AI-powered prediction mechanism. 6 by introducing task-based credit score improvement for customers with high credit risk. providing processes and rewarding low-risk and reliable customers. a system developed to provide focused advantages to accomplish. Another aim of this invention is to use artificial intelligence to identify customers with high credit risk and the likelihood of customers with low credit risk experiencing payment difficulties in advance taking proactive measures by anticipating risks, financial assistance for at-risk customers They should be encouraged to improve their situations through gamified tasks and Flexible revisions to payment plans based on risk factors when necessary. 10 the implementation, also for customers with good profiles and a clean payment history dynamically rewarding individuals based on their loyalty and performance, improving receivables management for lending institutions and for customers to ensure a fairer and more encouraging credit experience is offered It is about implementing a developed system. 15 Detailed Description of the Invention The "High Credit Risk and Credit" project was carried out to achieve the purpose of this invention. Gamified Revision and AI for Low-Risk Credit Profiles 20 The "Supported Delay Prediction System" is shown in the attached figure; Figure 1 shows a schematic view of the system that is the subject of the invention. The parts shown in the figure are individually numbered, and these numbers correspond to 25. The corresponding answers are given below. 1. System 2. Electronic Devices 3. Mobile Application 30 4. Server 7 A gamified refund system for customers with different credit profiles. by combining the revision with an artificial intelligence-powered prediction mechanism, Task-based credit score improvement processes for high-risk customers. 5 focused on rewarding customers with low credit risk and reliability. The system in question, developed to provide advantages (1); - exchanging data using any remote communication protocol at least one electronic device configured to perform (2), - to be carried out on electronic devices (2), banking of users 10 at least one mobile application configured to provide (3), - electronic device using any remote communication protocol (2) and connecting with the mobile application (3), customers' past data on credit payment performance, adherence to payment dates, delay times, minimum payment frequency, full payment rate, 15 Detailed financial figures in the form of debts paid off and asset debt amount. continuously monitoring data, and analyzing this data over time To measure credit discipline and identify payment behavior patterns. to use for the purpose of the user's mobile application (3) closely monitoring their behavior, mobile 20 during payment periods Number of logins to the application (3), response time to reminder notifications, campaign engagement rate, frequency of contact with customer service, and To continuously monitor behavioral data in the form of complaint content and this Using data in risk and motivation analysis, age, gender, marital status, education level, region of residence, income level, income 25 parameters such as resource diversity and employment stability using customer profile analysis, the user's location unemployment rate, inflation, interest rate by sector and region, macroeconomic parameters such as economic growth data are constantly to monitor, use this external data to create individual risk scores, and these 30 evaluating data as an environmental risk factor, the customer's bank 8 correspondence with, support requests, mobile application (3) natural language processing algorithms process information in the form of messages to analyze stress, anxiety and financial levels in the language used by the customer Identifying signs of boredom, helping users complete tasks rate, reward earning speeds, in-game progression level, and preference 5 tracking data in the form of types of rewards they receive, and using this data to optimize motivational strategies and payment behavior at least one server configured to analyze influencing factors (4) includes. The electronic device (2) in the system (1) which is the subject of the invention, any remote communication exchanging data using the protocol, on mobile to run the application (3), access and view notifications and the technique using any remote communication protocol available in the known situation It is configured to connect to the server (4). 15 The invention includes mobile application (3), electronic device (2) in the system (1) to be run on, allowing users to manage their banking transactions and It is configured to allow them access to notifications and alerts. The server (4) in the system in question (1) is located in the known state of the art. using any remote communication protocol, electronic devices (2) and Connecting with the mobile application (3), customers' past credit payments Data regarding performance, payment date adherence, late times, minimum Payment frequency, full payment rate, debts paid off, and outstanding debt amount 25 Continuously monitoring detailed financial data in this form, and using this data to inform users to measure credit discipline over time and payment behavior patterns It is configured to be used for extraction purposes. Server (4), the user’s behavior on the mobile application (3) in detail Monitoring, number of logins to the mobile application (3) during payment periods, reminder 30 response time to notifications, campaign engagement rate, customer service. 9 Continuously monitoring behavioral data such as communication frequency and complaint content. and to use this data in risk and motivation analysis is configured. Server (4), age, gender, marital status, education level, region of residence, income level, diversity of income sources, and employment stability 5. To analyze customer profiles using parameters such as these: The server (4) is configured according to the sector and region where the user is located. macroeconomic data such as unemployment rate, inflation, interest rate, and economic growth Continuously monitoring parameters and using this external data to create individual risk scores. and to evaluate this data as an environmental risk factor is being configured. Server (4), customer correspondence with the bank, support 10 requests, information in the form of in-app messaging (3) in natural language Analyzing stress, anxiety, and language in the customer's language using processing algorithms It is structured to detect signs of financial distress. Server (4), Users' task completion rate, reward earning speed, and in-game progress. By monitoring data such as their level and preferred reward types, and using this data to 15 Optimizing motivational strategies and factors influencing payment behavior It is structured for analysis. In the system in question (1), the server (4) in the customer's written communications Advanced level 20 to extract hidden emotional and behavioral signals. using transformer-based natural language processing models, Explanations in deferment applications, support messages and mobile application (3) By automatically analyzing internal expressions, it detects emotional tones and financial stress. to identify indicators and signals of risky behavior is configured. Server (4) uses supervised machine learning models 25 using customer financial and behavioral characteristics to process payments. the possibility of delay, the likelihood of a postponement request, predictions of score changes to calculate, by ranking the variables in order of importance, which factors have a risk score. It is structured to dynamically report which has a greater impact. Server (4) long-term trend on customer credit payment time series 30 to perform analyses, using time series analysis; possible future periods Anticipating delays, requests for postponements, or the possibility of early closure, With LSTM (Long Short-Term Memory) networks; seasonality, habit changes and external economic impacts can be analyzed by artificial intelligence. It is configured to include in the model. The server (4) is reinforced. With learning algorithms; it tells the customer when and which gamified tasks to perform. what will be offered, what rewards will be given, or what loan revision proposals will be presented Dynamically determining what to do, using algorithms based on customer behavior to enable them to develop the optimal interaction strategy through learning It is being structured. In the system in question (1), the server (4) in the invention distributes the data flow in layers. processing sentiment and content scores from natural language processing models, LSTM by combining time series predictions and data model results generated creating a holistic customer risk profile, combining the outputs of multiple models The system is configured to combine and calculate a weighted score for the customer. 15 Server (4) combines all model outputs and Reinforcement Learning processing with algorithms, current risk situation, gamification history and economic using environmental information to provide the most suitable award, revision proposal and / or information. It is configured to determine the action in the form of: Server (4), user The risk increases if there are any changes in their behavior or financial situation. It is configured to update its score automatically. Server (4) in the system in question (1) is for customers with high credit risk; Offers to extend payment terms and reduce installments, along with a temporary interest rate reduction, are being reconsidered. structuring, increasing payment loyalty with small rewards in the early stages, 25 in the form of granting the right to defer installments in exchange for successful completion of the task to ensure that revision and support mechanisms are proposed, and to implement these revisions, based on the user's risk score, game progress, and overall economic situation It is configured to optimize. Server (4), low credit risk For customers; interest rate reduction offers, early payment bonuses, free 30-day installment plan. extension rights, credit limit increases, and loyalty program points or cash 11 to develop motivation and reward strategies in the form of awards It is being structured. The server (4) in the system (1) which is the subject of the invention, natural language processing models If the user's stress level or motivation is determined by the system, the user will be given a 5 to ensure more empathetic messages and supportive tasks are presented, By incorporating macroeconomic data into the model, under different economic scenarios To predict changes in user behavior and design proactive interventions. Continuously analyze the user's gaming behavior and assign tasks if their motivation decreases. 10 to automatically optimize the difficulty or reward type It is being structured. In the system in question (1), the server (4) has credit repayment behavior, demographic characteristics, financial transactions, bank account activity, credit card expenses, income and expense statements, payment history for previous loans, loan 15 financial records and similar data, customer service interactions, or Text-based information in the form of application forms, websites or mobile applications (3) usage records, a bank customer's mobile application (3) customer included in payment inquiries or support requests It is configured to collect data in the form of complaints. Server (4), 20 Application programming interfaces within the framework of open banking standards providing access to consumer financial data through open banking practices Thanks to programming interfaces and data integration services, the customer securely accessing bank accounts and transaction history, thereby All of a user's bank accounts and credit card transactions in one place. 25 Expanding the data repository by observing, third-party access to bank account information. using third-party access channels, updating credit card payment information For this, interface integrations with payment gateways and lending systems. to enable users to perform (3) on their mobile applications 30 to collect and record transactions in real time is being configured. Server (4), call center in customer support systems 12 records, support chatbot messages, or free text petitions text data in the form of interactive interfaces or email integrations collecting this text data and analyzing it using natural language processing methods. social media plays a role in financial products offered to credit customers. 5 data from demographic statistics or credit rating agencies Third-party data sources in the form of scores are also included in the data collection process. to temporarily store the collected data in its raw form throughout this process, thus access controls in accordance with data protection and compliance policies It is structured to provide this. The server (4) in the system (1) which is the subject of the invention, analyzes all the raw data collected. and making it suitable for modeling, cleaning and converting it to ensure that the data used as input for the models are consistent and free of noise. to ensure, identify and correct inconsistent or incomplete data entries, credit payments If there are any missing installment values ​​or incorrect date formats in the past, these errors will be corrected on 15 to resolve by filling in the average estimate or last known values, unprocessable To identify and, if necessary, reject extreme values ​​using statistical methods. It is configured to transform the data for modeling. The server (4) converting it into a suitable format, taking into account the user's profession, education level, and place of residence. Categorical data in the form of credit type; one-hot encoding, 20 Converting digital data into digital form through labeling or embedding methods scaling using normalization or standardization techniques, through this process by eliminating differences in the unit of measurement between variables Learning models are structured to ensure effective learning. Server (4) uses time series data in the form of credit payment history; delay 25 to extract attributes such as number, average payment amount, and loss rate It is structured. The server (4) processes text and interaction data using textual properties. parsing to extract, using natural language processing techniques to support Sentiment analysis and keyword extraction on records and customer statements. or to perform text classification operations, for this purpose deep language 30 using models to obtain vector representations (embedding) from text. 13 It is structured to process the data into big data during the preprocessing phase. The server (4) by processing with processing libraries, to create a scalable database or machine. It is structured to transfer learning materials to libraries in a suitable format. The server (4) in the system (1) which is the subject of the invention, receives many 5 from the pre-processing stage. By using multi-dimensional financial data and user behavior information, various Developing artificial intelligence prediction models, conducting analyses, and generating scores. Financial behavior analysis is performed in predicting loan repayment and deferment. to provide, to use supervised learning algorithms in the analysis process, supervised learning algorithms; based on past credit payment examples 10 an estimate of whether future payments will be made on time It is structured to enable the user to learn how to perform the operation. Server (4) will run regression on structured numerical data, Building models in the form of classification or sorting algorithms, text or specialized models for data types such as time series data 15 It is configured to enable the use of the server (4), digital data Common methods used for this include logistic regression, random forest, and vector machines in support. or using tree models in the form of XGBoost, with these types of models; customer's demographic profile, loan type, previous credit history, and monthly spending. Data in the form of trends can be used to predict credit scoring or deferment probability. It is configured to enable the use of deep learning. The server (4) Using artificial neural networks, which are among the models, such as LSTM or GRU (Gated Repeating Unit) their networks, by evaluating the user's past payment sequence, determine debt repayment. to enable them to learn about payment interruptions or changing trends, over the last 25 With an LSTM model that takes the sequence of loan payments over twenty-four months as input, a an estimate of the probability of whether payment will be made by the following month It is configured to enable this. The server (4) is text-based. Using natural language processing-based models for data, support requests or customer feedback BERT (Bidirectional Encoder Representations from 30 Transformers - Representations of Bidirectional Encoders from Transformers), RoBERTa 14 (Robustly Optimized BERT Approach) (approach) by processing with transformative architectural models To make sense of things, these models represent each text fragment in a vector space. converting, thus reducing payment based on a customer's complaint content. to enable the extraction of clues about ability, sentiment analysis of the text, 5 Identifying problem-solving themed keyword clusters and similar inferences It is configured to perform. The server (4) uses different data sources. using a combined dataset in model training, with the predictions obtained and generating numerical outputs with scores, using historical data for training models. It is configured to create labeled training sets. Server (4), 10 We test model performance through cross-validation to measure accuracy, precision, and AUC. With various metrics such as Area Under The Curve to evaluate, select high-performing models and perform hyperparameter optimization to improve models periodically with updated data using continuous learning techniques. It is structured to retrain as such. Server (4), investment and credit 15 data on platforms, in the form of social media data or purchase history It is structured to ensure that its resources are used in the model. Server (3) assesses the possibility of a repayment delay for each credit customer or It is structured to generate a credit score. In the system in question (1), the server (4) is in case of delay in credit payment. For a customer with a high probability of being eligible, increasing the interest rate on the loan or introducing a new one. requesting collateral, offering a deferred payment plan to a customer identified as having a medium risk level. Offering an action plan in this way, according to predefined workflows, includes tax, interest, and It is configured to control the warning limits. Server (4), maximum 25 Which additional services are available to which customers in order to ensure a high repayment collection rate? will be presented, Multi-Criteria Decision Making methods or optimization solving with algorithms, maximizing total debt repayment collection with a limited optimization aimed at determining what credit limit or... is given to which customer. It is configured to calculate that a discount will be given. Server (4), 30 the flexibility of learned models and the financial institution's supervised policy meeting the requirements ensures a customer receives a certain amount of their upcoming loan installment payment. Deciding to send automatic notifications a day in advance, payment reminder defining time rules in the form of a time limit, if a customer is late with payment, to detect this event and initiate an automatic alert or additional enforcement action. It is being structured. 5 The server (4) in the system (1) which is the subject of the invention, incorporates game principles into business processes. integrating points, badges and levels into a classic credit payment tracking system. Adding game elements like this allows the user to make a certain number of payments on time. enabling them to earn points or badges as they do so, 10 for each planned loan installment Enabling users who pay on time to earn points, upfront payment Users who complete or are on time with payments will earn a special achievement badge. to ensure that the user's level increases as they accumulate points leaderboards as a social gaming element to provide or offer special advantages. to offer 15 on time payment among users with similar profiles It is structured to improve performance by creating competition. The server (4) in the system (1) that is the subject of the invention provides each user with real-time or provide scheduled information, payment reminders, status reports, sending notifications in the form of warnings and reward announcements, upcoming loan installments, 20 in the form of limit changes, point gains, or suspicious transaction alerts. send notifications, automatically create a payment reminder message, Personalizing the message text according to the user's language and preferences, mobile app notification system, text message, email or even call center The robot is configured to send messages via voice systems. Server 25 (4), both in-app alert and short message for urgent payment reminder. While available, sending the monthly financial summary report via email, notification enriching their content with user profile information and score status It is structured accordingly. 16 The server (4) in the system (1) which is the subject of the invention, processes and stores all information; to protect the privacy, integrity and accessibility of users' financial and to protect your personal data from unauthorized access, breach and misuse, data It is difficult to maintain security within the framework of legal regulations and industry standards. It is configured as follows: Server (4), encryption, authentication, 5 Implementing multi-layered measures in the form of authorization, monitoring, and auditing, data Using encryption techniques in the transmission and storage processes of sensitive information Storing financial data by encryption, symmetric encryption algorithms using algorithms to hash user passwords and similar information It is configured to protect. Server (4), authentication methods 10 using firewalls, intrusion detection systems, and protection software using, regularly updating security features, and actively combating malicious software. monitoring entrepreneurial attempts, anonymizing or masking personal data protecting aggregated data used in big data analytics with techniques Analyzing in an anonymized form, keeping only the minimum necessary data and 15 Ensuring privacy and regulatory compliance by deleting unnecessary data in a timely manner. providing data in the form of credit information or user notifications If there are legal restrictions, access records that comply with those restrictions and It is structured to generate transparency reports. Industrial application of the invention In the system that is the subject of the invention (1) server (5), customers with different credit profiles An AI-powered prediction with a gamified payback revision. by bringing together the mechanisms, task 25 for customers with high credit risk offering credit score improvement processes based on low credit risk and reliability. Reward-oriented benefits are provided for customers. The invention consists of the electronic device (2), mobile application (3) and (1) included in the system. Server (4) informs and consents to the user in accordance with data privacy principles 30 17 It provides services and operates within the scope of the Personal Data Protection Law (KVKK). It shows. Around these fundamental concepts, the invention focuses on "High Credit Risk and Credit Risk". 5 Gamified Revision and AI-Powered Solutions for Low Credit Profiles Development of a wide variety of applications related to the "Delay Prediction System (1)" It is possible, and the invention is not limited to the examples described here, but is essentially As stated in the requests. 1 TARIFF HIGH CREDIT RISK AND LOW CREDIT RISK CREDIT PROFILES GAMIFIED REVISION AND AI-POWERED FOR DELAY PREDICTION SYSTEM Technical Area This invention provides a gamified return system for customers with different credit profiles. A combination of payment revision and an AI-powered prediction mechanism. by introducing task-based credit score improvement for customers with high credit risk. 10 providing processes and rewarding low-risk and reliable customers. It relates to a system that enables the provision of focused advantages. Previous Technique In the known state of the art, credit risk estimation and performance are available in current financial technologies. Customer management relies primarily on fixed data and traditional scoring. It has remained limited to these models. Credit bureaus and banks assess customers' credit risk. When determining this, factors such as past payment history, current debt status, and income level are considered. It uses statistical models based on static variables. These systems 20 usually logistic regression, linear discriminant analysis or simple decision It is based on traditional methods in the form of trees, customer from instantly and proactively capturing dynamic changes in their behavior It is far away. Thus, risk assessment is often a passive process based on historical data. It remains limited to the forecasting process, adapting to real-time changes 25 It is unable to provide this. In recent years, the financial technology sector has been facing increased credit risk. It has begun integrating artificial intelligence and machine learning techniques into its analytics; especially decision trees, random forest, XGBoost (eXtreme Gradient Boosting- Default prediction using Extreme Gradient Boosting and artificial neural networks Significant improvements in accuracy have been achieved. For example, AI-based programs like "Upstart" have been used for 30 years. Lending platforms are modeling over a hundred variables, surpassing traditional methods. 2 Compared to other methods, they have reported default rate improvements of up to 16%. In this way, companies like "HighRadius" analyze a customer's financial behavior. Detecting changes in real time and providing credit teams with immediate risk information. They have developed systems that send warnings. However, these systems are mostly background checks. The analytical infrastructures working in the plan, and the interaction with the end user (customer) is 5 It does not establish. Credit risk systems developed to date, user It has lacked proactive intervention mechanisms for its behavior. Current practices mostly update the customer's risk score silently, It takes action when the customer defaults. However, early intervention... It is a critical requirement to be able to improve the customer's risk profile in a timely manner. 10 These systems do not offer the customer opportunities to change the risk situation. a missed potential for both financial institutions and borrowers. It represents... In addition, gamification in financial services. (Gamification) applications have been used in recent years to increase customer interaction. It has started to be used. 15 in the form of points, badges, levels, achievement rewards. Game mechanics elements have been adapted to the use of financial products. For example, some Banks use badges to encourage regular saving in savings accounts. Some mobile applications offer users the opportunity to set their spending goals. It offers virtual rewards if they accomplish these things. In the US, there's a guy named "Lenny". A financial technology startup is helping young users in the credit score building process. integrated gamified tasks; users who pay on time This allowed him to earn "Lenny Points," which in turn increased his credit limits. However, current gamification applications have several serious shortcomings: It is generally associated with financial behavior at a superficial level; for example, saving. It rewards doing so, but it also increases the risk of loan deferral or the dynamics of financial distress. 25 It cannot model it. It is not strongly integrated with artificial intelligence; the user They evaluate behavior solely based on rules (using if-then constructs), Deep learning-based personalization or behavioral trends over time It does not conduct analyses. Real behavioral change in risky customers. It does not offer proactive suggestions or structured interventions to provide this. 30 Some major banks (such as JPMorgan Chase) improve credit scores. 3 They made plans to offer rewards to customers in the form of interest rate reductions; however, The systems are not yet fully developed in the process of helping users improve their financial behavior. It has not yet been established with a structure that will provide support. It is still based on the principle of "reward if credit score improves." Reactive mechanisms such as these are prominent; how to increase the score The customer is not actively being guided. Furthermore, the current AI-powered 5 to encourage "good profile" users in credit risk systems Comprehensive solutions are also lacking. High-profile customers are often only found in the system in limited numbers. They are passively defined as low risk; these clients have specific strategic plans. Incentives are not being offered proactively. As a result of these shortcomings: Bad Active behavioral modification to reduce the risk of profiled customers 10 There is no mechanism in place to ensure the loyalty and long-term success of high-profile customers. Dynamic incentive systems that would increase its value are insufficient, artificial intelligence An effective integration between predictions and gamified user interaction. This cannot be achieved. In this context, other important observations in existing systems... Technical shortcomings include: financial data, behavioral data, and text-based 15. Customer communications are mostly processed in separate systems; this creates a multifaceted risk. This prevents the analysis of the risk score when user behavior changes. It is usually updated periodically (weekly / monthly); instant adaptation. This is not possible. In some systems, gamification elements and financial... This can undermine the seriousness of the responsibility and create a false sense of security in the user. It can create a feeling. Gamified systems have more users. The need to collect behavioral data is governed by the KVKK (Personal Data Protection Law). GDPR (General Data Protection Regulation) This can create a risk of non-compliance with regulations of this kind. So far... Most gamified credit systems have open and transparent data processing policies. 25 It is incomplete. Therefore, considering the studies and shortcomings in the current technique... when present, gamified tasks and rewards-backed, AI-powered credit score rehabilitation 30 the creation of, interest rate reductions for, and interest-free repayment period extensions for users with good profiles, 4 AI-optimized incentive in the form of cash support in exchange for early payment. creating mechanisms to use artificial intelligence to produce both good and bad profiles. implementing dynamic risk management for customers, for each customer offering specific task and reward combinations, natural language processing algorithms sentiment and risk analysis from customer statements, XGBoost (eXtreme Gradient 5 Tabular data risk modeling with Gradient Boosting (Extreme Gradient Boosting), LSTM (Long Terrestrial Transformation) Short-Term Memory (STM) and Time Series Payment Behavior optimal monitoring and RL (Reinforcement Learning) providing intervention options, fully compliant with KVKK and GDPR principles, data Providing transparent data usage supported by anonymization processes, task 10 Completion, leveling up, personalized rewards, and dynamic credit terms. Encouraging genuine behavioral change through revisions, rather than just risky a company that doesn't focus on getting rid of customers, but also rewards good customers. a system is needed that provides a holistic approach that increases loyalty It is understood. 15 United States Regulation US11514517B2, which is included in the known state of the art. The patent document describes a method that increases a user's credit score. The system is being discussed. The invention in question involves a credit associated with a user. 20 a credit change software tool configured to automate The system receives the appropriate accounts and application amounts for improvements. It can determine the application process and its impact on the user's credit score. The invention in question aims to reduce recurring debts in situations where funds are unavailable. 25 ways to improve a user's credit score by applying for a loan These scenarios can include various factors, thus resulting in a clear improvement in credit scores. Loans, negotiable and autonomous depending on the credit score obtained. It is feasible. Which methods are used to retrieve data and benefit a credit score? a transaction model that determines the order and amount in which the calculations will be applied A universal payment system is described to determine this. 30 Another document included in the known state of the art is AU2014233357B2. In an Australian patent document, number [number], a system for customer payment processing is described. to maintain interest in addressing concerns through gamification techniques The text describes a system that enables its use. The invention in question involves a... The processor is configured to store customer profile data in a persistent 5 It includes an intelligent automated tool that contains a storage device and memory. The processor; a customer from past interactions between the contact center and the customer to learn about and apply the learned information to future interactions to run an artificial intelligence engine structured for customer profiling It performs operations such as storing data on a storage device. 10 Maintaining customer profile data allows us to guide the customer at the start of a new interaction. retrieval of profile data, new interactions based on the retrieved customer profile data. to decide how to handle and new interaction Once completed, customer profile data and new interaction history will be displayed. It includes updating it to reflect one of the interactions. 15 Another document included in the known state of the art is US2018276710A1. In the United States patent document number [number], a registered customer is a registered customer. a loyalty program method that encourages transactions with the seller It is mentioned that in the invention in question, the method is performed by a supercomputer on 20... registered sellers and registered customers are identified by an artificial intelligence engine. It collects transaction data between them through data mining. The method is based on a recorded data. the probability of accepting an offer that includes incentives by conducting a transaction with the customer It estimates. The incentive is for a registered customer to conduct a transaction with a registered seller. a donation made by the seller to an organization with which they have established a close relationship, and in return 25 This could be in the form of a registered customer transacting with a registered seller. Brief Description of the Invention The aim of this invention is to create a gamified 30-point system for customers with different credit profiles. A combination of reimbursement revision and an AI-powered prediction mechanism. 6 by introducing task-based credit score improvement for customers with high credit risk. providing processes and rewarding low-risk and reliable customers. a system developed to provide focused advantages to accomplish. Another aim of this invention is to use artificial intelligence to identify customers with high credit risk and the likelihood of customers with low credit risk experiencing payment difficulties in advance taking proactive measures by anticipating risks, financial assistance for at-risk customers and Flexible revisions to payment plans based on risk factors when necessary. 10 the implementation, also for customers with good profiles and a clean payment history dynamically rewarding individuals based on their loyalty and performance, improving receivables management for lending institutions and for customers to ensure a fairer and more encouraging credit experience is offered It is about implementing a developed system. 15 Detailed Description of the Invention The "High Credit Risk and Credit" project was carried out to achieve the purpose of this invention. Gamified Revision and AI for Low-Risk Credit Profiles 20 The "Supported Delay Prediction System" is shown in the attached figure; Figure 1 shows a schematic view of the system that is the subject of the invention. The parts shown in the figure are individually numbered, and these numbers correspond to 25. The corresponding answers are given below. 1. System 2. Electronic Devices 3. Mobile Application 30 4. Server 7 A gamified refund system for customers with different credit profiles. by combining the revision with an artificial intelligence-powered prediction mechanism, Task-based credit score improvement processes for high-risk customers. 5 focused on rewarding customers with low credit risk and reliability. The system in question, developed to provide advantages (1); - exchanging data using any remote communication protocol at least one electronic device configured to perform (2), - to be carried out on electronic devices (2), banking of users 10 at least one mobile application configured to provide (3), - electronic device using any remote communication protocol (2) and connecting with the mobile application (3), customers' past data on credit payment performance, adherence to payment dates, delay times, minimum payment frequency, full payment rate, 15 Detailed financial figures in the form of debts paid off and asset debt amount. continuously monitoring data, and analyzing this data over time To measure credit discipline and identify payment behavior patterns. to use for the purpose of the user's mobile application (3) closely monitoring their behavior, mobile 20 during payment periods Number of logins to the application (3), response time to reminder notifications, campaign engagement rate, frequency of contact with customer service, and To continuously monitor behavioral data in the form of complaint content and this Using data in risk and motivation analysis, age, gender, marital status, education level, region of residence, income level, income 25 parameters such as resource diversity and employment stability using customer profile analysis, the user's location unemployment rate, inflation, interest rate by sector and region, macroeconomic parameters such as economic growth data are constantly to monitor, use this external data to create individual risk scores, and these 30 evaluating data as an environmental risk factor, the customer's bank 8 correspondence with, support requests, mobile application (3) natural language processing algorithms process information in the form of messages to analyze stress, anxiety and financial levels in the language used by the customer Identifying signs of boredom, helping users complete tasks rate, reward earning speeds, in-game progression level, and preference 5 tracking data in the form of types of rewards they receive, and using this data to optimize motivational strategies and payment behavior at least one server configured to analyze influencing factors (4) includes. The electronic device (2) in the system (1) which is the subject of the invention, any remote communication exchanging data using the protocol, on mobile to run the application (3), access and view notifications and the technique using any remote communication protocol available in the known situation It is configured to connect to the server (4). 15 The invention includes mobile application (3), electronic device (2) in the system (1) to be run on, allowing users to manage their banking transactions and It is configured to allow them access to notifications and alerts. The server (4) in the system in question (1) is located in the known state of the art. using any remote communication protocol, electronic devices (2) and Connecting with the mobile application (3), customers' past credit payments Data regarding performance, payment date adherence, late times, minimum Payment frequency, full payment rate, debts paid off, and outstanding debt amount 25 Continuously monitoring detailed financial data in this form, and using this data to inform users to measure credit discipline over time and payment behavior patterns It is configured to be used for extraction purposes. Server (4), the user’s behavior on the mobile application (3) in detail Monitoring, number of logins to the mobile application (3) during payment periods, reminder 30 response time to notifications, campaign engagement rate, customer service. 9 Continuously monitoring behavioral data such as communication frequency and complaint content. and to use this data in risk and motivation analysis is configured. Server (4), age, gender, marital status, education level, region of residence, income level, diversity of income sources, and employment stability 5. To analyze the customer profile using parameters such as these: The server (4) is configured according to the sector and region where the user is located. macroeconomic data such as unemployment rate, inflation, interest rate, and economic growth Continuously monitoring parameters and using this external data to create individual risk scores. and to evaluate this data as an environmental risk factor is being configured. Server (4), customer correspondence with the bank, support 10 requests, information in the form of in-app messaging (3) in natural language Analyzing stress, anxiety, and language in the customer's language using processing algorithms It is structured to detect signs of financial distress. Server (4), Users' task completion rate, reward earning speed, and in-game progress. By monitoring data such as their level and preferred reward types, and using this data to 15 Optimizing motivational strategies and factors influencing payment behavior It is structured for analysis. In the system in question (1), the server (4) in the customer's written communications Advanced level 20 to extract hidden emotional and behavioral signals. using transformer-based natural language processing models, Explanations in deferment applications, support messages and mobile application (3) By automatically analyzing internal expressions, it detects emotional tones and financial stress. to identify indicators and signals of risky behavior is configured. Server (4) uses supervised machine learning models 25 using customer financial and behavioral characteristics to process payments. the possibility of delay, the likelihood of a postponement request, predictions of score changes to calculate, by ranking the variables in order of importance, which factors have a risk score. It is structured to dynamically report which has a greater impact. Server (4) long-term trend on customer credit payment time series 30 to perform analyses, using time series analysis; possible future periods Anticipating delays, requests for postponements, or the possibility of early closure, With LSTM (Long Short-Term Memory) networks; seasonality, habit changes and external economic impacts can be analyzed by artificial intelligence. It is configured to include in the model. The server (4) is reinforced. With learning algorithms; it tells the customer when and which gamified tasks to perform. what will be offered, what rewards will be given, or what loan revision proposals will be presented Dynamically determining what to do, using algorithms based on customer behavior to enable them to develop the optimal interaction strategy through learning It is being structured. In the system in question (1), the server (4) in the invention distributes the data flow in layers. processing sentiment and content scores from natural language processing models, LSTM by combining time series predictions and data model results generated creating a holistic customer risk profile, combining the outputs of multiple models The system is configured to combine and calculate a weighted score for the customer. 15 Server (4) combines all model outputs and Reinforcement Learning processing with algorithms, current risk situation, gamification history and economic using environmental information to provide the most suitable award, revision proposal and / or information. It is configured to determine the action in the form of: Server (4), user The risk increases if there are any changes in their behavior or financial situation. It is configured to update its score automatically. Server (4) in the system in question (1) is for customers with high credit risk; Offers to extend payment terms and reduce installments, along with a temporary interest rate reduction, are being reconsidered. structuring, increasing payment loyalty with small rewards in the early stages, 25 in the form of granting the right to defer installments in exchange for successful completion of the task to ensure that revision and support mechanisms are proposed, and to implement these revisions, based on the user's risk score, game progress, and overall economic situation It is configured to optimize. Server (4), low credit risk For customers; interest rate reduction offers, early payment bonuses, free 30-day installment plan. extension rights, credit limit increases, and loyalty program points or cash 11 to develop motivation and reward strategies in the form of awards It is being structured. The server (4) in the system (1) which is the subject of the invention, natural language processing models If the user's stress level or motivation is determined by the system, the user will be given a 5 to ensure more empathetic messages and supportive tasks are presented, By incorporating macroeconomic data into the model, under different economic scenarios To predict changes in user behavior and design proactive interventions. Continuously analyze the user's gaming behavior and assign tasks if their motivation decreases. 10 to automatically optimize the difficulty or reward type It is being structured. In the system in question (1), the server (4) has credit repayment behavior, demographic characteristics, financial transactions, bank account activity, credit card expenses, income and expense statements, payment history for previous loans, loan 15 financial records and similar data, customer service interactions, or Text-based information in the form of application forms, websites or mobile applications (3) usage records, a bank customer's mobile application (3) customer included in payment inquiries or support requests It is configured to collect data in the form of complaints. Server (4), 20 Application programming interfaces within the framework of open banking standards providing access to consumer financial data through open banking practices Thanks to programming interfaces and data integration services, the customer securely accessing bank accounts and transaction history, thereby All of a user's bank accounts and credit card transactions in one place. 25 Expanding the data repository by observing, third-party access to bank account information. using third-party access channels, updating credit card payment information For this, interface integrations with payment gateways and lending systems. to enable users to perform (3) on their mobile applications 30 to collect and record transactions in real time is being configured. Server (4), call center in customer support systems 12 records, support chatbot messages, or free text petitions text data in the form of interactive interfaces or email integrations collecting this text data and analyzing it using natural language processing methods. social media plays a role in financial products offered to credit customers. data, demographic statistics or credit rating agencies 5 Third-party data sources in the form of scores are also included in the data collection process. to temporarily store the collected data in its raw form throughout this process, thus access controls in accordance with data protection and compliance policies It is structured to provide this. The server (4) in the system (1) which is the subject of the invention, analyzes all the raw data collected. and making it suitable for modeling, cleaning and converting it to ensure that the data used as input for the models are consistent and free of noise. to ensure, identify and correct inconsistent or incomplete data entries, credit payments If there are missing installment values ​​or incorrect date formats in the past, these errors will be corrected on 15 to resolve by filling in the average estimate or last known values, unprocessable To identify and, if necessary, reject extreme values ​​using statistical methods. It is configured to transform the data for modeling. The server (4) converting it into a suitable format, taking into account the user's profession, education level, and place of residence. Categorical data in the form of credit type; one-hot encoding, 20 Converting digital data into digital form through labeling or embedding methods scaling using normalization or standardization techniques, through this process by eliminating differences in the unit of measurement between variables Learning models are structured to ensure effective learning. Server (4) uses time series data in the form of credit payment history; delay 25 to extract attributes such as number, average payment amount, and loss rate It is structured. The server (4) processes text and interaction data using textual properties. parsing to extract, using natural language processing techniques to support Sentiment analysis and keyword extraction on records and customer statements. or to perform text classification operations, for this purpose deep language 30 using models to obtain vector representations (embedding) from text. 13 It is structured to process the data into big data during the preprocessing phase. The server (4) by processing with processing libraries, to create a scalable database or machine. It is structured to transfer learning materials to libraries in a suitable format. The server (4) in the system (1) which is the subject of the invention, receives many 5 from the pre-processing stage. By using multi-dimensional financial data and user behavior information, various Developing artificial intelligence prediction models, conducting analyses, and generating scores. Financial behavior analysis is performed in predicting loan repayment and deferment. to provide, to use supervised learning algorithms in the analysis process, supervised learning algorithms; based on past credit payment examples 10 an estimate of whether future payments will be made on time It is structured to enable the user to learn how to perform the operation. Server (4) will run regression on structured numerical data, Building models in the form of classification or sorting algorithms, text or specialized models for data types such as time series data 15 It is configured to enable the use of the server (4), digital data Common methods used for this include logistic regression, random forest, and vector machines in support. or using tree models in the form of XGBoost, with these types of models; customer's demographic profile, loan type, previous credit history, and monthly spending. Data in the form of trends can be used to predict credit scoring or deferment probability. It is configured to enable the use of deep learning. The server (4) Using artificial neural networks, which are among the models, such as LSTM or GRU (Gated Repeating Unit) their networks, by evaluating the user's past payment sequence, determine debt repayment. to enable them to learn about payment interruptions or changing trends, over the last 25 With an LSTM model that takes the sequence of loan payments over twenty-four months as input, a an estimate of the probability of whether payment will be made by the following month It is configured to enable this. The server (4) is text-based. Using natural language processing-based models for data, support requests or customer feedback BERT (Bidirectional Encoder Representations from 30 Transformers - Representations of Bidirectional Encoders from Transformers), RoBERTa 14 (Robustly Optimized BERT Approach) (approach) by processing with transformative architectural models To make sense of things, these models represent each text fragment in a vector space. converting, thus reducing payment based on a customer's complaint content. to enable the extraction of clues about ability, sentiment analysis of the text, 5 Identifying problem-solving themed keyword clusters and similar inferences It is configured to perform. The server (4) uses different data sources. using a combined dataset in model training, with the predictions obtained and generating numerical outputs with scores, using historical data for training models. It is configured to create labeled training sets. Server (4), 10 We test model performance through cross-validation to measure accuracy, precision, and AUC. With various metrics such as Area Under The Curve to evaluate, select high-performing models and perform hyperparameter optimization to improve models periodically with updated data using continuous learning techniques. It is structured to retrain as such. Server (4), investment and credit 15 data on platforms, in the form of social media data or purchase history It is structured to ensure that its resources are used in the model. Server (3) assesses the possibility of a repayment delay for each credit customer or It is structured to generate a credit score. In the system in question (1), the server (4) is in case of delay in credit payment. For a customer with a high probability of being eligible, increasing the interest rate on the loan or introducing a new one. requesting collateral, offering a deferred payment plan to a customer identified as having a medium risk level. Offering an action plan in this way, according to predefined workflows, includes tax, interest, and It is configured to control the warning limits. Server (4), maximum 25 Which additional services are available to which customers in order to ensure a high repayment collection rate? will be presented, Multi-Criteria Decision Making methods or optimization solving with algorithms, maximizing total debt repayment collection with a limited optimization aimed at determining what credit limit or... is given to which customer. It is configured to calculate that a discount will be given. Server (4), 30 the flexibility of learned models and the financial institution's supervised policy meeting the requirements ensures a customer receives a certain amount of their upcoming loan installment payment. Deciding to send automatic notifications a day in advance, payment reminder defining time rules in the form of a time limit, if a customer is late with payment, to detect this event and initiate an automatic alert or additional enforcement action. It is being structured. 5 The server (4) in the system (1) which is the subject of the invention, incorporates game principles into business processes. integrating points, badges and levels into a classic credit payment tracking system. Adding game elements like this allows the user to make a certain number of payments on time. enabling them to earn points or badges as they do so, 10 for each planned loan installment Enabling users who pay on time to earn points, upfront payment Users who complete or are on time with payments will earn a special achievement badge. to ensure that the user's level increases as they accumulate points leaderboards as a social gaming element to provide or offer special advantages. to offer 15 on time payment among users with similar profiles It is structured to improve performance by creating competition. The server (4) in the system (1) that is the subject of the invention provides each user with real-time or provide scheduled information, payment reminders, status reports, sending notifications in the form of warnings and reward announcements, upcoming loan installments, 20 in the form of limit changes, point gains, or suspicious transaction alerts. send notifications, automatically create a payment reminder message, Personalizing the message text according to the user's language and preferences, mobile app notification system, text message, email or even call center The robot is configured to send messages via voice systems. Server 25 (4), both in-app alert and short message for urgent payment reminder. While available, sending the monthly financial summary report via email, notification enriching their content with user profile information and score status It is structured accordingly. 16 The server (4) in the system (1) which is the subject of the invention, processes and stores all information; to protect the privacy, integrity and accessibility of users' financial and to protect your personal data from unauthorized access, breach and misuse, data It is difficult to maintain security within the framework of legal regulations and industry standards. It is configured as follows: Server (4), encryption, authentication, 5 Implementing multi-layered measures in the form of authorization, monitoring, and auditing, data Using encryption techniques in the transmission and storage processes of sensitive information Storing financial data by encryption, symmetric encryption algorithms using algorithms to hash user passwords and similar information It is configured to protect. Server (4), authentication methods 10 using firewalls, intrusion detection systems, and protection software using, regularly updating security features, and actively combating malicious software. monitoring entrepreneurial attempts, anonymizing or masking personal data protecting aggregated data used in big data analytics with techniques Analyzing in an anonymized form, keeping only the minimum necessary data and 15 Ensuring privacy and regulatory compliance by deleting unnecessary data in a timely manner. providing data in the form of credit information or user notifications If there are legal restrictions, access records that comply with those restrictions and It is structured to generate transparency reports. Industrial application of the invention In the system that is the subject of the invention (1) server (5), customers with different credit profiles An AI-powered prediction with a gamified payback revision. by bringing together the mechanisms, task 25 for customers with high credit risk offering credit score improvement processes based on low credit risk and reliability. Reward-oriented benefits are provided for customers. The invention consists of the electronic device (2), mobile application (3) and (1) included in the system. Server (4) informs and consents to the user in accordance with data privacy principles 30 17 It provides services and operates within the scope of the Personal Data Protection Law (KVKK). It shows. Around these fundamental concepts, the invention focuses on "High Credit Risk and Credit Risk". 5 Gamified Revision and AI-Powered Solutions for Low Credit Profiles Development of a wide variety of applications related to the "Delay Prediction System (1)" It is possible, and the invention is not limited to the examples described here, but is essentially As stated in the requests.

Claims

18 REQUESTS 1. A gamified return system for customers with different credit profiles. a payment revision with an AI-powered prediction mechanism By combining these, a task-based credit score of 5 is created for customers with high credit risk. providing improvement processes, low credit risk and reliable customers to ensure that reward-focused advantages are offered developed; - exchanging data using any remote communication protocol at least one electronic device configured to perform (2), 10 - to be carried out on electronic devices (2), banking of users to manage their operations and access notifications and alerts including at least one mobile application (3) configured to provide And; - using any remote communication protocol, electronic device 15 (2) and connecting with the mobile application (3), customers' past data on credit payment performance, adherence to payment dates, delay times, minimum payment frequency, full payment rate, Detailed financial figures in the form of debts paid off and asset debt amount. Continuously monitoring the data, using this data to analyze users' behavior over time. To measure credit discipline and identify payment behavior patterns. to use for the purpose of the user's mobile application (3) monitoring their behavior in detail, mobile during payment periods Number of logins to the application (3), response time to reminder notifications, campaign engagement rate, frequency of contact with customer service, and 25 To continuously monitor behavioral data in the form of complaint content and this Using data in risk and motivation analysis, age, gender, marital status, education level, region of residence, income level, income parameters such as resource diversity and employment stability By using customer profile analysis, the user's location is 30 unemployment rate, inflation, interest rate by sector and region, 19 macroeconomic parameters such as economic growth data are constantly to monitor, use this external data to perform individual risk scoring, and evaluating data as an environmental risk factor, the customer's bank correspondence with, support requests, mobile application (3) Information in the form of messages is processed using natural language processing algorithms. 5 to analyze stress, anxiety and financial levels in the language used by the customer Identifying signs of boredom, helping users complete tasks rate, reward earning speeds, in-game progression level, and preference tracking data in the form of types of rewards they receive, and using this data Optimizing motivational strategies and payment behavior 10 at least one server configured to analyze influencing factors a system characterized by (4) (1).

2. Data exchange using any remote communication protocol. to perform, run the mobile application (3), receive notifications 15 to access and view and any remote communication protocol electronic configured to connect to the server (4) using A system like the one in Claim 1, characterized by device (2) (1).

3. To be carried out on electronic devices (2), users' banking 20 to manage their operations and access notifications and alerts characterized by the mobile application (3) structured to provide a system like any of the above requests (1).

4. Using any remote communication protocol, electronic device (2) 25 and connecting with the mobile application (3), customers’ past credit history data on payment performance, payment due date, delay durations, minimum payment frequency, full payment rate, closed Detailed financial data in the form of debts and asset debt ratios are constantly being collected. monitoring this data, users' credit discipline over time 30 to measure and infer payment behavior patterns from the above requests characterized by the configured server (4) a system like any other (1).

5. Detailed user behavior on the mobile application (3) to monitor in this way, number of logins to the mobile application (3) during payment periods, 5 Response time to reminder notifications, campaign engagement rate, customer behavioral aspects such as frequency of communication with services and content of complaints continuously monitor the data and use this data in risk and motivation analysis. characterized by the server (4) configured for use a system like any of the above requests (1). 10 6. Age, gender, marital status, education level, region of residence, income level, parameters such as income source diversity and employment stability Server configured to analyze customer profile using (4) a 15 as in any of the above claims characterized by system (1).

7. Unemployment rate, inflation, specific to the user's sector and region, macroeconomic parameters such as interest rates and economic growth data continuously monitoring, using this external data to create individual risk scores, and these 20 structured to evaluate data as environmental risk factors any of the above requests characterized by the server (4) such a system (1).

8. Customer correspondence with the bank, support requests, mobile application 25 (3) information in the form of internal messaging with natural language processing algorithms to analyze stress, anxiety and financial difficulties in the language used by the customer characterized by the server (4) configured to detect its symptoms. a system like any of the above-mentioned requests (1). 21 9. Users' task completion rate, reward acquisition speed, in-game tracking data such as progress level and preferred types of rewards, With this data, we can optimize motivational strategies and payment behavior. with server (4) configured to analyze influencing factors a 5 as in any of the above characterized claims system (1).

10. The hidden feelings and behaviors of the customer in their written communications. advanced transducer-based natural language processing for extracting signals using their models, explanations in deferment applications, support 10 automatically analyze messages and expressions within the mobile application (3) by observing emotional tones, indicators of financial stress, and risky behavior. characterized by the server (4) configured to detect its signals. a system like any of the above-mentioned requests (1).

11. Using supervised machine learning models to analyze the customer's financial and processing behavioral characteristics, the likelihood of payment delay, postponement calculating the probability of application, predictions of score changes, variables By ranking the factors in order of importance, which factors carry a higher risk score? 20 with server (4) configured to dynamically report the impact. as in any of the above characterized claims system (1).

12. Long-term trend in customer credit payment time series. to conduct analyses, using time series analysis; 25 future periods possible delays, requests for postponement, or the possibility of early closure Predicting seasonality, habit changes, and with LSTM networks. to incorporate external economic impacts into the artificial intelligence model from the above requests characterized by the configured server (4) a system like any of them (1). 30 22 13. With reinforcement learning algorithms; when to tell the customer which that gamified tasks will be presented, what rewards will be given, or Dynamically determining which loan revision proposals will be made, Algorithms learn from customer behavior to achieve optimal interaction. 5 with the server (4) configured to enable it to develop its strategy as in any of the above characterized claims system (1).

14. Processing data streams in layers, drawing from 10 natural language processing models. sentiment and content scores, time series predictions generated with LSTM, and By combining data model results, a holistic customer risk profile is created. to create, combine the outputs of multiple models, for the client characterized by the server (4) configured to calculate weighted scores. a system like any of the above requests (1). 15 15. Combining all model outputs and using Reinforcement Learning algorithms. processing, current risk situation, gamification history and economic environment using their information to provide the most suitable award, revision offer and / or notification. Server (4) configured to determine the action in the form of 20 as in any of the above characterized claims system (1).

16. Any changes in user behavior or financial status. 25 configured to automatically update the risk score when any of the above requests characterized by the server (4) such a system (1).

17. For customers with high credit risk: extending the repayment period and reducing installments. proposals, restructuring with temporary interest rate reduction, 30 in the early stages 23 Enhancing loyalty through small rewards for successful task completion. revision and support in the form of granting the right to defer installments in return. to ensure that mechanisms are proposed, these revisions, user risk Optimized according to score, game progress and overall economic situation. The above 5 is characterized by the server (4) configured to do so. a system like any of the requests (1).

18. For customers with low credit risk; interest rate reduction offers, early payment bonuses, free loan extensions, increased credit limits, and loyalty programs. The program provides motivation and rewards in the form of points or cash prizes. 10 characterized by the server (4) configured to develop strategies a system like any of the above-mentioned requests (1).

19. User stress level or as determined by natural language processing models. If the motivation is identified, the user will receive more empathetic messages and 15 to ensure that supporting tasks are provided, macroeconomic data by including it in the model, user behavior under different economic scenarios Anticipating changes and designing proactive interventions, for the user Continuously analyze play behavior and assign tasks if motivation decreases. 20 to automatically optimize the difficulty or reward type from the above requests characterized by the configured server (4) a system like any other (1).

20. Credit repayment behavior, demographic characteristics, financial transactions, bank account statements, credit card spending, income and expense statements, previous 25 loan repayment history, credit reports, and similar financial records. data in the form of customer service interactions or petition forms text-based information, website or mobile application (3) use records made by a bank customer through the mobile application (3) Customer 30 included in payment inquiries or support tickets 24 Server configured to collect data in the form of complaints (4) as in any of the above claims characterized by system (1).

21. Application programming within the framework of open banking standards, intermediate 5 providing access to consumer financial data through their faces, openly banking application programming interfaces and data integration through its services, the customer's bank accounts and transaction history securely accessing all of the user's bank accounts and by monitoring credit card transactions in one place, you can create a wealth of data. expand third-party access channels for bank account information to use, payment network for updating credit card payment information. interface integrations with gateways and credit granting systems to provide users with what they do in their mobile applications (3) 15 to collect and record transactions in real time from the above requests characterized by the configured server (4) a system like any other (1).

22. Call center records and support chat logs in customer support systems. The robot's text in the form of correspondence or free-text petitions 20 data, through interactive interfaces or email integrations to collect and analyze this text data using natural language processing methods, Social media plays a role in financial products offered to credit customers. data from demographic statistics or credit rating agencies Data collection also includes third-party data sources in the form of scores obtained. 25 to include in the process, throughout this process, the collected data in their raw form. to retain it temporarily, so that it complies with data protection and compliance policies. server (4) configured to provide access controls as in any of the above characterized claims system (1). 30 23. All collected raw data is made suitable for analysis and modeling. to ensure the models are brought in, cleaned, and transformed. To ensure that the input data is consistent and noise-free, or inconsistent or Identifying and correcting missing data entries, missing 5 in credit payment history If there are any installment values ​​or incorrect date formats, these errors will be averaged out. to eliminate by filling with estimated or last known values, unprocessable Identifying and, if necessary, rejecting extreme values ​​using statistical methods. or characterized by the server configured to convert (4) a system like any of the above requests (1). 10 24. Converting the data into a suitable format for modeling allows the user to... categorical factors such as profession, education level, place of residence, and type of loan. The data is digitized using one-hot encoding, labeling, or embedding methods. converting to format, normalizing or standardizing numerical data 15 scaling with techniques, this process measures the relationship between variables. by eliminating unit differences, machine learning models characterized by the server (4) configured to enable effective learning. a system like any of the above-mentioned requests (1).

25. From time series data in the form of credit payment history; number of delays, to extract attributes such as average payment amount and loss rate from the above requests characterized by the configured server (4) a system like any other (1).

26. To extract textual features from text and interaction data. parsing, supporting records and using natural language processing techniques sentiment analysis, keyword extraction, or customer testimonials. to perform text classification processes, for this purpose deep linguistic analysis 30 using models to obtain vector representations from texts 26 from the above requests characterized by the configured server (4) a system like any other (1).

27. In the preprocessing stage, the data is processed using big data processing libraries. by processing a scalable database or machine learning 5 Server configured to transfer libraries in a suitable format (4) as in any of the above claims characterized by system (1).

28. Multidimensional financial data from the preprocessing stage and user 10 various artificial intelligence prediction models using behavioral data to improve, analyze and generate scores, credit repayment and In predicting procrastination, it is necessary to enable financial behavior analysis. Using supervised learning algorithms in the analysis process, supervised learning algorithms based on past credit payment examples 15 whether future payments will be made on time to enable it to learn how to perform the prediction process from the above requests characterized by the configured server (4) a system like any other (1).

29. Regression and classification systems that will work on structured numerical data. or establishing models in the form of sorting algorithms, text or Specialized models for data types in the form of time series data. characterized by the server (4) configured to enable its use. a system like any of the above requests (1). 25 30. For numerical data, logistic regression, random forest, and support methods are commonly used. Using vector machines or tree models in the form of XGBoost, this using typical models; a customer's demographic profile, loan type, previous loans Data such as credit history and monthly spending habits are used in credit scoring. 27 to enable its use in estimating the probability of postponement from the above requests characterized by the configured server (4) a system like any other (1).

31. Using artificial neural networks, which are among the deep learning models, 5 Recurrent neural networks, such as LSTM or GRU, analyze the user's past by evaluating the payment sequence, interruptions in debt repayment or to enable them to learn about changing trends, credit data from the last twenty-four months With an LSTM model that takes the payment sequence as input, for the next month 10 The above is characterized by the server (4) configured to provide a system like any of the requests (1).

32. Using natural language processing-based models for text-based data, BERT, RoBERTa 15, handles support requests or customer feedback. to make sense of it by processing it with transformative architectural models in this way, using models, transforming each text fragment into a vector space, thus regarding a customer's decreasing ability to pay based on the content of their complaint to enable the extraction of clues, sentiment analysis of the text, problem Detecting and inferring keyword clusters related to problem-solving. 20 characterized by the server (4) configured to perform a system like any of the above requests (1).

33. Using different data sources as a combined dataset in model training. to use, generate numerical outputs with the obtained predictions and scores, 25 labeled training sets from historical data for training models characterized by the server (4) configured to create a system like any of the above requests (1). 28 34. Model performance is tested through cross-validation to check for accuracy and precision. Evaluating with various metrics such as AUC indicates good performance. Selecting models and improving them through hyperparameter optimization, continuously periodically revise the models with current data using learning techniques. The above 5 is characterized by the server (4) configured to train. a system like any of the requests (1).

35. Investment and lending platforms, social media data or acquisitions. To enable the use of data sources such as historical data in the model. The above 10 is characterized by the server (4) configured to do so. a system like any of the requests (1).

36. The possibility of a potential repayment delay for each credit customer or Characterized by the server (4) configured to generate a credit score. a system like any of the above requests (1) 15 37. For a customer with a high probability of defaulting on loan payments, the loan... raising the interest rate or requesting new collateral, determining a medium risk level. Offering a customer a course of action such as a payment deferral proposal, Check tax, interest, and alert limits according to predefined workflows. 20 The above is characterized by the server (4) which is configured to do so. a system like any of the requests (1).

38. Which customers will receive the highest repayment collection rate? which additional services will be offered, Multi-Criteria Decision Making methods 25 or solve it with optimization algorithms, total debt repayment with a constrained optimization aimed at maximizing collection which to calculate how much credit limit or discount will be given to the customer from the above requests characterized by the configured server (4) a system like any of them (1). 30 29 39. The flexibility of learned models and the financial institution's supervised policy to meet the requirements, a customer's upcoming loan installment decided to send automatic notifications a certain number of days before the payment is due. giving, defining time rules in the form of payment reminder times, 5 If a customer is late on payment, the system detects this and sends an automatic alert. or the server (4) configured to initiate additional seizure proceedings as in any of the above characterized claims system (1).

40. Integrating game principles into business processes, a classic credit payment The tracking system uses game elements in the form of points, badges, and levels. adding points as the user makes a certain number of payments on time or enabling them to earn a badge, paying each scheduled loan installment on time. Enabling paying users to earn points, 15 who pay in advance or a special achievement badge for users who do not delay payments enabling the user to earn points as they accumulate them enabling promotion or offering special advantages, social gaming As an element, presenting leaderboards shows users with similar profiles. By creating competition among them regarding timely payments, performance will be increased to 20 The above is characterized by the server (4) configured to increase the above a system like any of the requests (1).

41. To provide each user with real-time or scheduled information, 25 in the form of payment reminders, status reports, alerts and reward announcements Sending notifications about upcoming loan installments, limit changes, and scores. to communicate notifications in the form of gains or suspicious transaction alerts, To automatically generate a payment reminder message, and to specify the message text, Personalizing this message according to the user's language and preferences, mobile application notification system, text message, email or even call center 30 Server configured to send messages via robot voice systems (4) like any of the above-mentioned claims characterized by system (1).

42. For an urgent payment reminder, both an in-app alert and a short message will appear. while available, sending the monthly financial summary report via email, notification content, user profile information and score status characterized by the server (4) configured to enrich a system like any of the above requests (1).

43. The confidentiality and integrity of all information processed and stored. to maintain accessibility to users' financial and personal data; To protect data security from unauthorized access, breach, and misuse. It is difficult to protect within the framework of legal regulations and industry standards. The above 15 is characterized by the server (4) configured to do so. a system like any of the requests (1).

44. In the form of encryption, authentication, authorization, monitoring, and auditing. implementing multi-layered measures in data transmission and storage processes using encryption techniques, storing sensitive information by encrypting it, 20 Using symmetric encryption algorithms for financial data, user to protect passwords and similar information using hashing algorithms from the above requests characterized by the configured server (4) a system like any other (1).

45. Using authentication methods, firewalls, intrusion detection Using security systems and protection software, regular security checks to make updates and actively counter malicious attempts monitoring, personal data through anonymizing or masking techniques protecting aggregated data used in big data analytics 30 31 Analyzing in an anonymized form, only the minimum necessary data. by maintaining privacy and deleting unnecessary data in a timely manner, ensuring regulatory compliance, credit information or user If there are legal restrictions on data usage in the form of notifications, 5. Creating access logs and transparency reports that comply with these limitations The above is characterized by the server (4) configured to do so. a system like any of the requests (1). 15 25 18 REQUESTS 1. A gamified return system for customers with different credit profiles. a payment revision with an AI-powered prediction mechanism By combining these, a task-based credit score of 5 is created for customers with high credit risk. providing improvement processes, low credit risk and reliable customers to ensure that reward-focused advantages are offered developed; - exchanging data using any remote communication protocol at least one electronic device configured to perform (2), 10 - to be carried out on electronic devices (2), banking of users to manage their operations and access notifications and alerts including at least one mobile application (3) configured to provide And; - using any remote communication protocol, electronic device 15 (2) and connecting with the mobile application (3), customers' past data on credit payment performance, adherence to payment dates, delay times, minimum payment frequency, full payment rate, Detailed financial figures in the form of debts paid off and asset debt amount. Continuously monitoring the data, using this data to analyze users' behavior over time. To measure credit discipline and identify payment behavior patterns. to use for the purpose of the user's mobile application (3) monitoring their behavior in detail, mobile during payment periods Number of logins to the application (3), response time to reminder notifications, campaign engagement rate, frequency of contact with customer service, and 25 To continuously monitor behavioral data in the form of complaint content and this Using data in risk and motivation analysis, age, gender, marital status, education level, region of residence, income level, income parameters such as resource diversity and employment stability By using customer profile analysis, the user's location is 30 unemployment rate, inflation, interest rate by sector and region, 19 macroeconomic parameters such as economic growth data are constantly to monitor, use this external data to perform individual risk scoring, and evaluating data as an environmental risk factor, the customer's bank correspondence with, support requests, mobile application (3) Information in the form of messages is processed using natural language processing algorithms. 5 to analyze stress, anxiety and financial levels in the language used by the customer Identifying signs of boredom, helping users complete tasks rate, reward earning speeds, in-game progression level, and preference tracking data in the form of types of rewards they receive, and using this data Optimizing motivational strategies and payment behavior 10 at least one server configured to analyze influencing factors a system characterized by (4) (1).

2. Data exchange using any remote communication protocol. to perform, run the mobile application (3), receive notifications 15 to access and view and any remote communication protocol electronic configured to connect to the server (4) using A system like the one in Claim 1, characterized by device (2) (1).

3. To be carried out on electronic devices (2), users' banking 20 to manage their operations and access notifications and alerts characterized by the mobile application (3) structured to provide a system like any of the above requests (1).

4. Using any remote communication protocol, electronic device (2) 25 and connecting with the mobile application (3), customers’ past credit history data on payment performance, payment due date, delay durations, minimum payment frequency, full payment rate, closed Detailed financial data in the form of debts and asset debt ratios are constantly being collected. monitoring this data, users' credit discipline over time 30 to measure and infer payment behavior patterns from the above requests characterized by the configured server (4) a system like any other (1).

5. Detailed user behavior on the mobile application (3) to monitor in this way, number of logins to the mobile application (3) during payment periods, 5 Response time to reminder notifications, campaign engagement rate, customer behavioral aspects such as frequency of communication with services and content of complaints continuously monitor the data and use this data in risk and motivation analysis. characterized by the server (4) configured for use a system like any of the above requests (1). 10 6. Age, gender, marital status, education level, region of residence, income level, parameters such as income source diversity and employment stability Server configured to analyze customer profile using (4) a 15 as in any of the above claims characterized by system (1).

7. Unemployment rate, inflation, specific to the user's sector and region, macroeconomic parameters such as interest rates and economic growth data continuously monitoring, using this external data to create individual risk scores, and these 20 structured to evaluate data as environmental risk factors any of the above requests characterized by the server (4) such a system (1).

8. Customer correspondence with the bank, support requests, mobile application 25 (3) information in the form of internal messaging with natural language processing algorithms to analyze stress, anxiety and financial difficulties in the language used by the customer characterized by the server (4) configured to detect its symptoms. a system like any of the above-mentioned requests (1). 21 9. Users' task completion rate, reward acquisition speed, in-game tracking data such as progress level and preferred types of rewards, With this data, we can optimize motivational strategies and payment behavior. with the server (4) configured to analyze the influencing factors a 5 as in any of the above characterized claims system (1).

10. The hidden feelings and behaviors of the customer in their written communications. advanced transducer-based natural language processing for extracting signals using their models, explanations in deferment applications, support 10 automatically analyze messages and expressions within the mobile application (3) by observing emotional tones, indicators of financial stress, and risky behavior. Characterized by the server (4) configured to detect its signals. a system like any of the above-mentioned requests (1).

11. Using supervised machine learning models to analyze the customer's financial and processing behavioral characteristics, the likelihood of payment delay, postponement calculating the probability of application, predictions of score changes, variables By ranking the factors in order of importance, which factors carry a higher risk score? 20 with server (4) configured to dynamically report the impact. as in any of the above characterized claims system (1).

12. Long-term trend in customer credit payment time series. to conduct analyses, using time series analysis; 25 future periods possible delays, requests for postponement, or the possibility of early closure Predicting seasonality, habit changes, and with LSTM networks. to incorporate external economic impacts into the artificial intelligence model from the above requests characterized by the configured server (4) a system like any other (1). 30 22 13. With reinforcement learning algorithms; when to tell the customer which that gamified tasks will be presented, what rewards will be given, or Dynamically determining which loan revision proposals will be made, Algorithms learn from customer behavior to achieve optimal interaction. 5 with the server (4) configured to enable it to develop its strategy as in any of the above characterized claims system (1).

14. Processing data streams in layers, drawing from 10 natural language processing models. sentiment and content scores, time series predictions generated with LSTM, and By combining data model results, a holistic customer risk profile is created. to create, combine the outputs of multiple models, for the client characterized by the server (4) configured to calculate weighted scores. a system like any of the above requests (1). 15 15. Combining all model outputs and using Reinforcement Learning algorithms. processing, current risk situation, gamification history and economic environment using their information to provide the most suitable award, revision offer and / or notification. Server (4) configured to determine the action in the form of 20 as in any of the above characterized claims system (1).

16. Any changes in user behavior or financial status. 25 configured to automatically update the risk score when any of the above requests characterized by the server (4) such a system (1).

17. For customers with high credit risk: extending the repayment period and reducing installments. proposals, restructuring with temporary interest rate reduction, 30 in the early stages 23 Enhancing loyalty through small rewards for successful task completion. revision and support in the form of granting the right to defer installments in return. to ensure that mechanisms are proposed, these revisions, user risk Optimized according to score, game progress and overall economic situation. The above 5 is characterized by the server (4) configured to do so. a system like any of the requests (1).

18. For customers with low credit risk; interest rate reduction offers, early payment bonuses, free loan extensions, increased credit limits, and loyalty programs. The program provides motivation and rewards in the form of points or cash prizes. 10 characterized by the server (4) configured to develop strategies a system like any of the above-mentioned requests (1).

19. User stress level or as determined by natural language processing models. If the motivation is identified, the user will receive more empathetic messages and 15 to ensure that supporting tasks are provided, macroeconomic data by including it in the model, user behavior under different economic scenarios Anticipating changes and designing proactive interventions, for the user Continuously analyze play behavior and assign tasks if motivation decreases. 20 to automatically optimize the difficulty or reward type from the above requests characterized by the configured server (4) a system like any other (1).

20. Credit repayment behavior, demographic characteristics, financial transactions, bank account statements, credit card spending, income and expense statements, previous 25 loan repayment history, credit reports, and similar financial records. data in the form of customer service interactions or petition forms text-based information, website or mobile application (3) use records made by a bank customer through the mobile application (3) Customer 30 included in payment inquiries or support tickets 24 Server configured to collect data in the form of complaints (4) as in any of the above claims characterized by system (1).

21. Application programming within the framework of open banking standards, intermediate 5 providing access to consumer financial data through their faces, openly banking application programming interfaces and data integration through its services, the customer's bank accounts and transaction history securely accessing all of the user's bank accounts and by monitoring credit card transactions in one place, you can create a wealth of data. expand third-party access channels for bank account information to use, payment network for updating credit card payment information. interface integrations with gateways and credit granting systems to provide users with what they do in their mobile applications (3) 15 to collect and record transactions in real time from the above requests characterized by the configured server (4) a system like any other (1).

22. Call center records and support chat logs in customer support systems. The robot's text in the form of correspondence or free-text petitions 20 data, through interactive interfaces or email integrations to collect and analyze this text data using natural language processing methods, Social media plays a role in financial products offered to credit customers. data from demographic statistics or credit rating agencies Data collection also includes third-party data sources in the form of scores obtained. 25 to include in the process, throughout this process, the collected data in their raw form. to retain it temporarily, so that it complies with data protection and compliance policies. server (4) configured to provide access controls as in any of the above characterized claims system (1). 30 23. All collected raw data is made suitable for analysis and modeling. to ensure the models are brought in, cleaned, and transformed. To ensure that the input data is consistent and noise-free, or inconsistent or Identifying and correcting missing data entries, missing 5 in credit payment history If there are any installment values ​​or incorrect date formats, these errors will be averaged out. to eliminate by filling with estimated or last known values, unprocessable Identifying and, if necessary, rejecting extreme values ​​using statistical methods. or characterized by the server configured to convert (4) a system like any of the above requests (1). 10 24. Converting the data into a suitable format for modeling allows the user to... categorical factors such as profession, education level, place of residence, and type of loan. The data is digitized using one-hot encoding, labeling, or embedding methods. converting to format, normalizing or standardizing numerical data 15 scaling with techniques, this process measures the relationship between variables. by eliminating unit differences, machine learning models characterized by the server (4) configured to enable effective learning. a system like any of the above-mentioned requests (1).

25. From time series data in the form of credit payment history; number of delays, to extract attributes such as average payment amount and loss rate from the above requests characterized by the configured server (4) a system like any other (1).

26. To extract textual features from text and interaction data. parsing, supporting records and using natural language processing techniques sentiment analysis, keyword extraction, or customer testimonials to perform text classification processes, for this purpose deep linguistic analysis 30 using models to obtain vector representations from texts 26 from the above requests characterized by the configured server (4) a system like any other (1).

27. In the preprocessing stage, the data is processed with big data processing libraries. by processing a scalable database or machine learning 5 Server configured to transfer libraries in a suitable format (4) as in any of the above claims characterized by system (1).

28. Multidimensional financial data from the preprocessing stage and user 10 various artificial intelligence prediction models using behavioral data to improve, analyze and generate scores, credit repayment and In predicting procrastination, it is necessary to enable financial behavior analysis. Using supervised learning algorithms in the analysis process, supervised learning algorithms based on past credit payment examples 15 whether future payments will be made on time to enable it to learn how to perform the prediction process from the above requests characterized by the configured server (4) a system like any other (1).

29. Regression and classification systems that will work on structured numerical data. or establishing models in the form of sorting algorithms, text or Specialized models for data types in the form of time series data. characterized by the server (4) configured to enable its use. a system like any of the above requests (1). 25 30. For numerical data, logistic regression, random forest, and support methods are commonly used. Using vector machines or tree models in the form of XGBoost, this using typical models; a customer's demographic profile, loan type, previous loans Data such as credit history and monthly spending habits are used in credit scoring. 27 to enable its use in estimating the probability of postponement from the above requests characterized by the configured server (4) a system like any other (1).

31. Using artificial neural networks, which are among the deep learning models, 5 Recurrent neural networks, such as LSTM or GRU, analyze the user's past by evaluating the payment sequence, interruptions in debt repayment or to enable them to learn about changing trends, credit data from the last twenty-four months With an LSTM model that takes the payment sequence as input, for the next month 10 The above is characterized by the server (4) configured to provide a system like any of the requests (1).

32. Using natural language processing-based models for text-based data, BERT, RoBERTa 15, handles support requests or customer feedback. to make sense of it by processing it with transformative architectural models in this way, using models, transforming each text fragment into a vector space, thus regarding a customer's decreasing ability to pay based on the content of their complaint to enable the extraction of clues, sentiment analysis of the text, problem Detecting and inferring keyword clusters related to problem-solving. 20 characterized by the server (4) configured to perform a system like any of the above requests (1).

33. Using different data sources as a combined dataset in model training. to use, generate numerical outputs with the obtained predictions and scores, 25 labeled training sets from historical data for training models characterized by the server (4) configured to create a system like any of the above requests (1). 28 34. Model performance is tested through cross-validation to check for accuracy and precision. Evaluating with various metrics such as AUC indicates good performance. Selecting models and improving them through hyperparameter optimization, continuously periodically revise the models with current data using learning techniques. The above 5 is characterized by the server (4) configured to train. a system like any of the requests (1).

35. Investment and lending platforms, social media data or acquisitions. To enable the use of data sources such as historical data in the model. The above 10 is characterized by the server (4) configured to do so. a system like any of the requests (1).

36. The possibility of a potential repayment delay for each credit customer or Characterized by the server (4) configured to generate a credit score. a system like any of the above requests (1) 15 37. For a customer with a high probability of defaulting on loan payments, the loan... raising the interest rate or requesting new collateral, determining a medium risk level. Offering a customer a course of action such as a payment deferral proposal, Check tax, interest, and alert limits according to predefined workflows. 20 The above is characterized by the server (4) which is configured to do so. a system like any of the requests (1).

38. Which customers will receive the highest repayment collection rate? which additional services will be offered, Multi-Criteria Decision Making methods 25 or solve it with optimization algorithms, total debt repayment with a constrained optimization aimed at maximizing collection which to calculate how much credit limit or discount will be given to the customer from the above requests characterized by the configured server (4) a system like any other (1). 30 29 39. The flexibility of learned models and the financial institution's supervised policy to meet the requirements, a customer's upcoming loan installment decided to send automatic notifications a certain number of days before the payment is due. giving, defining time rules in the form of payment reminder times, 5 If a customer is late on payment, the system detects this and sends an automatic alert. or the server (4) configured to initiate additional seizure proceedings as in any of the above characterized claims system (1).

40. Integrating game principles into business processes, a classic credit payment The tracking system uses game elements in the form of points, badges, and levels. adding points as the user makes a certain number of payments on time or enabling them to earn a badge, paying each scheduled loan installment on time. Enabling paying users to earn points, 15 who pay in advance or a special achievement badge for users who do not delay payments enabling the user to earn points as they accumulate them enabling promotion or offering special advantages, social gaming As an element, presenting leaderboards shows users with similar profiles. By creating competition among them regarding timely payments, performance will be increased to 20 The above is characterized by the server (4) configured to increase the above a system like any of the requests (1).

41. To provide each user with real-time or scheduled information, 25 in the form of payment reminders, status reports, alerts and reward announcements Sending notifications about upcoming loan installments, limit changes, and scores. to communicate notifications in the form of gains or suspicious transaction alerts, To automatically generate a payment reminder message, and to specify the message text, Personalizing this message according to the user's language and preferences, mobile application notification system, text message, email or even call center 30 Server configured to send messages via robot voice systems (4) like any of the above-mentioned claims characterized by system (1).

42. For an urgent payment reminder, both an in-app alert and a short message will appear. while available, sending the monthly financial summary report via email, notification content, user profile information and score status characterized by the server (4) configured to enrich a system like any of the above requests (1).

43. The confidentiality and integrity of all information processed and stored. to maintain accessibility to users' financial and personal data; To protect data security from unauthorized access, breach, and misuse. It is difficult to protect within the framework of legal regulations and industry standards. The above 15 is characterized by the server (4) configured to do so. a system like any of the requests (1).

44. In the form of encryption, authentication, authorization, monitoring, and auditing. implementing multi-layered measures in data transmission and storage processes using encryption techniques, storing sensitive information by encrypting it, 20 Using symmetric encryption algorithms for financial data, user to protect passwords and similar information using hashing algorithms from the above requests characterized by the configured server (4) a system like any other (1).

45. Using authentication methods, firewalls, intrusion detection Using security systems and protection software, regular security checks to make updates and actively counter malicious attempts monitoring, personal data through anonymizing or masking techniques protecting aggregated data used in big data analytics 30 31 Analyzing in an anonymized form, only the minimum necessary data. by maintaining privacy and deleting unnecessary data in a timely manner, ensuring regulatory compliance, credit information or user If there are legal restrictions on data usage in the form of notifications, 5. Creating access logs and transparency reports that comply with these limitations The above is characterized by the server (4) configured to do so. a system like any of the requests (1). 15 25

Images