To facilitate your discussion and sharing, please click "Follow" before reading. I will update the quality content daily, thank you for your support~
Wen 丨Sun Hu's study
Editor丨Sun Hu's study
Artificial intelligence technology applied in the financial field can be divided into several levels.
The basic layer refers to the layer that can provide the underlying basic software and hardware and computing power.
The artificial intelligence algorithm framework, intelligent hardware and system software are mainly to support the training, reasoning and application of artificial intelligence algorithms.
Using the resources provided by the basic layer, a series of general technologies are developed for the needs of the application layer in financial scenarios, which provides complete technical means for solving problems in financial scenarios, and provides comprehensive technical capabilities from the aspects of perception, cognition and process automation.
The application layer is an industry scenario generated by the integration of technology and business requirements.
The application layer, in general, mainly provides different intelligent solutions for business scenarios, which is a collection of multiple common layer technologies.
Artificial intelligence has a profound impact on the financial industry through hierarchical technological empowerment.
1. Basic layer:
The basic layer provides basic computing power, tools, and systems for algorithms by providing underlying resources to support efficient business implementation.
2. General layer:
Solve business pain points and create core application value.
(1) Biometric identification
Biometric technology, through camera capture, feature extraction, and then classification of features to complete matching.
With the online development of modern financial services, more and more business scenarios require remote identity verification and authentication of customers, and biometric technology has blossomed everywhere in the financial field.
(2) Computer vision
Computer vision technology refers to the process of using equipment to capture the information generated in the image or video produced by the observation object, and analyzing and processing it in a way that the computer can understand, so that the computer can identify the target.
(3) Knowledge graph
Knowledge graph is generally used to describe the relationship between information resources and resource carriers, and the relationship between the two generally includes mining, analysis, construction and display.
In financial application scenarios, knowledge graph technology, NLP and big data analysis are highly correlated.
In view of the original usage habits of securities firms, banks and other financial institutions, Databank Technology can effectively integrate the internal and external information of financial institutions, on the one hand, it can realize the orderly accumulation and knowledge precipitation of financial institutions' data, on the other hand, it can break information islands through the panorama, connect isolated data to improve the value of data analysis, and continuously extend the compatibility and correlation of industrial chain data.
At present, the SAM industry chain includes more than 1 million subdivided industry nodes, more than 10,000 standardized product nodes that can connect data, and more than 10.
Ten thousand sets of upstream and downstream industry relationships help financial institutions establish the tracking capability of intelligent industrial data, and further help financial institutions realize scenarios such as intelligent investment and research, intelligent risk control and intelligent marketing.
(4) Natural language processing
Natural language processing technology refers to the use of computer processing and analysis of human natural language, which can greatly improve the ability to obtain data, clean data, and deep processing data through natural language processing and knowledge graph to various forms of information, such as text information.
(5) Intelligent voice
Intelligent voice technology enables human-machine language communication, creating smarter customer interaction patterns, including speech recognition technology (ASR) and speech synthesis technology (TTS).
(6) RPA for robotic process automation
RPA, also known as robot process automation, is a technical application that helps robots simulate the interaction process between humans and computers by writing software to realize the automatic execution of workflows.
The core value of RPA is to realize the automation and intelligence of business processes, and RPA technology can be applied as long as it meets a large number of repetitive and clear rule feature scenarios in specific applications.
The financial industry is currently one of the main application fields of RPA, and RPA is used in many fields of the financial industry because of its high applicability in repetitive work processing.
In the insurance industry, RPA can complete basic work such as contract document submission, risk control index monitoring, and system liquidation.
In the securities industry, RPA can also be used to realize manual automatic market opening and closing, market opening monitoring and regular inspection.
(7) Others
With the continuous update and iteration of artificial intelligence technology, emerging cutting-edge technologies such as pre-training models and virtual digital humans are gradually making the financial industry possible.
In the high-frequency business scenario of the financial industry, NLP technology is mainly used to extract keywords for text or audio and perform high-precision sentiment analysis.
In the financial field, virtual digital humans can play a role in virtual reception, guiding business handling, answering business questions and other business scenarios.
In addition, many new research fields of artificial intelligence, such as knowledge computing, brain-computer interface, and multimodal analysis, will continue to release new value in the application of intelligent scenarios in finance with the continuous improvement and breakthrough of technology research and development and application capabilities.
Third, the application layer:
Combined with business needs, intelligent financial scenarios are derived.
(1) Intelligent marketing
Intelligent marketing refers to the use of artificial intelligence, Internet of Things, computers and Internet communication and other modern science and technology, through the establishment of customer portraits, map construction, from intelligent customer acquisition to precise policy to reach a new model of intelligent marketing.
Using artificial intelligence technologies such as knowledge graph and NLP, financial institutions can not only convert a large amount of customer information stored by themselves into marketing resources through in-depth data analysis, but also achieve accurate marketing positioning according to user portraits, and dig deep into the potential needs of customers on this basis, which greatly improves the conversion rate of marketing.
At the same time, RPA helps enterprises complete marketing automation, liberate corporate human resources, and help marketing reduce costs and increase efficiency.
1. Personalized financial services - the core of building inclusive finance
Personalized marketing is an important application scenario of smart finance.
The intelligent recommendation system based on machine learning and knowledge graph can provide customers with personalized marketing services through content recommendation algorithms and collaborative filtering algorithms based on internal data such as customers' historical transaction information, employee service information and product information.
2. GBC linkage marketing - innovative linkage, intelligent customer acquisition means
Banks use knowledge graph technology to build a knowledge graph covering individuals, institutions and legal persons to quickly and accurately locate capital leakage points, innovate and break through the linkage marketing business model, and improve the ability of the whole chain to expand accounts and increase deposits.
(2) Intelligent identification
The application of voiceprint recognition technology effectively makes up for the lack of voice interaction scenarios for face recognition.
Through voiceprint recognition technology, Shengyang Technology helps financial institutions further reduce the risk of loan fraud and insurance fraud.
This not only expands the business coverage of financial institutions, but also effectively promotes the further deepening of inclusive finance.
With the further integration of biometric technology and the financial industry, the authentication technology that has been applied is also being iteratively updated, for example, some city commercial banks hope to replace the existing fingerprint recognition technology with finger vein recognition.
Compared with other biometric technologies, finger vein recognition technology has higher security because it cannot be replicated and is not affected by the external environment.
(3) Intelligent customer service
Intelligent customer service is a technology engine based on large-scale knowledge processing using RPA, NLP, knowledge graph, intelligent voice, machine learning and other technologies.
It can provide semantic-based intelligent answering services to help enterprises complete customer-facing knowledge management.
Intelligent customer service has become the most efficient mode of communication between modern financial institutions and customers.
1. Intelligent customer service system - all-weather processing of professional questions and answers
The intelligent customer service system extracts and analyzes candidate answers by using logic, and then realizes the accurate expression of knowledge vectors through NLP.
It can greatly improve the efficiency of question extraction and semi-structured answers, thereby greatly improving the efficiency of knowledge extraction and semi-structured expression.
The intelligent customer service system effectively solves the pain points of traditional customer service such as large manpower occupation, high cost, difficult to effectively cover during peak business periods, and difficult to effectively integrate resources on multi-channel ports.
With the blessing of knowledge graph technology, not only can the knowledge base be used for professional Q&A, but also reasoning Q&A can be formed through knowledge association.
The intelligent customer service system can also complete the training of machine learning models in daily work, and use high-dimensional vector modeling training data to make the vector expression score of questions and correct answers as high as possible, so as to achieve self-development and self-optimization.
2. Intelligent telephone outbound call system - to create an intelligent remote service system
The smart phone outbound call system integrates intelligent voice, NLP, ASR, TTS and intelligent gateway technologies, which can replace manual completion of error-free multi-round conversations, and the system uses a semantic analysis engine based on NLP technology to decompose customer needs and respond by matching the corresponding scripts in the knowledge base.
In multiple outbound scenarios such as product marketing, overdue collection, risk early warning, information notification, and questionnaire survey, it can quickly and effectively reach user needs, greatly reducing labor costs.
3. Intelligent digital wealth management specialist - build the most intelligent wealth management bank
The intelligent digital wealth management specialist takes the pure online wealth management business as the starting point to realize the whole journey of intelligent services for customers' online wealth management business, including pre-sales product education and interpretation, in-sale product comparison, screening, recommendation and after-sales position analysis.
(4) Robo-advisory
Intelligent investment advisory refers to the construction of data-based models and algorithms based on asset portfolio theory through cloud computing, big data, artificial intelligence and other technical means, and then according to investors' risk appetite, financial status and income goals, input the model created by corresponding variables, so as to automatically form investment recommendations for customers, and can continuously track and dynamically adjust the investment portfolio.
Robo-advisors are more automated, personalized, and lower than human services.
With the blessing of artificial intelligence technology, robo-advisor has established a model that integrates investment analysis algorithms and post-investment automatic management, as long as the customer's investment preference is obtained, it can help customers obtain a customized set of investment portfolios that take into account active and passive investment strategies, so as to achieve effective automated investment management.
(5) Intelligent investment research
The operation mode of intelligent investment research can be basically divided into three steps: the first step is the acquisition of data
The crawler obtains real-time, dynamic, and multi-dimensional data; The second step is data collation, using artificial intelligence technologies such as NLP and knowledge graph to clean, screen, extract and calculate the original data, and then build the corresponding model accordingly;
The third step is data analysis, using the model trained on deep learning of investment business logic and analysis logic to analyze the collated data, find out the connection between core factors, and then make predictions to support investment decisions.
(6) Intelligent claims settlement
Intelligent claims settlement refers to the intelligent insurance claims system built by insurance companies based on artificial intelligence technologies such as machine learning, computer vision, knowledge graph, and intelligent voice to replace traditional labor-intensive operation methods.
Realize RPA automatic management of services such as identity verification, claim document identification, fraud detection, and image loss assessment, and establish a mechanism for pre-event risk warning, claim prevention, in-event fraud detection, accurate loss assessment, and post-event digital compensation.
The biggest pain point in the traditional insurance industry in the claim settlement process is the low claim efficiency and high claim cost, the limitation of manual efficiency makes the determination of claims and the speed of compensation can not meet customer satisfaction, and the lengthy compensation link consumes a lot of inefficient manual labor, which also increases the operating costs of the insurance industry.
The intelligent claims system obtains on-site images and other data through the mobile terminal after the loss occurs, then completes fraud detection while giving pricing through intelligent algorithm models, and finally completes fully automated business processing through digital compensation.
In this process, the machine learning in the intelligent claims system replaces manual operations through self-learning scripts such as claims and risk control, which can save labor costs and shorten the claim processing time, thereby improving the user experience.
(7) Intelligent risk control
Smart Risk Control will build a comprehensive intelligent financial risk prevention and control system through a number of artificial intelligence technologies such as machine learning, knowledge graph and computer vision, combined with big data and cloud computing.
It is the digitalization, automation and intelligent transformation of operational risk, credit risk, compliance risk, cross-risk and reputational risk.
The essence of intelligent risk control is to achieve a lean risk management model through data-driven risk management and operational optimization.
Therefore, the core of intelligent risk control lies in big data, and the risk control model uses NLP technology to semantically understand the data accumulated by financial institutions themselves, third-party cooperative databases and online public data obtained by crawlers, and labels them through the association model to form a knowledge graph of financial risk control.
The emergence of intelligent risk control marks that the driving force of risk control in the financial industry has shifted from the traditional supervision-driven model represented by compliance risk control to a profit-driven model, reducing the management cost of risk control, improving customer experience, and hedging unknown risks.
1. Credit risk prevention and control - improve business capabilities throughout the process
At present, in the financial field, intelligent risk control applies the OCR customer authorization letter recognition function, and the accuracy of early warning in scenarios such as deepening intelligent counterfeiting and anti-counterfeiting, and identifying invalid invoices exceeds the accuracy of manual identification.
2. Intelligent anti-fraud prevention and control - ensure the security of transactions throughout the process
Banking anti-fraud practices.
By using the knowledge graph to carry out automatic monitoring of the flow of credit funds, banks prevent the flow of funds in loans to prohibited fields such as real estate, and deploy ultra-high-dimensional intelligent models of transaction anti-fraud in online banking personal transfer scenarios.
(8) Intelligent compliance
Smart compliance technology is an application based on cognitive computing.
It uses machine-readable rules, facilitates the use of standardized rules through digital regulatory protocols, and reduces interpretation and interpretation errors by using standardized rule sets, thereby helping financial institutions monitor their own business processes for continuous compliance review and compliance assessment.
(9) Intelligent operation
Intelligent operation is based on artificial intelligence perception and cognitive technology, and with the blessing of RPA technology, it improves business standardization and strengthens the intensive utilization of resources by centrally managing similar business processes, and realizes platform sharing.
Obtain intensive effects, reduce the operating costs associated with business processing, release the intrinsic efficiency of data assets, and help financial institutions upgrade core business scenarios.
RPA Digital Workforce – Improve operational process automation
Using technologies such as robotic process automation to build RPA digital employees, replace manual completion of simple repetitive operations such as information entry, verification, and submission, promote cost reduction and efficiency increase, strengthen risk prevention and control, and help business expansion, and it is estimated that this technology can release hundreds of people's annual workload every year.
bibliography