System for fraud prevention

Abstract

A system for fraud prevention, the please invention extracts user behavior information and analyzes and compares it with the fraud behavior model established in the fraud behavior database, and then determines the level of risk of fraud in the user's behavior, and proposes a fraud prevention plan to the user. Fraud risk warning means that even before the fraud group has gained the user's trust, the user can continue to be warned of the risk of being deceived, so as to prevent fraud.

Description

[Technical Field]

[0001] This invention relates to an anti-fraud system, particularly a system that compares and analyzes continuous behavioral information established by a user's behavior with an established fraud model, thereby providing the user with a warning of fraud risk. [Previous Technology]

[0002] In recent years, fraud cases have emerged one after another. According to statistics, the average amount of money defrauded in Taiwan every day is about 300 to 500 million NT dollars, and more than 11 billion NT dollars per month. In addition to the elderly, the victims of fraud include PhDs, professors, senior managers and political and business celebrities. It is clear that the fraud methods are not easy for ordinary people to identify and prevent.

[0003] Currently, there are many common fraud methods, but most of them have the following stages: (a) contact stage, (b) communication stage, (c) trust stage, (d) baiting stage, and (e) closing stage. Most fraudsters will take the following actions in each stage: (a) Contact stage: Fraudsters use online communication software or phone lists to make their first contact with the victim. (b) Communication stage: Fraudsters and victims establish a communication channel so that they can contact each other at any time. They may become friends or make the victim a target for intimidation or blackmail. (c) Trust stage: Through frequent contact and conversation, fraudsters gradually gain the victim's trust, eventually making the victim believe the fraudster's instructions without question. (d) Baiting stage: Fraudsters deliberately reveal investment opportunities, invite victims to invest and make money together, and make the victims stable profits in the first few investments, further convincing the victims that this is a sure-fire way to make money. Or they may demand or coerce the victim to cooperate with their instructions in order to deposit money or transfer funds at a bank or ATM. (e) Closing the operation: Once the victim has transferred a large sum of money, the fraudster disappears instantly, completing the perfect scam.

[0004] In order to prevent the endless stream of fraud, the government launched the 165 anti-fraud hotline, hoping that the public can call the anti-fraud hotline to verify the authenticity when they find something suspicious. However, victims often only realize they have been scammed when the fraudsters are "closing the net". Therefore, when they call the "anti-fraud hotline" to verify, they have already transferred huge sums of money and been scammed. It is too late. Most victims are unable to recover the money they have been scammed, resulting in huge losses. The reason for this is that users are not on guard during the fraud scheme's setup stage before the fraudsters close the net. They are easily "brainwashed" by the fraudsters' words and have complete trust in the fraudsters.

[0005] Chinese Invention Patent No. I869013 discloses a system and method for preventing fraud via messaging. The system includes: a transceiver, which is connected to a text messaging center and an email server via the transceiver; a storage medium for storing multiple modules; and a processor coupled to the storage medium and the transceiver. The processor accesses and executes the message publishing API module, the fraud prevention assessment module, the fraud prevention detection module, the fraud prevention blocking alarm module, and the message batch sending module stored in the storage medium. The message publishing API module is used to receive the user's message sending request and the message content information, and to set the message publishing schedule.The anti-fraud assessment module is used to evaluate the message content information transmitted through the anti-fraud detection module and send the message assessment result. The anti-fraud assessment module includes at least the following components electrically connected to the anti-fraud detection module: a fraud message history record comparison unit, a fraud keyword retrieval unit, a phishing behavior detection unit, a homophone / variant word fuzzy comparison unit, an AI semantic logic recognition unit, and a suspected fraud content comparison unit. The fraud message history record comparison unit compares the message content information with known fraud message history records. If the message content information... If the message is a known scam, a first value is returned to the anti-fraud detection module. If the message is not a known scam, a second value is returned. The scam keyword search unit searches the message for specific keyword combinations. If a specific keyword combination is found, the first value is returned; otherwise, the second value is returned. The phishing detection unit checks whether the message contains a Uniform Resource Locator (URL). The system checks whether the URL (Resource Locator) or the number of messages sent exceeds a threshold. If the message content contains the URL or the number of messages sent exceeds the threshold, a first value is returned to the anti-fraud detection module. If the message content does not contain the URL and the number of messages sent does not exceed the threshold, a second value is returned to the anti-fraud detection module. The homophone / variant word fuzzy comparison unit uses Natural Language Processing (NLP) technology to fuzzy compare whether there are homophones or variations of known keywords in the message content. If the message content contains the homophone or variation of the known keyword, it calculates the similarity and similarity probability value between the homophone or variation and the known keyword, and returns the similarity probability value to the anti-fraud detection module. The AI ​​semantic logic recognition unit uses artificial intelligence (AI) to... The AI ​​semantic recognition technology calculates the probability that the message content is a fraudulent message and returns the probability to the anti-fraud detection module. The suspected fraud content comparison unit uses natural language processing technology to compare the message content with suspected fraudulent data, calculates the similarity and probability between the message content and the suspected fraudulent data, and returns the probability to the anti-fraud detection module.The anti-fraud detection module is electrically connected to both the message publishing API module and the anti-fraud assessment module. The anti-fraud detection module determines whether a message is fraudulent based on the assessment results. If the message is fraudulent, the anti-fraud detection module blocks the message and sends it to the anti-fraud blocking alarm module. If the message is not fraudulent, the anti-fraud detection module transmits the message to the message batch sending module, which then sends the message to the target via the SMS center. The anti-fraud blocking alarm module is electrically connected to the anti-fraud detection module and sends an alarm notification to the email server when the message is fraudulent.

[0006] Although the anti-fraud system and method disclosed in Patent No. I869013 can retrieve key information or signals such as "fraudulent messages" or "fraudulent keywords" from the APIs and URLs used by the user to remind the victim, the timing of the appearance of "fraudulent messages" or "fraudulent keywords" usually falls within the (d) baiting stage or (e) closing stage of the aforementioned fraudulent method. As mentioned above, in the fraudulent behavior setup stage before the (d) baiting stage in the fraudulent method, the fraudster has completely gained the victim's trust, and the victim has no doubt about the fraudster's instructions. Even if Patent No. I869013 can issue timely warnings in the (d) baiting stage or (e) closing stage, the victim will still choose to believe the fraudster. This can be confirmed by the fact that many victims have difficulty persuading the bank staff or even the police to stop the remittance when they go to the bank to make large withdrawals or remittances. This demonstrates that relying solely on extracting and analyzing commonly used keywords in fraudulent schemes to identify fraudulent activities is insufficient. Current known fraud prevention techniques only warn victims during the (d) bait-laying stage or (e) the closing stage of the fraudulent scheme, which is often too late, leaving victims with a high chance of being successfully scammed and losing money. Similar prior technologies, such as invention patent No. I860448, also rely solely on keyword comparison, just like No. I869013. They fail to provide timely warnings to users regarding the fraudulent scheme's unfolding actions after initial contact, highlighting the need for improvement in current fraud prevention methods and technologies.

[0007] In view of this, the inventor of this case conducted further research and finally revealed the anti-fraud system shown in this invention. [Summary of the Invention]

[0008] The purpose of this invention is to provide an anti-fraud system, comprising: a fraud behavior database for recording numerous fraud behavior models; a behavior information extraction unit, signal-connected to at least one behavior information extraction device, for extracting behavior information of users and fraudsters; a continuous process behavior recording unit for continuously recording the behavior information to form continuous process behavior information; a behavior comparison and analysis unit for extracting each fraud behavior model from the fraud behavior database, comparing and analyzing it with the continuous process behavior information, and a judgment unit for determining the degree to which the continuous process behavior information conforms to each fraud behavior model; and an alert unit for issuing an alert message based on the judgment result of the judgment unit. The information judged by the judgment unit is then analyzed and compared again by the behavior comparison and analysis unit with another new behavior information extracted by the behavior information extraction unit, and the judgment unit makes a new judgment.

[0009] The anti-fraud system disclosed in this invention further includes a behavior information preprocessing unit, which converts, decodes, and compiles the behavior data, and provides the processed behavior information to the continuous process behavior recording unit for recording.

[0010] The anti-fraud system disclosed in this invention, wherein the continuous process behavior recording unit continuously records the behavior information of each user towards a fraud group consisting of a large number of fraudsters, thereby forming a continuous process behavior information; or continuously records the behavior information formed by the user's operation of a program, thereby forming a continuous process behavior information. The anti-fraud system disclosed in this invention, wherein the continuous process behavior recording unit continuously records the behavior information of the user towards the same contact object, thereby forming a continuous process behavior information.

[0011] The anti-fraud system disclosed in this invention, wherein the behavior comparison and analysis unit compares and analyzes each of the continuous process behavior information and each of the fraud behavior models using a predetermined program or artificial intelligence.

[0012] The anti-fraud system disclosed in this invention, wherein the behavioral information acquisition device refers to one or any combination of a computer, mobile phone, portable device, program interface, scanner, photographic equipment, recording equipment, satellite locator, and biometric identifier.

[0013] In the anti-fraud system disclosed in this invention, each fraud behavior model in the fraud behavior database is established and updated by a system administrator or by an artificial intelligence training module.

[0014] The anti-fraud system disclosed in this invention includes behavioral information such as online time information, behavioral habit change information, visitor image record information, user-recorded image information, user footprint information, message receiving and publishing path information, text information, and electronic device usage history information, or any combination thereof.

[0015] The anti-fraud system disclosed in this invention further includes a signal sending unit for sending a warning message to the receiving path specified by the user.

[0016] The anti-fraud system disclosed in this invention includes a warning message which may be one or a combination of text, graphics, sound, termination of the user's electronic device usage, forced termination of the remittance process, automatic notification to a third party set by the user, or notification to the police.

[0017] The desirable entity of the present invention can be made clear from the following description and drawings.

Implementation Method

[0018] Please refer to Figure 1. This invention relates to an anti-fraud system, comprising: a fraud behavior database for recording numerous fraud behavior models; a behavior information extraction unit, signal-connected to at least one behavior information extraction device, for extracting behavior information of users and fraudsters; a continuous process behavior recording unit for continuously recording the behavior information to form continuous process behavior information; a behavior comparison and analysis unit for extracting each fraud behavior model from the fraud behavior database, comparing and analyzing it with the continuous process behavior information, and a judgment unit for determining the degree to which the continuous process behavior information conforms to each fraud behavior model; and an alert unit for issuing an alert message based on the degree to which the continuous process behavior information conforms to the fraud behavior model. The information judged by the judgment unit is then analyzed and compared again by the behavior comparison and analysis unit with another new behavior information extracted by the behavior information extraction unit, and the judgment unit makes a new judgment.

[0019] As shown in Figure 1, the anti-fraud system disclosed in this invention further includes a behavior information preprocessing unit, which is used to convert, decode, and compile the behavior data, such as converting images into text, compiling speech into text, etc., and providing the processed behavior information to the continuous process behavior recording unit for recording.

[0020] The anti-fraud system disclosed in this invention includes a continuous process behavior recording unit that continuously records the behavior information of each user towards a fraud group consisting of a large number of fraudsters, thereby forming a continuous process behavior information; or continuously records the behavior information of the user towards a program, thereby forming a continuous process behavior information.

[0021] The anti-fraud system disclosed in this invention mainly determines the degree of consistency between the user's and fraudster's behavioral patterns and fraud behavior models, or whether keywords listed as fraudulent behaviors appear in the behavior, rather than focusing solely on the detection of commonly used fraudulent keywords. The user or fraudster behavioral information referred to in this invention can be one or any combination of the following: internet access time information, changes in behavioral habits information, visitor image record information, user-recorded image information, user footprint information, message receiving and publishing path information, text information, process information of using electronic devices, etc., to establish a continuous process behavioral information. Through the behavior comparison and analysis unit, the continuous process behavioral information is compared and analyzed with each fraud behavior model to determine the degree of consistency with the fraud behavior model, and timely warnings are issued to the user, so as to stop the fraudulent behavior before the fraudster gains the victim's complete trust, thereby achieving the purpose of fraud prevention.

[0022] The anti-fraud system disclosed in this invention includes a behavior information acquisition device, which refers to one or any combination of a computer, mobile phone or portable device, program interface, scanner, photographic equipment, recording equipment, satellite positioning device, biometric identifier, etc., to correspondingly acquire voice behavior information of the user or fraudster, image behavior information of the user's field of vision or surroundings, user's movement behavior information, behavior information of the user or fraudster performing computer operations, call or text message record behavior information, behavior information of the execution program interface (API), or behavior information due to scanning biometric features.

[0023] The anti-fraud system disclosed in this invention uses a judgment unit to make judgments based on a predetermined program or artificial intelligence. This invention does not limit its judgment.

[0024] To achieve de-identification of personal privacy, reduce hardware storage, minimize processing space, and increase system processing speed, this invention does not need to retain all captured data. It only needs to extract behavioral information of key users or fraudsters. Contrary to most software, the less data sampled, the better. Therefore, this behavioral information preprocessing unit can use one or more methods to reduce data volume and increase processing speed, such as: 1. Setting the camera to black and white, 16-color, or low resolution. 2. Automatically capturing computer / mobile phone screens or text using software (API) / hardware (DisplayPort). If it is a screen image, it is also set to black and white or 16-color. 3. Filtering the captured screen image to remove image noise for binarization or image segmentation calculations (software data reduction). 4. Setting the microphone to a low sampling frequency. 5. Converting graphics to text using OCR. 6. Converting speech to text using artificial intelligence speech recognition, extracting key information, and then analyzing it.

[0025] The anti-fraud system disclosed in this invention refers to a behavioral information acquisition device that is built into the user's electronic device, such as a camera, recording device, satellite locator, biometric identifier, etc., built into a smart mobile device. Alternatively, it can refer to a peripheral video device, camera, recording device, satellite positioning receiver, biometric identifier, etc., connected to the user's electronic device. Or, it can refer to a peripheral video device, camera, recording device, satellite positioning receiver, biometric identifier, etc., connected to a cloud server by this invention. This invention is not intended to limit itself.

[0026] The application of the anti-fraud system disclosed in this invention is that after the system is turned on, the continuous process behavior recording unit starts recording the user or fraudster behavior information captured by the behavior information extraction unit, and continuously records and builds a continuous process behavior information. At the same time, the behavior comparison and analysis unit analyzes and compares it with each fraud behavior model in the fraud behavior database, and the judgment unit judges the degree of matching with the fraud model. When the system disclosed in this invention detects that a stranger (or fraudster) is contacting a user for the first time via a messaging app, it judges the user based on a five-stage behavioral pattern established in one of the fraud behavior models in the fraud behavior database: (a) contact stage, (b) communication stage, (c) trust stage, (d) baiting stage, and (e) closing stage. If the stranger's behavior matches the (a) contact stage pattern, this invention can issue a low-risk warning to the user, such as "This person is contacting you for the first time" or "Risk of being scammed is 10%". Afterward, if the stranger and the user add each other as friends on the messaging app and obtain each other's contact information, and the stranger frequently contacts the user daily, with the contact time and frequency showing abnormal increases... Therefore, the analysis and comparison results of this invention match the (b) contact stage of the fraud behavior model. This invention can issue warnings to users such as "You have been in contact with this person unusually frequently, please be careful of being scammed" or "The risk of being scammed is 40%". When, in a very short period of time, the user complies with the other party's instructions (such as uploading photos of their home decoration) or discloses the user's private information (such as salary, savings, asset status, home assets, etc.), this invention judges the user's behavior to match the (c) trust stage of the fraud behavior model. This invention can issue warnings to users such as "Your behavior is overly trusting of others, please do not disclose your personal information to others" or "Your behavior has reached the risk of being scammed of 60%".When the present invention analyzes and compares the content or voice recognition of the program interface API, if the communication behavior information of the user or fraudster mentions investment opportunities, the judgment unit of the present invention can then match these communication behaviors with the (d) baiting stage of the fraud behavior model. The present invention can then issue a warning to the user through the warning unit, such as "Your communication conversation has reached a 90% risk of being defrauded. Please contact the police or discuss with your family." If the present invention subsequently obtains information on the user's behavior that includes online transfer operations, or predicts through the user's footprint information that the user is going to a bank or ATM, or judges through images or voice that the user is exchanging money with others (drivers), the present invention can be considered to match the (e) closing stage of the fraud behavior model. The warning unit can then issue a text warning to the user, such as "You have been defrauded. Please stop the remittance or payment operation. Your risk of being defrauded has reached 100%", or issue an alarm sound, terminate the user's electronic device usage, forcibly stop the remittance process, or automatically notify the third party or police unit set by the user for reporting.

[0027] The anti-fraud system disclosed in this invention can determine that a user is engaging in suspected money transfer activities, such as entering bank account numbers, ID card numbers, or passwords, when video or video recording devices capture images of a user continuously using keypad numbers. This information is then used to establish corresponding behavioral data. Furthermore, when a user is using a mobile phone or communicating with a fraudster via a messaging app like LINE, and the user's location is recorded by a satellite locator as either heading to or currently in a bank or convenience store, this invention can determine that the user is suspected of making money transfers according to the fraudster's instructions. The system identifies the act of making or receiving money to establish corresponding behavioral information. As explained above, the system disclosed in this invention makes a comprehensive judgment based on the behavior of the user or the fraudster, which serves as the basis for determining the degree of fraud suffered by the user. This is not based on whether the user uses keywords commonly used in fraudulent schemes. From the moment the fraudster first contacts the victim, this invention can analyze and compare fraudulent behavior models to continuously provide the user with prompts or reminders regarding the degree of fraud suffered. This is to prevent the user from completely trusting the fraudster during a series of psychological and persuasive fraud schemes, which could ultimately lead to being deceived and suffering huge financial losses.

[0028] The anti-fraud system disclosed in this invention, when analyzing and recording the continuous process behavior information of users, if the user's behavior information is represented by a time-fraud severity coordinate graph, as shown in Figure 2, where the horizontal axis is the time axis and the vertical axis is the fraud severity axis, the fraud severity can be roughly divided into low fraud severity (L), medium fraud severity (M) and high fraud severity (H). When the user's fraud severity increases over time from low fraud severity (L) to high fraud severity (H), the anti-fraud system disclosed in this invention can issue a warning to the user each time the user uses the system or enters a higher level of fraud severity.

[0029] In the normal process of making friends, the two parties may eventually become good friends who trust each other. As shown in Figure 3, after the user and the stranger have their first contact, they eventually reach a level of mutual trust through frequent contact. Although this meets the (a) contact stage, (b) contact stage and (c) trust stage established in the aforementioned example fraud behavior model, it can be clearly seen from the user's continuous process behavior information in Figure 3 that the other party has not fallen into a high level of fraud (H), that is, the behavior information of the suspected (d) baiting stage. Therefore, the warning reminders of the present invention can be maintained at the level of moderate fraud (M) for the user, and can issue a message such as "This anti-fraud system reminds you that you and the other party have reached a level of mutual trust. Please do not disclose your personal information to others." Because this invention analyzes and judges a series of continuous behavioral information of the user, it can intervene in the early stages of fraud schemes, that is, when the fraudster contacts and talks to the user, and continuously issue different levels of warnings to the user. Therefore, when the fraudster offers the user bait such as investment opportunities, it can reduce the probability that the user will suffer huge financial losses due to the trust relationship with the fraudster and complete trust in the fraudster.

[0030] The anti-fraud system disclosed in this invention can also be applied to prevent common fraud perpetrated by impersonating prosecutors or investigation bureaus. This invention can analyze and compare user behavior information to determine if it matches the fraudulent behavior model of a fake prosecutor or investigation bureau member in the following ways: 1. By using a camera, microphone, or mobile phone to capture user behavior information such as meeting or talking with police officers, court clerks, or prosecutors, this behavior information is recorded in a continuous process behavior recording unit. This behavior information is then analyzed and compared with the fraud behavior model in the fraud behavior database. At this point, this invention can issue a warning to the user: "The prosecutor you are contacting may be an imposter. Please verify their identity. Do not easily trust unfamiliar public authority representatives. The fraud level is 20%." 2. Subsequently, by using a recording device or other behavior information capture device to capture user behavior information, if the analysis shows behavior information such as "assisting the police" or "cooperating with the police," this invention can issue a warning to the user: "Do not easily trust the instructions of strangers whose identities have not been verified. The fraud level is 50%." 3. Subsequently, when the analysis of behavioral information acquisition devices such as recording devices reveals information such as "cash flow" or "supervisory account", or when the user's footprint behavior information shows visits to banks or ATMs, this invention can issue a warning to the user, "Please verify the identity of the public authority with your real name first and suspend the remittance behavior. This system determines that the degree of fraud is 100%", or notify a third party set by the user to intervene and persuade.

[0031] The above-described embodiments use a behavior information acquisition device such as a camera, video device or telephone recording device installed in the user's home to obtain the identity of the person who comes into contact with the user. For example, when a visitor presents a prosecutor's, police officer's, or investigation bureau ID or identifies himself as a prosecutor, police officer's, or investigation bureau personnel, the present invention can determine that the person who comes into contact with the user may be a prosecutor, police officer's, or investigation bureau personnel, and thus determine whether the subsequent behavior matches fraudulent behavior.

[0032] As shown in Figure 4, fraudsters (F) usually contact users (U) through communication software such as LINE. The fraud prevention system (100) disclosed in this invention is based on a cloud server. The fraud behavior database (101), the continuous process behavior recording unit (102), and the behavior comparison and analysis unit (103) are connected to the user's mobile device or computer (P) through a mobile application (APP) or user interface. The system can send warning signals to the outside world through the display or speaker of the user's mobile device (such as a smartphone) or computer (P). The behavior information acquisition device of the behavior information acquisition unit (104) refers to the camera, recording, video, satellite positioning device, etc. connected to the user's mobile device or computer (P) or the remote camera, recording, video device connected to the server of this invention. However, the present invention does not limit the signal connection method of the fraud behavior database (101), behavior information extraction unit (104), continuous process behavior recording unit (102), behavior comparison and analysis unit (103) and warning unit.

[0033] As shown in Figure 1, the anti-fraud system disclosed in this invention includes a continuous process behavior recording unit that individually and continuously records the user's behavior information on the same contact object to form continuous process behavior information for a specific contact object, and compares and analyzes the continuous process behavior information of each contact object with each fraud behavior model.

[0034] In the anti-fraud system disclosed in this invention, each fraud behavior model in the fraud behavior database is established and updated by a system administrator or by an artificial intelligence training module.

[0035] The anti-fraud system disclosed in this invention further includes a signal sending unit for sending a warning message to a receiving path specified by the user (e.g., a third-party telephone or SMS message specified by the user), so that the third party can intervene and persuade the user to reduce the chance of being defrauded.

[0036] The anti-fraud system disclosed in this invention includes a warning message which may be one or a combination of text, graphics, sound, termination of the user's electronic device usage, forced suspension of the remittance process, automatic notification to a third party set by the user, or notification to the police.

[0037] Compared with traditional anti-fraud technologies, this invention changes the conventional method and technology of anti-fraud based solely on keywords. Instead, it focuses on analyzing a series of events over a period of time, supplemented by a fraud process progress chart, to alert users to be more vigilant in the early stages, rather than solely analyzing the content of the messages. Therefore, it can effectively alert the user before the fraud is completed. This invention can utilize the following peripheral technologies to collect behavioral information to simulate a lifelike anti-fraud assistant, such as: visual (camera), auditory (microphone), computer / mobile phone screen or text capture, keyboard / mouse monitoring, etc. This allows the invention to simulate a real person to the greatest extent possible, thereby providing anti-fraud advice and preventing users from being "brainwashed" by fraudsters and blindly believing their instructions in the future, thus reducing the chance of being scammed.

[0038] The system disclosed in this invention can be modified and applied without departing from the spirit and scope of this invention, and is not limited to the above-described embodiments. [Simplified Explanation of the Diagram]

[0039] Figure 1: System flowchart of the present invention. Figure 2: Coordinate diagram of the points where the present invention matches the characteristics of fraud (I). Figure 3: Coordinate diagram of the points where the present invention matches the characteristics of fraud (II). Figure 4: Application diagram of the present invention.

Claims

1. A fraud prevention system, comprising: A database of fraudulent activities, used to record most fraudulent behavior models; A behavior information acquisition unit, signal-connected to at least one behavior information acquisition device, is used to acquire behavior information of users and fraudsters; a continuous process behavior recording unit is used to continuously record the behavior information to form continuous process behavior information; a behavior comparison and analysis unit is used to acquire each fraud behavior model from the fraud behavior database, compare and analyze it with the continuous process behavior information, and a judgment unit determines the degree to which the continuous process behavior information matches each fraud behavior model; and an alert unit issues an alert message based on the judgment result of the judgment unit; the information judged by the judgment unit is then analyzed and compared again with another new behavior information acquired by the behavior information acquisition unit in the behavior comparison and analysis unit, and the judgment unit makes a new judgment.

2. The anti-fraud system as described in claim 1 further includes a behavior information preprocessing unit for converting, decoding, and compiling the behavior data, and providing the processed behavior information to the continuous process behavior recording unit for recording.

3. The anti-fraud system as described in claim 1, wherein the continuous process behavior recording unit continuously records each of the user's behavioral information on a fraud group consisting of a large number of fraudsters to form continuous process behavior information; or continuously records each of the user's behavioral information on a program to form continuous process behavior information.

4. The anti-fraud system as described in claim 1, wherein the continuous process behavior recording unit continuously records the user's behavior information on the same contact object individually to form a continuous process behavior information.

5. The anti-fraud system as described in claim 1, wherein the judgment unit makes judgments using a predetermined program or artificial intelligence.

6. The anti-fraud system as described in claim 1, wherein the behavioral information acquisition device refers to one or any combination of a computer, mobile phone or portable device, program interface, scanner, photographic equipment, recording equipment, satellite locator, and biometric identifier.

7. The anti-fraud system as described in claim 1, wherein each fraud behavior model in the fraud behavior database is created and updated by a system administrator or by an artificial intelligence training module.

8. The anti-fraud system as described in claim 1, wherein the behavioral information refers to one or any combination of the following: online time information, behavioral habit change information, visitor image record information, user-recorded image information, user footprint information, message receiving and publishing path information, text information, and electronic device usage history information.

9. The anti-fraud system as described in claim 1, wherein the alert unit further includes a signal sending unit for sending a warning message to a receiving path specified by the user.

10. The anti-fraud system as described in claim 1, wherein the warning message refers to one or any combination of text, graphics, sound, termination of the user's electronic device usage, forced suspension of the remittance process, automatic notification to a third party designated by the user, and notification to the police.

Images