Sanction scanning method and device for transactions

Abstract

The embodiments of this specification provide a method and device for scanning sanctions for transactions. The method involves a scanning engine, which is associated with multiple review robots, including a first review robot. The method includes: the scanning engine sends the current review task generated by it to the first review robot, and the current review task is generated according to the sanction scanning request for the current transaction; the first review robot searches for a target manual review task that is the same as the current review task in a local database, wherein the local database stores a number of manual review tasks and the review results corresponding to the several manual review tasks; if found, the review result corresponding to the target manual review task is returned to the scanning engine; if not found, a decision-making failure message is returned to the scanning engine.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of computer technology, and in particular, to a method and device for scanning sanctions against transactions. Background Art

[0002] At present, in the sanctions scanning business for transactions, in order to ensure comprehensive coverage of sanctions risks, the scanning engine usually makes some sacrifices in accuracy. Therefore, in the actual sanctions scanning business, analysts usually review the review tasks produced by the scanning engine to avoid misaudits and customer complaints due to accuracy factors. However, a large number of sanctions scanning requests usually result in the output of a large number of review tasks, and the labor cost and review efficiency of the review face huge challenges.

[0003] Therefore, there is an urgent need for a reasonable and reliable solution that can save the manpower cost of the trial and improve the trial efficiency in the sanctions scanning business while ensuring the accuracy of the trial. Summary of the invention

[0004] The embodiments of this specification provide a sanctions scanning method and device for transactions, which can save the human cost of the review and improve the review efficiency in the sanctions scanning business while ensuring the accuracy of the review.

[0005] In the first aspect, an embodiment of the present specification provides a sanction scanning method for transactions, involving a scanning engine, wherein the scanning engine is associated with multiple review robots, and the multiple review robots include a first review robot. The method includes: the scanning engine sends a current review task generated by it to the first review robot, and the current review task is generated according to a sanction scanning request for the current transaction; the first review robot searches a local database for a target manual review task that is the same as the current review task, wherein the local database stores a number of manual review tasks and the review results corresponding to the several manual review tasks; if found, the review result corresponding to the target manual review task is returned to the scanning engine; if not found, a decision-making failure message is returned to the scanning engine.

[0006] In some embodiments, the current review task includes several scanning parameters and hit parameters corresponding to the several scanning parameters respectively, the several scanning parameters are derived from the current transaction, the hit parameters match the corresponding scanning parameters and are included in a preset sanctions list; the multiple review robots also include a second review robot, and the method also includes: the scanning engine sends the current review task to the second review robot in response to receiving an inability to make a decision message returned by the first review robot; the second review robot selects a scanning parameter from the several scanning parameters as the first scanning parameter, and performs feature extraction on the first scanning parameter and its corresponding hit parameter to obtain the extracted first feature information; predicts the sub-review result corresponding to the first scanning parameter based on the first feature information; if the sub-review result is not predicted, returns an inability to make a decision message for the current review task to the scanning engine.

[0007] In some embodiments, after predicting the sub-review result corresponding to the first scanning parameter based on the first characteristic information, it also includes: if the sub-review result is predicted and the sub-review result indicates rejection, a review result indicating rejection is generated for the current review task, and the generated review result is returned to the scanning engine.

[0008] In some embodiments, after predicting the sub-review result corresponding to the first scanning parameter according to the first feature information, it also includes: if the sub-review result is predicted and the sub-review result indicates pass, determining whether there are any unselected scanning parameters among the plurality of scanning parameters; if the determination result is yes, selecting a scanning parameter from the unselected scanning parameters as the first scanning parameter, and continuing to perform feature extraction on the first scanning parameter and its corresponding hit parameter; if the determination result is no, generating a review result indicating pass for the current review task, and returning the generated review result to the scanning engine.

[0009] In some embodiments, the method further includes: the scanning engine, in response to receiving the unable to make a decision message returned by the second review robot, sending the current review task to a review terminal used by a relevant reviewer;

[0010] The scanning engine receives the review result of the reviewer on the current review task from the review end.

[0011] In some embodiments, the second review robot saves feature-related data, which includes a number of second feature information and scores corresponding to the second feature information respectively, where the scores are rejection rates or pass rates; the second feature information and its corresponding scores are generated according to a number of manual review tasks and the review results corresponding to the manual review tasks respectively.

[0012] In some embodiments, predicting the sub-review result corresponding to the first scanning parameter based on the first characteristic information includes: searching for target second characteristic information matching the first characteristic information among the plurality of second characteristic information; if the target second characteristic information is found, determining whether the score corresponding to the target second characteristic information reaches a score threshold; if it is determined that the score corresponding to the target second characteristic information reaches the score threshold, generating the sub-review result corresponding to the first scanning parameter based on the score corresponding to the target second characteristic information.

[0013] In some embodiments, predicting the sub-review result corresponding to the first scanning parameter based on the first characteristic information also includes: if the target second characteristic information is not found, or it is determined that the score corresponding to the target second characteristic information does not reach the score threshold, it is determined that the sub-review result cannot be predicted for the first scanning parameter.

[0014] In some embodiments, the feature-related data also includes several patterns, any pattern corresponds to at least one of the several second feature information, and is also associated with a feature extraction strategy; and the feature extraction of the first scanning parameter and its corresponding hit parameter includes: determining the target pattern that the first scanning parameter and its corresponding hit parameter conform to among the several patterns; performing feature extraction on the first scanning parameter and its corresponding hit parameter using the feature extraction strategy associated with the target pattern; searching for the target second feature information that matches the first feature information among the several second feature information includes: searching for the target second feature information that matches the first feature information in the second feature information corresponding to the target pattern.

[0015] In some embodiments, the feature-related data also includes several scanning categories, any scanning category corresponds to at least one of the several modes; and determining the target mode that the first scanning parameter and its corresponding hit parameter conform to in the several modes includes: determining the target scanning category that the first scanning parameter conforms to in the several scanning categories; and determining the target mode that the first scanning parameter and its corresponding hit parameter conform to in the mode corresponding to the target scanning category.

[0016] In some embodiments, the sub-review result corresponding to the first scanning parameter is generated based on the score corresponding to the target second characteristic information, including: if the score corresponding to the target second characteristic information is a rejection rate, then a sub-review result corresponding to the first scanning parameter and indicating rejection is generated; if the score corresponding to the target second characteristic information is a pass rate, then a sub-review result corresponding to the first scanning parameter and indicating pass is generated.

[0017] In some embodiments, after receiving the review result of the current review task by the reviewer from the review end, it also includes: the scanning engine takes the current review task as a manual review task, and updates the manual review task and its corresponding review result to the local database.

[0018] In some embodiments, after receiving the review result of the current review task by the reviewer from the review end, it also includes: the scanning engine takes the current review task as a manual review task, and updates the data of the second review robot according to the manual review task and its corresponding review result.

[0019] In some embodiments, the method also includes: the scanning engine obtains a number of review tasks reviewed by the second review robot, and the review results corresponding to the number of review tasks, and sends the number of review tasks to the review end used by relevant review personnel; the scanning engine receives the review results of the number of review tasks by the review personnel from the review end; the scanning engine determines whether the review results of the second review robot and the review personnel for any review task among the number of review tasks are consistent, and when the result is determined to be no, the second feature information referenced in the feature-related data for the any review task is invalidated.

[0020] In some embodiments, the invalidation processing of the second feature information referenced by the arbitrary trial task in the feature-related data includes: resetting the score corresponding to the second feature information referenced by the arbitrary trial task to zero.

[0021] In some embodiments, the second feature information referenced by the arbitrary review task includes a number of cumulative features; and the invalidation processing of the second feature information referenced by the arbitrary review task in the feature-related data also includes: resetting the numerical values ​​of the several cumulative features to zero.

[0022] In some embodiments, the method also includes: the scanning engine obtains at least one second feature information that meets preset quality inspection conditions from the feature-related data, and the scores corresponding to the at least one second feature information; the scanning engine sends the at least one second feature information and the scores corresponding to them to the quality inspection terminal used by relevant quality inspectors, so that the quality inspectors can perform quality inspection.

[0023] In the second aspect, an embodiment of the present specification provides a sanction scanning method for transactions, which is applied to a first review robot, and the first review robot is included in multiple review robots associated with a scanning engine. The method includes: receiving a current review task sent by the scanning engine, and the current review task is generated by the scanning engine according to a sanction scanning request for the transaction; searching for a target manual review task that is the same as the current review task in a local database, wherein the local database stores a number of manual review tasks and the review results corresponding to the several manual review tasks; if found, the review result corresponding to the target manual review task is returned to the scanning engine; if not found, a decision-making failure message is returned to the scanning engine.

[0024] In a third aspect, an embodiment of the present specification provides a sanction scanning method for a transaction, which is applied to a second review robot, wherein the second review robot is included in a plurality of review robots associated with a scanning engine, wherein the plurality of review robots also include a first review robot, and the method includes: receiving a current review task sent by the scanning engine, wherein the current review task is sent when the first review robot does not store the same target manual review task, wherein the current review task is generated by the scanning engine according to a sanction scanning request for the current transaction, and includes a plurality of scanning parameters and hit parameters corresponding to the plurality of scanning parameters, wherein the plurality of scanning parameters are derived from the current transaction, and the hit parameters match the corresponding scanning parameters and are included in a preset sanction list; selecting a scanning parameter from the plurality of scanning parameters as a first scanning parameter, and performing feature extraction on the first scanning parameter and the corresponding hit parameter to obtain extracted first feature information; predicting a sub-review result corresponding to the first scanning parameter based on the first feature information; and returning a decision-making failure message for the current review task to the scanning engine if the sub-review result is not predicted.

[0025] In a fourth aspect, an embodiment of the present specification provides a sanction scanning device for transactions, which is applied to a first review robot, wherein the first review robot is included in a plurality of review robots associated with a scanning engine, and the device comprises: a receiving unit, configured to receive a current review task sent by the scanning engine, wherein the current review task is generated by the scanning engine according to a sanction scanning request for the current transaction; a searching unit, configured to search a local database for a target manual review task that is the same as the current review task, wherein the local database stores a number of manual review tasks and the review results corresponding to the several manual review tasks; an output unit, configured to return the review result corresponding to the target manual review task to the scanning engine if found; and if not found, return a decision-making failure message to the scanning engine.

[0026] In a fifth aspect, an embodiment of the present specification provides a sanction scanning device for transactions, which is applied to a second review robot, wherein the second review robot is included in a plurality of review robots associated with a scanning engine, wherein the plurality of review robots also include a first review robot, and the device includes: a receiving unit, configured to receive a current review task sent by the scanning engine, wherein the current review task is sent when the first review robot does not store the same target manual review task, wherein the current review task is generated by the scanning engine according to a sanction scanning request for the current transaction, and includes a plurality of scanning parameters and hit parameters corresponding to the plurality of scanning parameters, wherein the plurality of scanning parameters are derived from the current transaction, and the hit parameters match the corresponding scanning parameters and are included in a preset sanction list; a feature extraction unit, configured to select a scanning parameter from the plurality of scanning parameters as a first scanning parameter, and perform feature extraction on the first scanning parameter and the corresponding hit parameter to obtain extracted first feature information; a prediction unit, configured to predict a sub-review result corresponding to the first scanning parameter based on the first feature information; and an output unit, configured to return a decision-making failure message for the current review task to the scanning engine if the sub-review result is not predicted.

[0027] In a sixth aspect, an embodiment of the present specification provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed in a computer, the computer is caused to execute the method described in any one of the implementations in the second and third aspects.

[0028] In a seventh aspect, an embodiment of the present specification provides a computing device, comprising a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method described in any one of the implementation modes in the second aspect and the third aspect is implemented.

[0029] In an eighth aspect, an embodiment of the present specification provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the method described in any implementation manner in the second aspect and the third aspect.

[0030] The solution provided in the above-mentioned embodiments of this specification involves a scanning engine, which is associated with multiple review robots. The scanning engine can use the review robots among the multiple review robots associated with it to perform intelligent review on the current review task, thereby saving the manpower cost of the review and improving the review efficiency.

[0031] In addition, the multiple review robots include a first review robot, and the local database of the first review robot stores a number of manual review tasks and the review results corresponding to the manual review tasks. It should be understood that the manual review tasks are review tasks that have been manually reviewed, and the review results corresponding to the manual review tasks are the results of manual review, which have high accuracy. By using the first review robot to review the current review task, the scanning engine can enable the first review robot to make a decision directly when the current review task is a repeated task or a retry task, that is, when the current review task is the same as a manual review task stored in the local database. This can not only improve the review efficiency, but also ensure the accuracy of the review. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the multiple embodiments disclosed in this specification, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only the multiple embodiments disclosed in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 is an exemplary system architecture diagram to which some embodiments of the present specification may be applied;

[0034] Figure 2 is a flow chart of one embodiment of a sanctions scanning method for a transaction;

[0035] Figure 3 It is a schematic diagram of the trial process of the second trial robot;

[0036] Figure 4 It is a schematic diagram of the structure of a sanction scanning device for transactions;

[0037] Figure 5 It is a structural diagram of a sanctions scanning device for transactions. DETAILED DESCRIPTION

[0038] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. The described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of this application.

[0039] It should be noted that, for the convenience of description, only the parts related to the invention are shown in the drawings. The embodiments and features in the embodiments in this specification can be combined with each other without conflict. In addition, the words "first", "second" and the like in this specification are only used for information distinction and do not play any limiting role.

[0040] As mentioned above, in actual sanctions scanning business, analysts usually review the review tasks generated by the scanning engine to avoid misaudits and customer complaints due to accuracy factors. However, a large number of sanctions scanning requests usually result in a large number of review tasks being generated, and the human cost and review efficiency of the review face huge challenges.

[0041] Based on this, some embodiments of this specification provide a sanctions scanning method for transactions, which can save the human cost of review and improve the review efficiency in the sanctions scanning business while ensuring the accuracy of the review.

[0042] Specifically, Figure 1 An exemplary system architecture diagram applicable to some embodiments of this specification is shown. Figure 1 As shown, the system architecture may include a scanning engine and a plurality of review robots associated with the scanning engine, such as Figure 1 The first trial robot and the second trial robot are shown in FIG.

[0043] In practice, the scanning engine can receive a sanction scanning request for a transaction, and when the transaction meets the review requirements, generate a review task based on the sanction scanning request, and use a review robot among the multiple review robots to review the generated review task.

[0044] The transaction targeted by the sanctions scan request may be a real-time transaction in any business scenario, such as merchant entry, payment or shopping. The transaction may include multiple parameters. For example, a payment transaction may include parameters such as the payee's name and bank card number. It should be noted that the sanctions scan request may include the transaction itself, or include the multiple parameters.

[0045] The scanning engine may store several preset sanctions lists. The several sanctions lists may include, for example, sanctions lists issued by international organizations and / or national organizations. Specifically, the several sanctions lists may include, for example, lists of names, lists of companies, lists of ships, lists of addresses, and / or lists of bank card numbers, etc. It should be noted that the sanctions list may also be referred to as a sanctions blacklist, in which the objects indicated by each list item, such as a person, company, ship, address or bank card number, are objects that have been sanctioned by relevant organizations.

[0046] The above review requirements may, for example, indicate that there are parameters in the transaction that comply with (hit) a sanctions list in the above-mentioned sanctions lists, and the parameter matches a list item in the sanctions list that complies with the parameters.

[0047] The above-mentioned multiple trial robots can be implemented as software modules or servers, etc., respectively, which are not specifically limited here. In addition, the above-mentioned multiple trial robots generally adopt different trial methods.

[0048] For example, the first review robot can make decisions directly on repeated tasks and retry tasks. Specifically, the local database of the first review robot can store a number of manual review tasks, and the review results corresponding to the several manual review tasks. It should be understood that the several manual review tasks are review tasks that have been manually reviewed, and the review results corresponding to the several manual review tasks are the results of manual review. When the first review robot conducts task review, if it recognizes that the review task it is reviewing is the same as a certain manual review task among the several manual review tasks, it can determine that the review task is a repeated task or a retry task, so that it can directly make a decision on the review task, for example, using the review result corresponding to the certain manual review task as the review result of the review task.

[0049] The second review robot may, for example, extract features of the review task it is reviewing, and conduct task review based on the extracted features.

[0050] According to the description of the first trial robot and the second trial robot in the previous text, it can be known that the accuracy of the trial result of the first trial robot is relatively high. Therefore, the scanning engine can, for example, first use the first trial robot to review the generated trial task. Then, when the first trial robot cannot make a decision on the trial task, the second trial robot can be used to review the trial task.

[0051] In one embodiment, the scanning engine can also support manual review. For example, the above system architecture can also include a review terminal used by relevant reviewers. When the above multiple review robots are unable to make a decision on the review task, the scanning engine can send the review task to the review terminal, and the reviewer can manually review the review task, so as to obtain the review result of the reviewer on the review task.

[0052] Below, take the current transaction as an example. Figure 1 As shown, the scanning engine can receive a sanction scanning request for the current transaction, for example, from a business system corresponding to the business to which the current transaction belongs, or from other systems (such as a risk control system, etc.). After determining that the current transaction meets the above review requirements according to the sanction scanning request, the scanning engine can generate a current review task for the current transaction.

[0053] Then, the scanning engine can send the current trial task to the first trial robot as shown by label 102, and the first trial robot will review the current trial task. After completing the trial, the first trial robot can provide information feedback to the scanning engine, for example, it can return the trial result or a message that a decision cannot be made to the scanning engine as shown by label 104. If the scanning engine receives the message that a decision cannot be made returned by the first trial robot, it can send the current trial task to the second trial robot as shown by label 106, and the second trial robot will review the current trial task. After completing the trial, the second trial robot can also provide information feedback to the scanning engine, for example, it can return the trial result or a message that a decision cannot be made to the scanning engine as shown by label 108.

[0054] Next, if the scanning engine receives a decision-making failure message returned by the second review robot, the current review task can be sent to the review terminal used by the relevant review personnel as shown in label 110, and the review personnel can manually review the current review task. For example, the review terminal can provide the current review task to the review personnel as shown in label 112. The review personnel can manually review the current review task, and then submit the review result of the current review task to the review terminal as shown in label 114. Then, the review terminal can return the review result to the scanning engine as shown in label 116.

[0055] By adopting the above-described review process, it is possible to save the human cost of review and improve the review efficiency in the sanctions scanning business while ensuring the accuracy of the review.

[0056] The specific implementation steps of the above method are described below in conjunction with specific embodiments.

[0057] See also Figure 2, which is a schematic diagram of an embodiment of a sanction scanning method for a transaction. The method involves a scanning engine, the scanning engine is associated with a plurality of trial robots, and the plurality of trial robots include a first trial robot. The method comprises the following steps:

[0058] Step 202, the scanning engine sends the current review task generated by it to the first review robot, and the current review task is generated according to the sanction scanning request for the current transaction;

[0059] Step 204, the first review robot searches the local database for a target manual review task that is the same as the current review task;

[0060] Step 206: If the first review robot finds the target manual review task, it returns the review result corresponding to the target manual review task to the scanning engine; if the target manual review task is not found, it returns a decision-making failure message to the scanning engine.

[0061] In practice, before step 202, the scanning engine may receive a sanction scanning request for the transaction, for example, from a business system corresponding to the business to which the transaction belongs, or from other systems (such as a risk control system, etc.). The transaction may include multiple parameters. The sanction scanning request may include the transaction itself, or include the multiple parameters. After determining that the transaction meets the review requirements as described above according to the sanction scanning request, the scanning engine may generate a current review task for the transaction.

[0062] As an example, the scanning engine may support several scanning categories, which may include, for example, scanning people, scanning companies, scanning ships, and / or scanning addresses, etc. In addition, a single scanning category may correspond to at least one of the several sanctions lists described above. The scanning engine may, for example, determine several parameters to be scanned from multiple parameters of the transaction based on the several scanning categories. It should be understood that the parameters in the several parameters conform to the scanning categories in the several scanning categories. For example, if the transaction includes the names of the people involved in the transaction, and the several scanning categories include scanning people, it can be determined that the name conforms to the scanning category of scanning people, so that the name can be used as a parameter to be scanned.

[0063] For any parameter among the above-mentioned several parameters, the scanning engine can, for example, determine the sanctions list that the arbitrary parameter matches in the above-mentioned several sanctions lists, for example, determine the sanctions list that the arbitrary parameter matches in the sanctions list corresponding to the scanning category that the arbitrary parameter matches, and then match the arbitrary parameter with the list items in the sanctions list, for example, perform similarity calculation. It should be pointed out that the sanctions list that the arbitrary parameter matches can be called the hit list of the arbitrary parameter. After matching the list item that matches the arbitrary parameter from the hit list, it can be determined that the transaction meets the review requirements, so that the arbitrary parameter can be used as a scanning parameter, the list item can be used as the hit parameter corresponding to the scanning parameter, and the scanning parameter and its corresponding hit parameter can be included in the current review task for the transaction.

[0064] Optionally, the current review task may also include other content, such as the business scenario to which the transaction belongs, and / or the language involved in the transaction (such as Chinese or English, etc.), etc., which may be included in the current review task.

[0065] Based on the generation process of the current review task, it can be known that the current review task may include several scanning parameters and hit parameters corresponding to the several scanning parameters respectively, the several scanning parameters are derived from the current transaction, the hit parameters match the corresponding scanning parameters and are included in the preset sanctions list.

[0066] After generating the current review task for the transaction, the scanning engine may send the current review task to the first review robot by executing step 202. The local database of the first review robot may store a number of manual review tasks and the review results corresponding to the manual review tasks. It should be understood that the manual review tasks are review tasks that have been manually reviewed, and the review results corresponding to the manual review tasks are the results of manual review, which have high accuracy.

[0067] Next, in step 204, the first review robot can search the local database for the target manual review task that is the same as the current review task. If the target manual review task is found, the first review robot can determine that the current review task is a repeated task or a retry task, and make a decision directly based on the review result corresponding to the target manual review task, such as determining the review result corresponding to the target manual review task as the review result corresponding to the current review task. If the target manual review task is not found, the first review robot can determine that the current review task is not a repeated task or a retry task, and further determine that no decision can be made for the current review task.

[0068] Next, in step 206, the first review robot may return the review result corresponding to the target manual review task, or a message indicating that a decision cannot be made, to the scanning engine. For example, when the target manual review task is found, the review result corresponding to the target manual review task is returned to the scanning engine; when the target manual review task is not found, a message indicating that a decision cannot be made is returned to the scanning engine.

[0069] exist Figure 2 In the corresponding embodiment, by executing steps 202-206, the scanning engine can use the first review robot to perform intelligent review on the current review task, which can save the human cost of review and improve the review efficiency while ensuring the accuracy of the review in the sanction scanning business.

[0070] In one embodiment, the plurality of trial robots may further include a second trial robot. The second trial robot may adopt a trial method different from that of the first trial robot. If, in step 206, the first trial robot returns a message indicating that a decision cannot be made to the scanning engine, the scanning engine may then use the second trial robot to perform intelligent trial on the current trial task.

[0071] Specifically, after step 206, the following steps may be performed: Figure 2 The steps within steps 208-226 shown in .

[0072] In step 208, the scanning engine may send the current review task to the second review robot in response to receiving the decision-making failure message returned by the first review robot. It is understandable that the current review task may be sent when the first review robot does not store the same target manual review task.

[0073] Next, the second review robot may, for example, extract features of the current review task, and conduct task review based on the extracted features, and then provide information feedback to the scanning engine, such as returning a review result or a decision-making failure message.

[0074] Specifically, according to the description in the foregoing text, the current review task may include a number of scanning parameters and hit parameters corresponding to the scanning parameters. The second review robot may select a scanning parameter from the scanning parameters as the first scanning parameter by executing step 210, and perform feature extraction on the first scanning parameter and its corresponding hit parameter by executing step 212 to obtain extracted first feature information. Here, the second review robot may, for example, use a preset feature extraction strategy to perform feature extraction on the first scanning parameter and its corresponding hit parameter.

[0075] Next, in step 214, the second review robot may predict the sub-review result corresponding to the first scanning parameter according to the first feature information.

[0076] As an example, the second review robot may store feature-related data, which may include a number of second feature information and scores corresponding to the second feature information, respectively, and the scores may be a rejection rate or a pass rate. The second feature information and the scores corresponding thereto may be generated according to a number of manual review tasks and the review results corresponding to the manual review tasks.

[0077] Based on this, in step 214, the second review robot can search for the target second feature information that matches the first feature information in the above-mentioned plurality of second feature information. If the target second feature information is found, it can be determined whether the score corresponding to the target second feature information reaches the score threshold. If it is determined that the score reaches the score threshold, a sub-review result corresponding to the first scanning parameter can be generated according to the score.

[0078] For example, if the score is a rejection rate, a sub-trial result corresponding to the first scanning parameter and indicating rejection can be generated, and then step 218 can be executed. If the score is a pass rate, a sub-trial result corresponding to the first scanning parameter and indicating pass can be generated, and then step 220 can be executed. It should be noted that "reject" can mean that there is a risk and needs to be blocked. "Pass" can mean that there is no risk and needs to be released.

[0079] In addition, if the target second characteristic information is not found, or it is determined that the score corresponding to the target second characteristic information does not reach the score threshold, it can be determined that the sub-examination result cannot be predicted for the first scanning parameter, and then it can be determined that no decision can be made for the current examination task, so step 216 can be executed.

[0080] In step 216, when the second review robot fails to predict the sub-review result corresponding to the first scanning parameter, it may return a decision-making failure message to the scanning engine.

[0081] In step 218, when a sub-review result corresponding to the first scanning parameter is predicted, if the sub-review result indicates rejection, the second review robot can generate a review result indicating rejection for the current review task, and return the review result to the scanning engine by executing step 226.

[0082] In step 220, when the sub-trial result corresponding to the first scanning parameter is predicted, if the sub-trial result indicates passing, the second trial robot can determine whether there are any unselected scanning parameters among the above scanning parameters. If the determination result is yes, step 222 can be executed. If the determination result is no, step 224 can be executed.

[0083] In step 222 , in response to the determination result of step 220 being yes, a scanning parameter may be selected from the above unselected scanning parameters as the first scanning parameter, and step 212 may be continued based on the first scanning parameter.

[0084] In step 224 , in response to the determination result of step 220 being negative, a review result indicating a pass may be generated for the current review task, and the review result may be returned to the scanning engine by executing step 226 .

[0085] According to the description of steps 208-226 in the previous text, it can be known that only when the sub-trial results corresponding to the above-mentioned scanning parameters all indicate passing, the second review robot will give a passing conclusion for the current review task, and the passing conclusion can indicate that the transaction is risk-free and needs to be released. As long as the sub-trial result corresponding to one of the above-mentioned scanning parameters indicates rejection, the second review robot will give a rejection conclusion for the current review task, and the rejection conclusion can indicate that the transaction is risky and needs to be blocked. Moreover, as long as the sub-trial result corresponding to one of the above-mentioned scanning parameters cannot be predicted, the second review robot will not make a decision for the current review task.

[0086] In one embodiment, the feature-related data may further include several patterns, any of which may correspond to at least one of the several second feature information, and may also be associated with a feature extraction strategy. Based on this, step 212 in the foregoing may further include: Figure 3 Step 302 and step 304 shown in FIG.

[0087] Specifically, in step 302, the target pattern that the first scanning parameter and its corresponding hit parameter meet can be determined in the above-mentioned several patterns. In step 304, the feature extraction strategy associated with the target pattern can be used to extract features from the first scanning parameter and its corresponding hit parameter to obtain the extracted first feature information. Accordingly, step 214 in the above text can further include, for example, Figure 3 Step 306 shown in FIG.

[0088] In step 306, the target second feature information matching the first feature information can be searched in the second feature information corresponding to the target mode. In this way, it is possible to avoid matching the first feature information with each piece of second feature information in the feature-related data, which can effectively save feature matching time.

[0089] It should be noted that for any of the above-mentioned patterns, the pattern can reflect the characteristics of the scanning parameters and hit parameters associated with the corresponding second feature information, and the pattern can be generated based on the characteristics. Taking the name as an example, assuming that the scanning parameters include "Liu Yier", the hit parameter corresponding to "Liu Yier" is "Liu Eryi". By observing these two names, it can be seen that the two have the characteristics of the same surname and the same first name in reverse order, so the pattern can be, for example, a "name reversal pattern".

[0090] Next, taking the name reversal pattern as an example, its corresponding second feature information can, for example, indicate that the surname is the same, the given name is the same and the order is reversed. The feature extraction strategy associated with the name reversal pattern can, for example, indicate that the extracted first feature information needs to indicate that the surname is the same, the given name is the same and the order is reversed. Furthermore, the second feature information can also indicate other content, and accordingly, the feature extraction strategy can also indicate that the extracted first feature information needs to indicate other content. For example, the second feature information can also indicate the business scenario to which the associated transaction belongs, the language involved in the transaction, and / or the hit category of the scanning parameter associated with the second feature information (such as a person), etc.; accordingly, the feature extraction strategy can also indicate that the extracted first feature information needs to indicate the business scenario to which the transaction belongs, the language involved in the transaction, and / or the hit category of the scanning parameter, etc.

[0091] Continuing to take the name as an example, for any of the above several patterns, assuming that a certain scanning parameter associated with the second characteristic information corresponding to the pattern is "A1B1C1", and the hit parameter corresponding to the scanning parameter is "A2B2C2", where A1 and A2, B1 and B2 are completely matched or aligned (such as reaching the alignment match threshold), and the word pairs C1 and C2 are not aligned. In this case, the pattern can be a word pair pattern.

[0092] In practice, the second feature information corresponding to the word pair pattern can indicate, for example, the number of word pairs, the number of fully matched word pairs, the number of aligned matched word pairs, the number of unaligned matched word pairs, the number of historical manual review tasks for unaligned matched word pairs rejected, and the number of historical manual review tasks for unaligned matched word pairs passed. Accordingly, the feature extraction strategy associated with the word pair pattern can indicate that the extracted first feature information needs to indicate the number of word pairs, the number of fully matched word pairs, the number of aligned matched word pairs, and the number of unaligned matched word pairs. Optionally, the second feature information can also indicate the business scenario to which the associated transaction belongs, the language involved in the transaction, and / or the hit category of the scanning parameter associated with the second feature information, etc.; accordingly, the feature extraction strategy can also indicate that the extracted first feature information needs to indicate the business scenario to which the transaction belongs, the language involved in the transaction, and / or the hit category of the scanning parameter, etc.

[0093] In one embodiment, the feature-related data may also include several scanning categories, which may correspond to at least one of the several modes. Based on this, in step 302, the target scanning category that the first scanning parameter matches may be determined in the several scanning categories, and then the target mode that the first scanning parameter and its corresponding hit parameter match may be determined in the mode corresponding to the target scanning category.

[0094] It should be noted that the second review robot may, for example, store first correspondence information, which is used to characterize the correspondence between parameter categories and scan categories. In step 302, the parameter category to which the first scan parameter belongs may be first determined, and the parameter category may be referred to as the hit category of the first scan parameter. Furthermore, the scan category corresponding to the parameter category may be determined based on the first correspondence information, and the scan category may be determined as the target scan category that the first scan parameter complies with. When determining the target pattern that the first scan parameter and its corresponding hit parameter comply with, the target pattern may be determined based on the characteristics of the first scan parameter and its corresponding hit parameter.

[0095] Continue to view Figure 2 When the second review robot is unable to make a decision on the current review task, the scanning engine can also adopt manual review.

[0096] Specifically, after the second review robot returns a decision-making failure message to the scanning engine by executing step 216, the scanning engine can respond to the decision-making failure message returned by the second review robot by executing step 228 to send the current review task to the review terminal used by the relevant review personnel. Then, the review terminal can provide the current review task to the review personnel, who will conduct the review. After completing the review, the review personnel can submit the review result of the current review task to the review terminal, and then the review terminal can return the review result of the current review task to the scanning engine by executing step 230.

[0097] According to the description of step 228 and step 230 in the foregoing text, it can be known that the scanning engine adopts the manual review method only when the above-mentioned multiple review robots are unable to make a decision for the current review task, which can effectively save the labor cost of the review.

[0098] In one embodiment, manually reviewed review tasks and their corresponding review results can be used to update the data of the review robot, which can effectively improve the comprehensiveness and review accuracy of the review robot's review reference data.

[0099] Specifically, after step 230, the scanning engine can respond to receiving the review result of the current review task from the reviewer, treat the current review task as a manual review task, and update the manual review task and its corresponding review result to the local database of the first intelligent robot.

[0100] In addition, the scanning engine can also update the data of the second review robot according to the manual review task and its corresponding review result, for example, update the feature-related data in the second review robot.

[0101] For example, when the feature-related data contains second feature information related to the manual review task, the score corresponding to the second feature information can be updated according to the review result corresponding to the manual review task. Furthermore, when the second feature information includes several cumulative features, the numerical values ​​of the several cumulative features can be updated according to the manual review task and the review result.

[0102] For another example, when the feature-related data does not contain the second feature information related to the manual review task, the preset template feature extraction strategy can be used to extract the features of the manual review task, thereby obtaining the second feature information used as the template feature, and the score corresponding to the second feature information can be determined according to the review result corresponding to the manual review task. Afterwards, the second feature information and its corresponding score can be updated to the feature-related data.

[0103] In one embodiment, the scanning engine can also implement quality control of the second feature information in the second review robot to avoid the risk of systematic decision-making errors, ensure the accuracy of intelligent review, and meet compliance requirements.

[0104] Specifically, in one example, the scanning engine can obtain (for example, periodically obtain) a number of review tasks reviewed by the second review robot, and the review results corresponding to the several review tasks, and send the several review tasks to the review terminal used by the relevant review personnel. Afterwards, the scanning engine can receive the review results of the several review tasks by the review personnel from the review terminal. Then, the scanning engine can determine whether the review results of the second review robot and the review personnel for any review task among the several review tasks are consistent, and when the result is determined to be no, the second feature information referenced for the arbitrary review task in the above-mentioned feature-related data is invalidated. For example, the score corresponding to the second feature information can be reset to zero. Furthermore, when the second feature information includes several cumulative features, the numerical values ​​of the several cumulative features can also be reset to zero.

[0105] In another example, the scanning engine can obtain (e.g., periodically obtain) at least one second feature information that meets the preset quality inspection conditions from the above feature-related data, as well as the scores corresponding to the at least one second feature information. Afterwards, the scanning engine can send the at least one second feature information and the scores corresponding to the at least one second feature information to the quality inspection terminal used by the relevant quality inspectors, so that the quality inspectors can perform quality inspections. The quality inspectors and the reviewers mentioned above can be the same or different, and are not specifically limited here.

[0106] The preset quality inspection condition may include one or more sub-conditions. For example, the preset quality inspection condition may include that the corresponding score has been reset to zero and / or the number of times it has been cited has reached a threshold number of times. It should be noted that when the preset quality inspection condition includes multiple sub-conditions, any second feature information in the at least one second feature information may satisfy one or more of the multiple sub-conditions, which is not specifically limited here.

[0107] Further references Figure 4 This specification provides an embodiment of a sanction scanning device for transactions, which can be applied to Figure 1 The first trial robot is shown in FIG. The first trial robot is included in a plurality of trial robots associated with the scanning engine.

[0108] like Figure 4 As shown, the sanction scanning device 400 for transactions in this embodiment includes: a receiving unit 401, a searching unit 402 and an output unit 403. The receiving unit 401 is configured to receive the current review task sent by the scanning engine, and the current review task is generated by the scanning engine according to the sanction scanning request for the current transaction; the searching unit 402 is configured to search for the target manual review task that is the same as the current review task in the local database, wherein the local database stores a number of manual review tasks and the review results corresponding to the several manual review tasks; the output unit 403 is configured to return the review result corresponding to the target manual review task to the scanning engine if it is found; if it is not found, return a decision-making failure message to the scanning engine.

[0109] In some embodiments, the receiving unit 401 can also be configured to: receive the manual review task sent by the scanning engine and the review result corresponding to the manual review task; the above-mentioned device 400 can also include: a storage unit (not shown in the figure), which is configured to store the manual review task and the review result in a local database.

[0110] Further references Figure 5 This specification provides an embodiment of a sanction scanning device for transactions, which can be applied to Figure 1The second trial robot shown in the figure. The second trial robot is included in a plurality of trial robots associated with the scanning engine, and the plurality of trial robots also include the first trial robot.

[0111] like Figure 5 As shown, the sanction scanning device 500 for transactions in this embodiment includes: a receiving unit 501, a feature extraction unit 502, a prediction unit 503 and an output unit 504. The receiving unit 501 is configured to receive the current trial task sent by the scanning engine. The current trial task is sent when the first trial robot does not store the same target manual trial task. The current trial task is generated by the scanning engine according to the sanction scanning request for the current transaction, and includes a number of scanning parameters and hit parameters corresponding to the scanning parameters. The scanning parameters are derived from the current transaction. The hit parameters match the scanning parameters corresponding to the scanning parameters and are included in the preset sanction list. The feature extraction unit 502 is configured to select a scanning parameter from the scanning parameters as the first scanning parameter, and perform feature extraction on the first scanning parameter and the hit parameter corresponding to the first scanning parameter to obtain the extracted first feature information. The prediction unit 503 is configured to predict the sub-trial result corresponding to the first scanning parameter according to the first feature information. The output unit 504 is configured to return a decision-making failure message for the current trial task to the scanning engine if the sub-trial result is not predicted.

[0112] In some embodiments, the above-mentioned device 500 may also include: a first generation unit (not shown in the figure), which is configured to generate a review result indicating rejection for the current review task if the prediction unit 503 predicts the above-mentioned sub-review result and the above-mentioned sub-review result indicates rejection; the output unit 504 may also be configured to return the review result to the scanning engine.

[0113] In some embodiments, the apparatus 500 may further include a determination unit (not shown in the figure) and a second generation unit (not shown in the figure). The determination unit may be configured to determine whether there are any unselected scanning parameters among the above-mentioned scanning parameters if the prediction unit 503 predicts the above-mentioned sub-trial result and the above-mentioned sub-trial result indicates passing; the feature extraction unit 502 may also be configured to select a scanning parameter from the unselected scanning parameters as the first scanning parameter if the determination result of the determination unit is yes, and continue to perform feature extraction on the first scanning parameter and its corresponding hit parameter; the second generation unit may be configured to generate a trial result indicating passing for the current trial task if the determination result of the determination unit is no; the output unit 504 may also be configured to return the trial result to the scanning engine.

[0114] In some embodiments, the second review robot may store feature-related data, which may include a plurality of second feature information and scores corresponding to the plurality of second feature information, which may be a rejection rate or a pass rate. The second feature information and its corresponding score may be generated according to a plurality of manual review tasks and the review results corresponding to the plurality of manual review tasks.

[0115] In some embodiments, the prediction unit 503 can be further configured to: search for target second feature information matching the first feature information among the above-mentioned plurality of second feature information; if the target second feature information is found, determine whether the score corresponding to the target second feature information reaches the score threshold; if it is determined that the score corresponding to the target second feature information reaches the score threshold, generate a sub-examination result corresponding to the first scanning parameter based on the score corresponding to the target second feature information.

[0116] In some embodiments, the prediction unit 503 may be further configured to: if the target second feature information is not found, or if it is determined that the score corresponding to the target second feature information does not reach the score threshold, determine that the sub-trial result cannot be predicted for the first scanning parameter.

[0117] In some embodiments, the feature-related data may also include several patterns, any of which corresponds to at least one of the several second feature information, and is also associated with a feature extraction strategy; and the feature extraction unit 502 may be further configured to: in the several patterns, determine the target pattern that the first scanning parameter and its corresponding hit parameter conform to, and use the feature extraction strategy associated with the target pattern to perform feature extraction on the first scanning parameter and its corresponding hit parameter; the prediction unit 503 may be further configured to: in the second feature information corresponding to the target pattern, search for the target second feature information that matches the first feature information.

[0118] In some embodiments, the feature-related data may also include several scanning categories, any of which corresponds to at least one of the several modes; and the feature extraction unit 502 may be further configured to: determine, among the several scanning categories, the target scanning category to which the first scanning parameter conforms, and, among the modes corresponding to the target scanning categories, determine the target mode to which the first scanning parameter and its corresponding hit parameter conform.

[0119] In some embodiments, the prediction unit 503 can be further configured to: if the score corresponding to the target second characteristic information is a rejection rate, then a sub-examination result corresponding to the first scanning parameter and indicating rejection is generated; if the score corresponding to the target second characteristic information is a pass rate, then a sub-examination result corresponding to the first scanning parameter and indicating pass is generated.

[0120] exist Figure 4 , Figure 5 In the corresponding device embodiments, the specific processing of each unit and the technical effects brought about by it can be referred to the relevant description of the method embodiment in the previous text, and will not be repeated here.

[0121] The embodiments of this specification also provide a computer-readable storage medium on which a computer program is stored. When the computer program is executed in a computer, the computer is caused to execute the sanction scanning method for transactions described in the above method embodiments.

[0122] An embodiment of the present specification also provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the sanction scanning method for transactions described in the above method embodiments is implemented.

[0123] The embodiments of this specification also provide a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the sanction scanning method for transactions described in the above method embodiments.

[0124] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the multiple embodiments disclosed in this specification can be implemented by hardware, software, firmware, or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0125] In some cases, the actions or steps described in the claims may be performed in a different order than in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0126] The specific implementation methods described above further illustrate in detail the purposes, technical solutions and beneficial effects of the multiple embodiments disclosed in this specification. It should be understood that the above description is only the specific implementation methods of the multiple embodiments disclosed in this specification, and is not used to limit the protection scope of the multiple embodiments disclosed in this specification. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the multiple embodiments disclosed in this specification should be included in the protection scope of the multiple embodiments disclosed in this specification.

Claims

1. A sanction scanning method for a transaction, involving a scanning engine, wherein the scanning engine is associated with a plurality of review robots, wherein the plurality of review robots include a first review robot and a second review robot; the method comprises: The scanning engine sends the current review task generated by it to the first review robot, wherein the current review task is generated according to the sanction scanning request for the current transaction, and the current review task includes a plurality of scanning parameters and hit parameters corresponding to the plurality of scanning parameters, wherein the plurality of scanning parameters are derived from the current transaction, and the hit parameters match the corresponding scanning parameters and are included in a preset sanction list; The first review robot searches for a target manual review task that is the same as the current review task in a local database, wherein the local database stores a plurality of manual review tasks and review results corresponding to the plurality of manual review tasks; if the target manual review task is found, the review result corresponding to the target manual review task is returned to the scanning engine; if the target manual review task is not found, a decision-making failure message is returned to the scanning engine; In response to receiving the decision-making failure message returned by the first review robot, the scanning engine sends the current review task to the second review robot; The second review robot selects a scanning parameter from the several scanning parameters as the first scanning parameter, and performs feature extraction on the first scanning parameter and its corresponding hit parameter to obtain extracted first feature information; predicts a sub-review result corresponding to the first scanning parameter based on the first feature information; if the sub-review result is not predicted, returns a decision-making failure message for the current review task to the scanning engine.

2. The method according to claim 1, wherein: After predicting the sub-trial result corresponding to the first scanning parameter according to the first feature information, the method further includes: If the sub-examination result is predicted and the sub-examination result indicates rejection, a review result indicating rejection is generated for the current review task, and the generated review result is returned to the scanning engine.

3. The method according to claim 1, wherein: After predicting the sub-trial result corresponding to the first scanning parameter according to the first feature information, the method further includes: If the sub-trial result is predicted and the sub-trial result indicates passing, determining whether there are any unselected scanning parameters among the plurality of scanning parameters; If the determination result is yes, a scanning parameter is selected from the unselected scanning parameters as the first scanning parameter, and the feature extraction of the first scanning parameter and the corresponding hit parameter is continued; If the determination result is no, a review result indicating passing is generated for the current review task, and the generated review result is returned to the scanning engine.

4. The method according to claim 1, further comprising: In response to receiving the decision-making failure message returned by the second review robot, the scanning engine sends the current review task to a review terminal used by relevant reviewers; The scanning engine receives the review result of the reviewer on the current review task from the review end.

5. The method according to claim 1, wherein: The second review robot saves feature-related data, which includes a number of second feature information and scores corresponding to the second feature information, respectively, where the scores are rejection rates or pass rates; the second feature information and its corresponding scores are generated based on a number of manual review tasks and the review results corresponding to the manual review tasks.

6. The method according to claim 5, wherein: The predicting, according to the first feature information, a sub-trial result corresponding to the first scanning parameter includes: Searching for target second feature information matching the first feature information among the plurality of second feature information; If the target second characteristic information is found, determining whether the score corresponding to the target second characteristic information reaches the score threshold; If it is determined that the score corresponding to the target second characteristic information reaches the score threshold, a sub-examination result corresponding to the first scanning parameter is generated according to the score corresponding to the target second characteristic information.

7. The method according to claim 6, wherein: The predicting, according to the first feature information, a sub-trial result corresponding to the first scanning parameter also includes: If the target second characteristic information is not found, or it is determined that the score corresponding to the target second characteristic information does not reach the score threshold, it is determined that the sub-trial result cannot be predicted for the first scanning parameter.

8. The method according to claim 6, wherein: The feature-related data further includes a plurality of patterns, wherein any pattern corresponds to at least one item of the plurality of second feature information and is also associated with a feature extraction strategy; as well as The extracting features of the first scanning parameter and the corresponding hit parameter includes: Among the several modes, determining a target mode to which the first scanning parameter and its corresponding hit parameter conform; Using the feature extraction strategy associated with the target pattern, feature extraction is performed on the first scanning parameter and its corresponding hit parameter; The step of searching, among the plurality of second feature information, target second feature information matching the first feature information includes: In the second characteristic information corresponding to the target pattern, search for target second characteristic information matching the first characteristic information.

9. The method according to claim 8, wherein: The feature-related data further includes a plurality of scanning categories, wherein any scanning category corresponds to at least one of the plurality of modes; as well as Determining, among the plurality of modes, a target mode in which the first scanning parameter and its corresponding hit parameter conform to each other includes: Determine, among the plurality of scan categories, a target scan category to which the first scan parameter complies; In the pattern corresponding to the target scanning category, a target pattern that the first scanning parameter and its corresponding hit parameter match is determined.

10. The method according to claim 6, wherein: The generating, according to the score corresponding to the second characteristic information of the target, a sub-trial result corresponding to the first scanning parameter includes: If the score corresponding to the second characteristic information of the target is a rejection rate, generating a sub-trial result corresponding to the first scanning parameter and indicating rejection; If the score corresponding to the target second characteristic information is a pass rate, a sub-examination result corresponding to the first scanning parameter and indicating pass is generated.

11. The method according to claim 4, wherein: After receiving the review result of the current review task from the reviewer from the review end, the method further includes: The scanning engine regards the current review task as a manual review task, and updates the manual review task and its corresponding review result to the local database.

12. The method according to claim 4, wherein: After receiving the review result of the current review task from the reviewer from the review end, the method further includes: The scanning engine regards the current review task as a manual review task, and updates the data of the second review robot according to the manual review task and its corresponding review result.

13. The method according to claim 5, further comprising: The scanning engine obtains a number of review tasks reviewed by the second review robot and review results corresponding to the number of review tasks, and sends the number of review tasks to a review terminal used by relevant review personnel; The scanning engine receives the review results of the review personnel on the plurality of review tasks from the review end; The scanning engine determines, for any review task among the several review tasks, whether the review results of the second review robot and the review personnel on the any review task are consistent, and when the result is determined to be no, invalidates the second feature information referenced in the feature-related data for the any review task.

14. The method according to claim 13, wherein: The invalidation processing of the second feature information referenced by the arbitrary trial task in the feature-related data includes: The score corresponding to the second characteristic information referenced for the arbitrary trial task is reset to zero.

15. The method according to claim 14, wherein: The second characteristic information referenced for the arbitrary trial task includes a plurality of cumulative characteristics; as well as The invalidation processing of the second feature information referenced by the arbitrary trial task in the feature-related data further includes: The numerical values ​​of the plurality of cumulative features are reset to zero.

16. The method according to claim 5, further comprising: The scanning engine obtains at least one piece of second feature information satisfying a preset quality inspection condition and scores corresponding to the at least one piece of second feature information from the feature-related data; The scanning engine sends the at least one second feature information and the scores respectively corresponding thereto to a quality inspection terminal used by relevant quality inspectors, so that the quality inspectors can perform quality inspection.

17. A sanction scanning method for a transaction, applied to a first review robot and a second review robot, wherein the first review robot and the second review robot are included in a plurality of review robots associated with a scanning engine, the method comprising: The first review robot receives a current review task sent by the scanning engine, wherein the current review task is generated by the scanning engine according to a sanction scanning request for the current transaction, and the current review task includes a plurality of scanning parameters and hit parameters corresponding to the plurality of scanning parameters, wherein the plurality of scanning parameters are derived from the current transaction, and the hit parameters match the corresponding scanning parameters and are included in a preset sanction list; The first review robot searches a local database for a target manual review task that is the same as the current review task, wherein the local database stores a plurality of manual review tasks and review results corresponding to the plurality of manual review tasks; If found, the first review robot returns the review result corresponding to the target manual review task to the scanning engine; If the search is not successful, the first review robot returns a decision-making failure message to the scanning engine; The second review robot receives the current review task sent by the scanning engine, where the current review task is sent by the scanning engine when the first review robot returns the unable to make a decision message; The second review robot selects a scanning parameter from the plurality of scanning parameters as a first scanning parameter, and performs feature extraction on the first scanning parameter and its corresponding hit parameter to obtain extracted first feature information; The second review robot predicts a sub-review result corresponding to the first scanning parameter according to the first feature information; If the sub-trial result is not predicted, the second trial robot returns a message of being unable to make a decision for the current trial task to the scanning engine.

18. A sanction scanning method for a transaction, applied to a second review robot, the second review robot being included in a plurality of review robots associated with a scanning engine, the plurality of review robots also including a first review robot, the method comprising: receiving a current review task sent by the scanning engine, wherein the current review task is sent when the first review robot does not store the same target manual review task, the current review task is generated by the scanning engine according to a sanction scanning request for the current transaction, and includes a plurality of scanning parameters and hit parameters corresponding to the plurality of scanning parameters, the plurality of scanning parameters are derived from the current transaction, and the hit parameters match the corresponding scanning parameters and are included in a preset sanction list; Selecting a scanning parameter from the plurality of scanning parameters as a first scanning parameter, and performing feature extraction on the first scanning parameter and its corresponding hit parameter to obtain extracted first feature information; predicting a sub-trial result corresponding to the first scanning parameter according to the first feature information; If the sub-trial result is not predicted, a decision-making failure message for the current trial task is returned to the scanning engine.

19. A scanning device for sanctions against transactions, applied to a first review robot and a second review robot, wherein the first review robot and the second review robot are included in a plurality of review robots associated with a scanning engine, the device comprising: A first receiving unit is configured as a first review robot to receive a current review task sent by the scanning engine, wherein the current review task is generated by the scanning engine according to a sanction scanning request for the current transaction, and the current review task includes a plurality of scanning parameters and hit parameters corresponding to the plurality of scanning parameters, wherein the plurality of scanning parameters are derived from the current transaction, and the hit parameters match the corresponding scanning parameters and are included in a preset sanction list; A search unit is configured to enable the first trial robot to search a local database for a target manual trial task that is the same as the current trial task, wherein the local database stores a plurality of manual trial tasks and trial results corresponding to the plurality of manual trial tasks; The first output unit is configured such that if the target manual review task is found, the first review robot returns the review result corresponding to the target manual review task to the scanning engine; if the target manual review task is not found, the first review robot returns a decision-making failure message to the scanning engine; A second receiving unit, configured to receive, as a second trial robot, the current trial task sent by the scanning engine, wherein the current trial task is sent by the scanning engine when the first trial robot returns the unable to make a decision message; The feature extraction unit is configured such that the second trial robot selects a scanning parameter from the plurality of scanning parameters as a first scanning parameter, and performs feature extraction on the first scanning parameter and its corresponding hit parameter to obtain extracted first feature information; A prediction unit, configured as a second trial robot to predict a sub-trial result corresponding to the first scanning parameter according to the first feature information; The second output unit is configured to, if the sub-trial result is not predicted, cause the second trial robot to return a message indicating that a decision cannot be made with respect to the current trial task to the scanning engine.

20. A sanction scanning device for transactions, applied to a second review robot, the second review robot being included in a plurality of review robots associated with a scanning engine, the plurality of review robots also including a first review robot, the device comprising: a receiving unit, configured to receive a current review task sent by the scanning engine, wherein the current review task is sent when the first review robot does not store the same target manual review task, wherein the current review task is generated by the scanning engine according to a sanction scanning request for the current transaction, and comprises a plurality of scanning parameters and hit parameters corresponding to the plurality of scanning parameters, wherein the plurality of scanning parameters are derived from the current transaction, and the hit parameters match the corresponding scanning parameters and are included in a preset sanction list; a feature extraction unit configured to select a scanning parameter from the plurality of scanning parameters as a first scanning parameter, and perform feature extraction on the first scanning parameter and its corresponding hit parameter to obtain extracted first feature information; A prediction unit, configured to predict a sub-trial result corresponding to the first scanning parameter according to the first feature information; The output unit is configured to return a decision-making failure message for the current trial task to the scanning engine if the sub-trial result is not predicted.

21. A computer-readable storage medium having a computer program stored thereon, wherein: When the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 17 to 18.

22. A computing device comprising a memory and a processor, wherein: The memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 17 to 18 is implemented.

23. A computer program, wherein When the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 17 to 18.

Images