Method, device, equipment, medium and program product for detecting blacklist

Abstract

The present disclosure provides a method for detecting a blacklist, which can be used in the financial field and the information security field. The specific implementation scheme is: obtaining information to be processed; using a semantic analysis model to parse the information to be processed to obtain multiple structured data for the information to be processed; and using a preset blacklist search engine to determine the blacklist data in the multiple structured data. The present disclosure also provides a device, equipment, medium and program product for detecting a blacklist.

Description

Technical Field

[0001] The present disclosure relates to the field of information processing technology, specifically to the field of finance and information security, and more specifically to a method, device, equipment, medium and program product for detecting a blacklist. Background Art

[0002] In the process of handling financial business, financial institutions need to conduct blacklist detection on the business information generated during the business process in order to prevent and control business risks.

[0003] In the related technology, a purchased professional blacklist checking system and blacklist database are usually used to perform blacklist detection on business information. The use of methods in the related technology often leads to many false hits, especially for cross-border financial business, which involves many business objects and the probability of false hits will increase significantly. The false hit business information usually needs to be manually confirmed, which will bring high labor costs when there are many links in the financial business. Summary of the invention

[0004] In view of the above problems, the present disclosure provides a method, apparatus, device, medium and program product for detecting a blacklist to reduce the probability of false hits.

[0005] According to the first aspect of the present disclosure, a method for checking a blacklist is provided, comprising: obtaining information to be processed; parsing the information to be processed using a semantic analysis model to obtain a plurality of structured data for the information to be processed; and determining the blacklist data in the plurality of structured data using a preset blacklist retrieval engine.

[0006] According to an embodiment of the present disclosure, the above-mentioned information to be processed includes multiple pieces of information; using a semantic analysis model to parse the information to be processed includes: identifying the information to be processed to determine at least one piece of information with a predetermined identifier in the information to be processed as the information to be parsed; and using a semantic analysis model to parse each piece of information in the information to be parsed to obtain structured data representing each piece of information.

[0007] According to an embodiment of the present disclosure, the above-mentioned method for detecting a blacklist further includes, after determining the blacklist data in the plurality of structured data: determining the misidentified data in the blacklist data according to a preset rule.

[0008] According to an embodiment of the present disclosure, the above-mentioned determining the erroneous data in the blacklist data according to the preset rule includes: determining the data in the blacklist data that belongs to the preset whitelist as the erroneous data.

[0009] According to an embodiment of the present disclosure, the information to be processed includes information generated by processing a business; the above-mentioned determination of the erroneous data in the blacklist data according to preset rules includes: determining that the data in the blacklist data that meets the preset erroneous rules is erroneous data, and the preset erroneous rules include at least one of the following: the blacklist data is the same as the first data that has been confirmed to be erroneous, and the first data and the blacklist data are data generated by processing the same business; the blacklist data indicates a business object, and the type of business object indicated by the blacklist data is inconsistent with the type of business object targeted by the business; the blacklist data is part of the preset information; the blacklist data includes unstructured data; or the blacklist data is data of a preset type.

[0010] According to an embodiment of the present disclosure, the above-mentioned determination of the erroneous data in the blacklist data according to preset rules includes: using a preset machine learning model to determine the probability that each blacklist data in the blacklist data is erroneous data; and determining whether each blacklist data is erroneous information based on the probability that each blacklist data is erroneous data and a preset threshold.

[0011] According to an embodiment of the present disclosure, determining whether each blacklist data is false positive information includes: determining that each blacklist data is false positive data when the probability that each blacklist data is false positive data is greater than or equal to a first preset threshold.

[0012] According to an embodiment of the present disclosure, determining whether each blacklist data is false hit information also includes: when the probability of each blacklist data being false hit data is less than a first preset threshold and greater than or equal to a second preset threshold, determining that each blacklist data is suspected false hit data; when the probability of each blacklist data being false hit data is less than the second preset threshold, determining that each blacklist data is hit data.

[0013] According to an embodiment of the present disclosure, the above-mentioned preset machine learning model includes at least two models of different types, and determining the probability that each blacklist data in the blacklist data is erroneous data includes: determining the input data of each model in the at least two models according to each blacklist data; inputting the input data into each model, and outputting output information for each blacklist data to obtain at least two output information for each blacklist data; and determining the probability that each blacklist data is erroneous data according to the at least two output information for each blacklist data.

[0014] The second aspect of the present disclosure provides a device for detecting a blacklist, including: an information acquisition module, used to acquire information to be processed; an information parsing module, used to parse the information to be processed using a semantic analysis model to obtain multiple structured data for the information to be processed; and a blacklist determination module, used to use a preset blacklist retrieval engine to determine the blacklist data in the multiple structured data.

[0015] The third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above-mentioned method for detecting a blacklist.

[0016] A fourth aspect of the present disclosure further provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the above-mentioned method for detecting a blacklist.

[0017] A fifth aspect of the present disclosure further provides a computer program product, including a computer program, which implements the above-mentioned method for detecting a blacklist when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0019] Figure 1 The following schematically shows an application scenario diagram of a method, apparatus, device, medium, and program product for detecting a blacklist according to an embodiment of the present disclosure;

[0020] Figure 2 A flowchart of a method for detecting a blacklist according to an embodiment of the present disclosure is schematically shown;

[0021] Figure 3 Schematically shows a flow chart of parsing information to be processed using a semantic analysis model according to an embodiment of the present disclosure;

[0022] Figure 4 A flowchart of a method for detecting a blacklist according to another embodiment of the present disclosure is schematically shown;

[0023] Figure 5 A schematic diagram schematically shows the principle of determining the misselected data in the blacklist data according to the preset rules according to an embodiment of the present disclosure;

[0024] Figure 6 Schematically shows a flow chart of determining misselected data in blacklist data according to preset rules according to an embodiment of the present disclosure;

[0025] Figure 7 A block diagram schematically shows a structure of a blacklist screening device for executing a method for detecting a blacklist according to an embodiment of the present disclosure;

[0026] Figure 8 A structural block diagram of a device for detecting a blacklist according to an embodiment of the present disclosure is schematically shown; and

[0027] Fig. 9 A block diagram of an electronic device suitable for implementing a method for detecting a blacklist according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0028] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0029] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0030] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0031] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0032] The embodiment of the present disclosure provides a method for detecting a blacklist, which first obtains information to be processed, then uses a semantic analysis model to parse the information to be processed, and obtains a plurality of structured data for the information to be processed, and finally uses a preset blacklist search engine to determine the blacklist data in the plurality of structured data.

[0033] Figure 1 The application scenario diagram of the method, apparatus, device, medium and program product for detecting a blacklist according to an embodiment of the present disclosure is schematically shown.

[0034] like Figure 1As shown, the application scenario 100 may include a first terminal 110 and a second terminal 120. The first terminal 110 and the second terminal 120 may be connected via a network communication, for example. The network may include a wired or wireless communication link.

[0035] In one embodiment, the first terminal 110 may be, for example, various electronic devices used in a financial institution to process financial services. The first terminal 110 may, for example, generate a business message during the business processing. The business message may include, for example, a remittance message, etc. In a cross-border transaction business, the business message may be, for example, a SWIFT (Society for Worldwide Interbank Financial Telecommunications) message. For example, the first terminal 110 may be a self-service business processing terminal (such as an ATM machine), a terminal device set up at a counter, etc.

[0036] In one embodiment, the second terminal 120 may be, for example, a terminal device that checks the business message generated by the first terminal 110 to ensure the legitimacy and accuracy of the business message before completing the financial business according to the business message. For example, the second terminal 120 may perform a blacklist check on the business message. The second terminal may be, for example, various electronic devices with processing functions, including but not limited to servers, desktop computers, laptop portable computers, tablet computers, or smart phones, etc.

[0037] In one embodiment, the second terminal 120 in the application scenario 100 can be connected to an electronic device 130 provided with a blacklist detection engine via a network, for example, and the second terminal 120 can be used to identify, parse, and process the service message, and obtain structured data as input of the blacklist detection engine. The second terminal 120 can call the blacklist detection engine provided in the electronic device 130 via the network, so as to implement blacklist detection of the service message by running the blacklist detection engine. For example, the electronic device 130 can provide an interface for other electronic devices to call the blacklist detection engine, and the second terminal 120 can call the blacklist detection engine through the interface.

[0038] In one embodiment, the second terminal 120 may be provided with a display screen, for example, to provide a human-computer interaction interface. The application scenario 100 may also include a business person 140, for example, to input information for the second terminal 120 to perform a blacklist check via the human-computer interaction interface. The information input by the business person 140 may be, for example, information directly used for checking, or may also be an uploaded text file, etc. In the case where the business person 140 uploads a text file, the second terminal 120 may also use text recognition technology (such as OCR technology) to recognize the text in the text file, and obtain the information required for blacklist detection.

[0039] In one embodiment, the second terminal 120 may have an image acquisition function, so that the image of the paper document provided by the business personnel 140 can be acquired, and the information required for blacklist detection can be obtained by identifying the acquired image.

[0040] In one embodiment, if Figure 1 As shown, the application scenario 100 may also include a server 150, which may be a server of a financial institution data center. The server 150 may, for example, be a server that provides support for the operation of the first terminal 110, and the server 150 stores a full amount of business messages. Exemplarily, the second terminal 120 may be connected to the server 150 via a network, and the second terminal 120 may obtain business messages from the server 150 to identify, parse, and process the obtained business messages.

[0041] It should be noted that the method for detecting a blacklist provided in the present disclosure may be executed by, for example, the second terminal 120. Accordingly, the apparatus for detecting a blacklist provided in the present disclosure may be arranged in the second terminal 120.

[0042] Understandably, Figure 1 The number and type of the first terminal 110, the second terminal 120, the electronic device 130 and the server 150 described are only examples to facilitate understanding of the present disclosure. According to actual needs, there may be any number and type of the first terminal 110, the second terminal 120, the electronic device 130 and the server 150.

[0043] The following will be based on Figure 1 The scene described by Figure 2 to Figure 6 The method for detecting the blacklist in the disclosed embodiment is described in detail.

[0044] Figure 2 The flowchart of the method for detecting a blacklist according to an embodiment of the present disclosure is schematically shown.

[0045] like Figure 2As shown, the method 200 for detecting a blacklist of this embodiment may include operations S210 to S230.

[0046] In operation S210, information to be processed is obtained.

[0047] According to the embodiments of the present disclosure, the information to be processed may be, for example, the business message described above and other information that needs to be blacklisted. For example, the information to be processed may be the relevant information generated by financial institutions in handling business in accordance with the three principles of business development and the national anti-money laundering compliance management requirements.

[0048] Exemplarily, the information to be processed may be message information including information related to any type of object, such as the applicant, beneficiary, issuing bank, notifying bank, accepting bank, presenting bank, negotiating bank, etc. Alternatively, the information to be processed may also be the transportation information of goods, the circuit information of funds, SWIFT message, etc. For example, the information to be processed may be a message generated at any link in the business transaction link.

[0049] According to the embodiments of the present disclosure, for example, the element information that needs to be checked can be manually entered by a business personnel, and the information to be processed obtained in operation S210 can be the entered element information. Alternatively, the business personnel can upload a business attachment, and the second terminal described above can identify the business attachment through the OCR automatic recognition technology to obtain the information to be processed. The information to be processed can be, for example, structured data or unstructured data. Alternatively, the business processing information stored in the business system of the financial institution can be obtained by calling the API interface provided by the business system of the financial institution, and this can be used as the information to be processed. Among them, the information to be processed can be, for example, structured data, unstructured data or semi-structured data.

[0050] In operation S220, the information to be processed is parsed using a semantic analysis model to obtain a plurality of structured data for the information to be processed.

[0051] According to an embodiment of the present disclosure, the semantic analysis model may be, for example, a natural language processing (NPL) model. The model may be used, for example, to perform semantic analysis on information to be processed, extract dependencies between multiple words in the information to be processed, and obtain structured data based on the extracted dependencies. For example, for a remittance message, there is a dependency relationship between the payee and the payee's name, and the embodiment forms a structured data with the payee and the payee's name in the form of a key-value pair.

[0052] Exemplarily, the natural language processing model may predefine a structured data architecture and the type of data in the architecture. The embodiment may parse and identify the information to be processed, obtain data of the type of data in the architecture, and fill the data into the predefined structured data architecture to obtain structured data. The data type may be set according to actual needs, for example, the data type may include date, document, description information, etc.

[0053] In operation S230, a preset blacklist search engine is used to determine blacklist data in the plurality of structured data.

[0054] According to an embodiment of the present disclosure, for example, a blacklist search engine call interface may be called to determine blacklist data. Specifically, the plurality of structured data obtained in operation S220 is used as input to the call interface, and the data fed back by the call interface is used as blacklist data.

[0055] Among them, the blacklist search engine can maintain a blacklist library, for example. After receiving the structured data, the blacklist search engine can compare the structured data with each blacklist in the blacklist library. If a certain structured data belongs to the blacklist library, the structured data is determined to be blacklist data. It is understandable that the above method for the blacklist search engine to determine the blacklist data is only used as an example to facilitate understanding of the present disclosure, and the present disclosure does not limit this.

[0056] According to the embodiments of the present disclosure, when the information to be processed is a relatively long text message, if a blacklist search is directly performed on the information to be processed, it is easy to make a false hit. For example, for the information to be processed "we further undertake that, on the maturity data...", when the blacklist database includes the blacklist data "that, on", if a blacklist search is directly performed on the information to be processed, it is easy to output the data "that, on" in the information to be processed as blacklist data. However, the embodiments of the present disclosure can avoid listing "take, on" as a blacklist data by parsing the information to be processed and performing blacklist detection on the structured data of the information to be processed, and thus can improve the accuracy of the determined blacklist data to a certain extent. Therefore, the method for detecting blacklists provided by the present disclosure can achieve refined inspection of the information to be processed by parsing the information to be processed to obtain structured data, and thus can effectively reduce the false hit rate of the blacklist.

[0057] Figure 3 The flowchart of using a semantic analysis model to parse information to be processed according to an embodiment of the present disclosure is schematically shown.

[0058] According to the embodiments of the present disclosure, when the semantic analysis model is used to parse the information to be processed, for example, only the information with a high proportion of false hits in the information to be processed can be parsed, and structured data for the information with a high proportion of false hits can be obtained to perform blacklist retrieval on the structured data. In this way, the pertinence of the blacklist retrieval can be improved to a certain extent, and the efficiency of the blacklist retrieval can be improved.

[0059] like Figure 3 As shown, in this embodiment, the operation of parsing the information to be processed using the semantic analysis model may include operations S321 to S322.

[0060] In operation S321, the information to be processed is identified to determine at least one piece of information with a predetermined identifier in the information to be processed as the information to be parsed.

[0061] According to an embodiment of the present disclosure, the information to be processed obtained by the embodiment may include, for example, multiple pieces of data, and the multiple pieces of data may be data located at different positions in the information to be processed, and the number of the multiple pieces of data and the position of each piece of data may be determined according to the message format. After obtaining the business message, the embodiment of the present disclosure may, for example, add a predetermined identifier to the data at the position of the pre-set key information in the business message. It is understandable that the key information may be set according to actual needs, or the key information may be determined according to the historically determined erroneous data, so as to set the data with a high probability of being erroneous in history as the key information.

[0062] Exemplarily, for the issuance message of a documentary credit, the multiple data included may include, for example, "total number of pages", "form of documentary credit", "documentary credit number", "advance notice reference number", "issuance date", "expiration date and place", "applicant bank", "applicant", "beneficiary bank", "currency code", "issue conditions", "payee", ..., "goods and services description", "required documents", "additional conditions", "fees", "presentation period", ..., "notification via bank", "remarks", etc. If "data goods and services description", "required documents" and "additional conditions" are pre-set key information, a predetermined mark is added at the position of "data goods and services description", "required documents" and "additional conditions" in the issuance message. For example, if "data goods and services description" is located at the 10th byte in the format of the issuance message, a predetermined mark is added to the data of the 10th byte of the issuance message. The predetermined mark can be set according to actual needs, and the present disclosure does not limit this.

[0063] According to an embodiment of the present disclosure, the acquired information to be processed may also be added with a predetermined identifier, which is added by a device generating the information to be processed using a similar method as shown above, and will not be described in detail here.

[0064] In this embodiment, a predetermined identifier may be firstly identified on the information to be processed, and a character string with the predetermined identifier may be intercepted from the information to be processed as the information to be parsed.

[0065] In operation S322, a semantic analysis model is used to parse each piece of information in the information to be parsed, and structured data representing each piece of information is obtained.

[0066] After obtaining the data to be parsed, the NPL model described above can be used to parse each piece of data in the data to be parsed, so as to obtain structured data of each piece of information.

[0067] Based on the above analysis, the disclosed embodiment intercepts the information to be processed before parsing to obtain structured data, and only parses and extracts structured data from key information with identifiers, thereby improving the efficiency of blacklist detection while ensuring refined processing of information with a high false positive rate.

[0068] Figure 4 The following schematically shows a flow chart of a method for detecting a blacklist according to another embodiment of the present disclosure.

[0069] like Figure 4 As shown, the method for detecting the blacklist of this embodiment may further include operation S440 in addition to the operations S210 to S230 described above, to determine the erroneous data in the blacklist data according to a preset rule. Through this operation, this embodiment can select the erroneous data from the blacklist data, thereby only providing the other blacklist data except the erroneous data to the business personnel for manual review. Compared with the technical solution in the related art that requires business personnel to review all blacklist data, it can effectively reduce the cost of manual review and improve the review efficiency.

[0070] According to an embodiment of the present disclosure, a whitelist database may be maintained based on historical false positive data, and the whitelist database may have multiple preset whitelists. After determining the blacklist data, the embodiment may match the blacklist data with the preset whitelist one by one, and if the determined blacklist data contains data belonging to the preset whitelist, the data is determined to be false positive data. It is understandable that the whitelist in the whitelist database may be flexibly set, and the present disclosure does not limit this.

[0071] According to an embodiment of the present disclosure, the information to be processed may be, for example, information generated by processing a service. In this embodiment, data in the blacklist data that meets a preset false positive rule may be determined as false positive data.

[0072] Exemplarily, the preset mis-hit rule may be that when the blacklist data is repeated mis-hit data, the blacklist data is determined to be mis-hit data. Wherein, repeated mis-hit may, for example, refer to repeated mis-hit under the same business, or may refer to repeated mis-hit of information in the business message generated by the entire business system. In one embodiment, if the blacklist data is the same as the first data that has been confirmed to be mis-hit, the blacklist data is determined to be mis-hit data. Wherein, the first data and the blacklist data are data generated for processing the same business. For example, two messages are generated successively for processing a certain business, and both messages include certain data. When a blacklist detection is performed on a message generated first, if it is determined that the certain data in the message is mis-hit data, then when a blacklist detection is performed on another message generated later to determine that the certain data is blacklist data, it can be determined that the certain data in the other message is also mis-hit data.

[0073] Exemplarily, the preset mismatch rule may be that the type of the business object indicated by the blacklist data is inconsistent with the type of the business object in the actual transaction. The actual transaction is the processed business. The business object indicated by the blacklist data may include, for example, a country name, a commodity name, a company name, or an individual name. For example, when the actual transaction is a transaction between companies, if the business object indicated by the blacklist data is an individual, then the data determining that the type of the indicated business object is inconsistent with the type of the business object in the actual transaction is mismatch data.

[0074] Exemplarily, the preset mismatch rule can be a mismatch rule with the same name. If the business object indicated by the blacklist data is the same as the target business object actually belonging to the blacklist, but the business object indicated by the blacklist data belongs to different categories from the target business object, the blacklist data is determined to be mismatch data. For example, different cities or ports may be set with the same name in different countries. For example, both country A and country B have cities named a, and city a of country A is blacklisted. When city a of country B is determined to be blacklist data through the previous operation, the blacklist data can be determined to be mismatch data. For the preset mismatch rule, after obtaining the blacklist data via the blacklist search engine, if the blacklist data indicates a business object, the category to which the business object belongs can also be obtained by parsing the business message. It is then determined whether the category to which the belonging belongs is consistent with the category to which the target business object belongs. If consistent, the blacklist data is determined to be mismatch data.

[0075] Exemplarily, the preset false hit rule may be that the blacklist data is part of the preset information. The preset information may be any information such as a company name, a country name, etc. For example, if the preset information is "YOKOHAMA", then when the blacklist data is "HAMA", the blacklist data determined as part of the preset information is false hit information.

[0076] Exemplarily, the preset false positive rule may include unstructured data in the blacklist data. For example, for the message content "We further undertake that, on the maturity data...", if the determined blacklist data includes "take, on" (in the blacklist, "take, on" indicates prohibited materials), the unstructured data is determined to be false positive data. By setting such a preset false positive rule, when the accuracy of the method used to extract structured data in the previous text is low, the low accuracy can be compensated.

[0077] Exemplarily, the preset false positive rule may include: the blacklist data is data of a preset type. For example, if the blacklist type is data indicating a name, a number, or a ship name, then the data of the preset type in the blacklist data may be determined to be false positive data.

[0078] According to the embodiments of the present disclosure, the present disclosure can be pre-trained with a machine learning model. During the training process of the machine learning model, the existing blacklist data and the result information of confirming whether it is erroneous data can be extracted first, and the blacklist data and the corresponding result information can be used as a sample data to obtain a large amount of sample data. Subsequently, each sample data is processed and corrected, and the processing and correction process can include data enhancement processing, etc. For example, the sample data can be cleaned, and the result information can be used as a label for the blacklist data. Then, the existing models of natural language processing and machine learning in the industry are selected as the initial model, and the sample data is used to train the initial model. Specifically, the sample data is used as the input of the initial model, and according to the difference between the input of the initial model and the label in the sample data, the gradient descent algorithm is used to optimize the parameters in the initial model to obtain the optimized model as the preset machine learning model.

[0079] Accordingly, in this embodiment, when determining the false positive data in the blacklist data according to the preset rules, the preset machine learning model can be used to first determine the probability that each blacklist data in the blacklist data belongs to the false positive data. Then, based on the probability that each blacklist data belongs to the false positive data and the preset threshold, it is determined whether each blacklist data belongs to the false positive information. Among them, the preset threshold can be set according to actual needs, for example, it can be 0.9, 0.8, etc., which is greater than 0.5, and the present disclosure does not limit this.

[0080] Exemplarily, the preset machine learning model may be, for example, a FastText model, a TextCNN (an algorithm for classifying text using a convolutional neural network) model, a ULMFit model (a universal language model based on fine-tuning), or a bidirectional encoder (Bidirectional Encoder Representations from Transformers, BERT). Among them, the FastText model is a fast text classifier developed by Facebook. Among them, when using a preset machine learning model to determine the probability, for example, the structured data can be first converted into input information of the preset machine learning model, such as a word sequence, etc. The input information is then input into the preset machine learning model, and the preset machine learning model outputs the probability that the blacklist data belongs to the erroneous data. Among them, the output of the preset machine learning model may be, for example, the probability that the blacklist data belongs to the erroneous data, or it may be the probability that the blacklist data belongs to each category in the preset categories. Among them, the preset categories include erroneous data categories.

[0081] Exemplarily, each blacklist data may be determined to be erroneous data when the probability that each blacklist data is erroneous data is greater than or equal to a first preset threshold, wherein the first preset threshold may be the preset threshold described above.

[0082] By adopting a machine learning model to determine the probability that blacklist data is false data and setting a threshold for the probability, it is possible to comprehensively consider the blacklist data characters and the semantic information of the blacklist data, and thus improve the accuracy of the determined false data.

[0083] Based on the above analysis, the embodiments of the present disclosure can effectively reduce the amount of blacklist data that needs to be reviewed by business personnel by screening out the erroneous data from the blacklist data, thereby reducing the manual review cost and improving the blacklist detection efficiency.

[0084] According to the embodiments of the present disclosure, for example, a second preset threshold value less than the first preset threshold value may be set for the probability of being false hit data, and the hit data that actually belongs to the blacklist data may be screened out from the blacklist data according to the second preset threshold value. Finally, the other data in the blacklist data except the hit data and the false hit data are regarded as suspicious false hit data and manually reviewed by business personnel. In this way, the amount of blacklist data that needs to be reviewed by business personnel is further reduced, the cost of manual review is further reduced, and the efficiency of blacklist detection is improved.

[0085] Accordingly, the operation of determining whether each blacklist data is false hit information includes, in addition to determining that each blacklist data is false hit data when the probability of each blacklist data being false hit data is greater than or equal to a first preset threshold, also including, when the probability of each blacklist data being false hit data is less than the first preset threshold and greater than or equal to a second preset threshold, determining that each blacklist data is suspected false hit data, and when the probability of each blacklist data being false hit data is less than the second preset threshold, determining that each blacklist data is hit data.

[0086] Exemplarily, each blacklist data in the blacklist data determined by the preset blacklist retrieval engine in the foregoing text can be used as the input of the preset machine learning model, and the probability vector for each blacklist data is output. The probability vector represents the probability that each blacklist data belongs to each category in the preset category. Among them, the preset category can include, for example, a false data category, a hit data category, and a suspicious false data category. The probability vector can be, for example, a normalized vector, that is, the sum of each value in the probability vector is 1. In this way, after obtaining the probability vector for each blacklist data, the probability indicating that each blacklist data belongs to the false data can be determined first. Then the probability is compared with the first preset threshold and the second preset threshold to determine whether each blacklist data belongs to one of the false data, the hit data, and the suspicious false data. It can be understood that the second preset threshold can be set according to actual needs. For example, the value of the second preset threshold can be any value less than or equal to 0.7 and greater than or equal to 0.3. The present disclosure does not limit this, as long as the second preset threshold is less than the first preset threshold.

[0087] Figure 5 The schematic diagram schematically shows the principle of determining the erroneous data in the blacklist data according to the preset rules according to the embodiment of the present disclosure.

[0088] According to an embodiment of the present disclosure, the preset machine learning model may include, for example, at least two models of different types. This embodiment may determine whether each blacklist data is erroneous data based on the output of the at least two models. By selecting different types of models, different models may focus on different aspects of feature extraction of blacklist data, so that the final determined result can comprehensively consider the characteristics of multiple aspects of the blacklist data, and thus improve the accuracy of the determined erroneous data.

[0089] For example, Figure 5In the embodiment 500 shown, for each blacklist data 510 in the blacklist data determined by the preset blacklist retrieval engine described above, the input data of each model in at least two models can be first determined based on each blacklist data 510. For example, when the at least two models include a first machine learning model 531, a second machine learning model 532, and a third machine learning model 533, each blacklist data 510 can be first converted into the first input data 521 of the first machine learning model 531, the second input data 522 of the second machine learning model 532, and the third input data 523 of the third machine learning model 533. After the input data is obtained, the input data is input into each model, and the output information for each blacklist data 510 is output by each model, so as to obtain at least two output information. For example, the first input data 521, the second input data 522, and the third input data 523 are respectively input into the first machine learning model 531, the second machine learning model 532, and the third machine learning model 533, to obtain the first output information 541 output by the first machine learning model 531, the second output information 542 output by the second machine learning model 532, and the third output information 543 output by the third machine learning model 533. Finally, according to at least two output information for each blacklist data, the probability of each blacklist data 510 being a false positive data is determined. For example, according to the first output information 541, the second output information 542, and the third output information 543, it is determined whether the blacklist data 510 is a false positive data.

[0090] For example, the probability values ​​indicating that the blacklist data 510 is erroneous data in at least two output information may be determined to obtain at least two probability values. Finally, the average of the at least two probability values ​​is used as the probability 550 that the blacklist data 510 is erroneous data.

[0091] Exemplarily, the output of at least two models can also be used as the input of the fusion layer, and the fusion layer outputs the probability that each blacklist data 510 belongs to the false data. The fusion layer and the at least two models are trained simultaneously. The fusion layer is used to fuse the feature data extracted from the at least two models and classify the blacklist data according to the feature data. The probability indicating that the blacklist data 510 belongs to the false data output by the fusion layer is finally determined as the final determined probability.

[0092] Exemplarily, the method of converting blacklist data into input data for each of at least two models is similar to the method in the related art. For example, the input data of the model can be obtained by converting the blacklist data into a word vector or a word sequence, which will not be described in detail here.

[0093] Figure 6A flowchart of determining misselected data in blacklist data according to preset rules according to an embodiment of the present disclosure is schematically shown.

[0094] According to the embodiments of the present disclosure, Figure 6 As shown, the operation of determining the misselected data in the blacklist data according to the preset rule in the embodiment 600 may include, for example, the following process executed for each blacklist data.

[0095] In operation S641, it is determined whether the blacklist data belongs to a preset whitelist. If so, operation S642 is performed, otherwise operation S643 is performed. The preset whitelist is similar to the preset whitelist described above, and will not be described in detail here.

[0096] In operation S642, it is determined that the blacklist data belongs to mis-selected data.

[0097] In operation S643, it is determined whether the blacklist data satisfies the preset false hit rule. If so, operation S642 is executed, otherwise operation S644 is executed. The preset false hit rule is similar to the preset false hit rule described above, and will not be described in detail here.

[0098] In operation S644, a preset machine learning model is used to determine the probability that the blacklist data is erroneous data. The preset machine learning model is similar to that described above and will not be described in detail here.

[0099] In operation S645, it is determined whether the probability of the data being misidentified is less than a first preset threshold. If so, operation S646 is performed, otherwise operation S642 is performed.

[0100] In operation S646, it is determined whether the probability of the data being misidentified is less than a second preset threshold. If so, operation S647 is performed, otherwise, operation S648 is performed.

[0101] In operation S647, it is determined that the blacklist data belongs to hit data.

[0102] In operation S648, it is determined that the blacklist data belongs to the suspected mis-hit data.

[0103] This embodiment uses a combination of a whitelist, preset false hit rules and a machine learning model to determine whether blacklist data is false hit data, which can effectively avoid the omission of false hit data and thus improve the accuracy of the determined false hit data. At the same time, by screening out hit data from blacklist data, only suspicious false hit data can be provided to business personnel for manual review, which can effectively reduce the amount of data for manual review and improve the cost and efficiency of blacklist detection.

[0104] Based on the above blacklist detection method, the present disclosure also provides a blacklist screening device suitable for executing the blacklist detection method. Figure 7 The blacklist screening device is described in detail.

[0105] Figure 7 The structure block diagram of a blacklist screening device suitable for executing the method for detecting a blacklist according to an embodiment of the present disclosure is schematically shown.

[0106] like Figure 7 As shown, the blacklist screening device 700 may include a test information collection module 710, a test information analysis module 720, a data transmission module 730, a test result review module 740, a review result output module 750, a review result confirmation module 760 and a data storage module 770.

[0107] like Figure 7 As shown, the inspection information collection module 710 may include, for example, an interface entry collection unit 711, an image upload collection unit 712, and an API interface collection unit 713. The interface entry collection unit 711 is used to provide an interactive interface to business personnel, and the business personnel enter the key information that needs to be inspected. In the present device 700, the interface entry can be configured with a custom input field through the entry style. The image upload collection unit 712 is used to use OCR automatic recognition technology to perform character recognition on the business attachments uploaded by the business personnel, and complete the extraction and collection of the information to be processed. The API interface collection unit 713 is used to connect to the business system of the financial institution to directly obtain the business handling information generated by the business system of the financial institution as the information to be processed.

[0108] The inspection information analysis module 720 is mainly used to further analyze the data collected by the inspection information collection module 710. Figure 7 As shown, the information analysis module 720 may include a semantic analysis unit 721 and a first model synchronization unit 722. The semantic analysis unit 721 is used to obtain structured data after semantically parsing and structuring the content of the text input field in the information to be inspected (i.e., the information to be processed) using a semantic analysis model. The structured data is sent to a preset blacklist retrieval engine via a data transmission module 730 for blacklist detection. The first model synchronization unit 722 is used to synchronize and update the semantic analysis model used by the semantic analysis unit 721 in real time.

[0109] The data transmission module 730 is responsible for data transmission between the modules in the blacklist screening device 700. Figure 7As shown, the data transmission module 730 includes an information inspection unit 731, a detection result receiving unit 732 and a data updating unit 733. Among them, the information inspection unit 731 is used to send structured data to the preset blacklist search engine. The detection result receiving unit 732 is used to receive the blacklist data fed back by the preset blacklist search engine and store it locally in the blacklist screening device 700 (for example, stored in the data storage module 770). The data updating unit 733 is used to update the audit status of the data.

[0110] The detection result review module 740 is used to review the blacklist data fed back by the preset blacklist search engine, determine whether the blacklist data is erroneous data, and set the blacklist information list of the determined erroneous data to an erroneous state. At the same time, whether the erroneous data needs to be output through the review result confirmation module 760 is determined based on whether the preset review result requires manual confirmation. If the preset review result requires manual confirmation, the blacklist data is sent to the review result confirmation module 760 via the data transmission module 730 for review by business personnel. Alternatively, the review result can also be sent to the review result output module 750 to be displayed to business personnel. Figure 7 As shown, the detection result review module 740 includes a review unit 741 and a second model synchronization unit 742. The review unit 741 is used to review the blacklist data using the operation S440 described above to determine whether the blacklist data is erroneous data. The second model synchronization unit 742 is used to synchronize and update the preset machine learning model used by the review unit 741 in real time.

[0111] The audit result output module 750 is used to display the audit results of the detection result audit module 740, including whether it is false data and detailed information on the blacklist data that is hit data.

[0112] The audit result confirmation module 760 is used to display a human-computer interaction interface, which can, for example, display blacklist data to be manually reviewed and a control for business personnel to select whether the audit result is passed.

[0113] The data storage module 770 is used to store data involved in the process of data processing by the blacklist screening device 700, such as image pictures, structured data, audit results and model data.

[0114] Based on the above blacklist detection method, the present disclosure also provides a blacklist detection device. Figure 8 The device for detecting the blacklist is described in detail.

[0115] Figure 8 The structure block diagram of the device for detecting a blacklist according to an embodiment of the present disclosure is schematically shown.

[0116] like Figure 8 As shown, the apparatus 800 for detecting a blacklist of this embodiment may include, for example, an information acquisition module 810 , an information parsing module 820 , and a blacklist determination module 830 .

[0117] The information acquisition module 810 is used to acquire information to be processed. In one embodiment, the information acquisition module 810 can be used to perform the operation S210 described above, which will not be described in detail here.

[0118] The information parsing module 820 is used to parse the information to be processed using a semantic analysis model to obtain a plurality of structured data for the information to be processed. In one embodiment, the information parsing module 820 can be used to perform the operation S220 described above, which will not be described in detail here.

[0119] The blacklist determination module 830 is used to use a preset blacklist search engine to determine the blacklist data in the plurality of structured data. In one embodiment, the blacklist determination module 830 can be used to perform the operation S230 described above, which will not be described in detail here.

[0120] According to an embodiment of the present disclosure, the information to be processed includes multiple pieces of information. The above-mentioned information acquisition module 810 may, for example, include an information identification submodule and an information parsing submodule. Among them, the information identification submodule is used to identify the information to be processed to determine at least one piece of information with a predetermined identifier in the information to be processed as the information to be parsed. The information parsing submodule is used to parse each piece of information in the information to be parsed using a semantic analysis model to obtain structured data representing each piece of information. In one embodiment, the information identification submodule and the information parsing submodule may, for example, be used to respectively perform operations S321 to S322 described above, which will not be repeated here.

[0121] According to an embodiment of the present disclosure, the above-mentioned device 800 for detecting a blacklist may further include, for example, an erroneous data determination module, which is used to determine the erroneous data in the blacklist data according to a preset rule.

[0122] According to an embodiment of the present disclosure, the erroneous data determination module is specifically used to determine, for example, data in the blacklist data that belongs to a preset whitelist as erroneous data.

[0123] According to an embodiment of the present disclosure, the information to be processed includes information generated by processing a business. The above-mentioned erroneous data determination module is specifically used to determine, for example, that the data in the blacklist data that meets the preset erroneous rules is erroneous data. Among them, the preset erroneous rules include at least one of the following: the blacklist data is the same as the first data that has been confirmed to be erroneous, and the first data and the blacklist data are data generated by processing the same business; the blacklist data indicates a business object, and the type of the business object indicated by the blacklist data is inconsistent with the type of the business object targeted by the business; the blacklist data is part of the preset information; the blacklist data includes unstructured data; or the blacklist data is data of a preset type.

[0124] According to an embodiment of the present disclosure, the above-mentioned erroneous data determination module may include, for example, a probability determination submodule and an erroneous information determination submodule. The probability determination submodule is used to determine the probability that each blacklist data in the blacklist data belongs to erroneous data by using a preset machine learning model. The erroneous information determination submodule is used to determine whether each blacklist data belongs to erroneous information based on the probability that each blacklist data belongs to erroneous data and a preset threshold.

[0125] According to an embodiment of the present disclosure, the above-mentioned false hit information determination submodule can be specifically used to determine that each blacklist data is false hit data when the probability that each blacklist data is false hit data is greater than or equal to a first preset threshold.

[0126] According to an embodiment of the present disclosure, the above-mentioned false hit information determination submodule can also be used, for example, to determine that each blacklist data is suspected false hit data when the probability that each blacklist data is false hit data is less than a first preset threshold and is greater than or equal to a second preset threshold; and to determine that each blacklist data is hit data when the probability that each blacklist data is false hit data is less than the second preset threshold.

[0127] According to an embodiment of the present disclosure, the preset machine learning model includes at least two models of different types, and the above-mentioned probability determination submodule may, for example, include an input data determination unit, a model processing unit, and a probability determination unit. The input data determination unit is used to determine the input data of each model in the at least two models according to each blacklist data. The model processing unit is used to input each model with the input data, and output output information for each blacklist data to obtain at least two output information for each blacklist data. The probability determination unit is used to determine the probability that each blacklist data belongs to the false data according to the at least two output information for each blacklist data.

[0128] According to an embodiment of the present disclosure, any multiple of the information acquisition module 810, the information parsing module 820 and the blacklist determination module 830 can be combined in one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the information acquisition module 810, the information parsing module 820 and the blacklist determination module 830 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or encapsulating the circuit, or implemented in any one of the three implementation modes of software, hardware and firmware or in an appropriate combination of any of them. Alternatively, at least one of the information acquisition module 810, the information parsing module 820 and the blacklist determination module 830 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.

[0129] Fig. 9 A block diagram of an electronic device suitable for implementing a method for detecting a blacklist according to an embodiment of the present disclosure is schematically shown.

[0130] like Fig. 9 As shown, the electronic device 900 according to an embodiment of the present disclosure includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage part 908 into a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include an onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0131] In RAM 903, various programs and data required for the operation of electronic device 900 are stored. Processor 901, ROM 902 and RAM 903 are connected to each other via bus 904. Processor 901 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 902 and / or RAM 903. It should be noted that the program can also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0132] According to an embodiment of the present disclosure, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to the bus 904. The electronic device 900 may further include one or more of the following components connected to the I / O interface 905: an input portion 906 including a keyboard, a mouse, etc.; an output portion 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 908 including a hard disk, etc.; and a communication portion 909 including a network interface card such as a LAN card, a modem, etc. The communication portion 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I / O interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage portion 908 as needed.

[0133] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0134] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus, or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 902 and / or RAM 903 described above and / or one or more memories other than ROM 902 and RAM 903.

[0135] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the item recommendation method provided by the embodiment of the present disclosure.

[0136] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 901. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0137] In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and downloaded and installed through the communication part 909, and / or installed from a removable medium 911. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0138] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0139] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0140] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0141] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0142] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A method for detecting a blacklist, comprising: Get the information to be processed; Parsing the information to be processed by using a semantic analysis model to obtain a plurality of structured data for the information to be processed; as well as Using a preset blacklist search engine to determine the blacklist data in the plurality of structured data; The information to be processed includes multiple pieces of information; and the semantic analysis model is used to parse the information to be processed, including: Identifying the information to be processed to determine at least one piece of information with a predetermined identifier in the information to be processed as the information to be parsed; and Using the semantic analysis model to parse each piece of information in the information to be parsed, obtaining structured data representing each piece of information, wherein there is a dependency relationship between words constituting the structured data, and the dependency relationship enables the structured data to represent each piece of information in the information to be parsed; The method further comprises: Using a preset machine learning model, determining the probability that each blacklist data in the blacklist data is a false positive data; and Determining whether each blacklist data is false information according to the probability that each blacklist data is false information and a preset threshold; The preset machine learning model includes at least two models of different types, and determining the probability that each blacklist data in the blacklist data is erroneous data includes: Determining input data of each model in the at least two models according to each blacklist data; Inputting the input data into each model, and outputting output information for each blacklist data, so as to obtain at least two output information for each blacklist data; and According to the at least two output information for each blacklist data, the probability that each blacklist data is erroneous data is determined.

2. The method according to claim 1, further comprising, after determining the blacklist data in the plurality of structured data: The misselected data in the blacklist data is determined according to a preset rule.

3. The method according to claim 2, wherein: Determining the erroneous data in the blacklist data according to the preset rules includes: Determine that the data in the blacklist data that belongs to the preset whitelist is the mis-selected data.

4. The method according to claim 3, wherein: The information to be processed includes information generated by processing business; Determining the erroneous data in the blacklist data according to the preset rule includes: determining the data in the blacklist data that meets the preset erroneous rule as the erroneous data, and the preset erroneous rule includes at least one of the following: The blacklist data is the same as the first data that has been confirmed to be in error, and the first data and the blacklist data are data generated for processing the same service; The blacklist data indicates a service object, and the type of the service object indicated by the blacklist data is inconsistent with the type of the service object targeted by the service; The blacklist data is part of the preset information; The blacklist data includes unstructured data; or The blacklist data is data of a preset type.

5. The method according to claim 1, wherein: Determining whether each blacklist data is erroneous information includes: When the probability that each blacklist data is erroneous data is greater than or equal to a first preset threshold, it is determined that each blacklist data is erroneous data.

6. The method according to claim 5, wherein: Determining whether each blacklist data is erroneous information further includes: When the probability that each blacklist data is erroneous data is less than the first preset threshold and is greater than or equal to the second preset threshold, determining that each blacklist data is suspicious erroneous data; When the probability that each blacklist data is false hit data is less than the second preset threshold, it is determined that each blacklist data is hit data.

7. A device for detecting a blacklist, comprising: An information acquisition module, used to acquire information to be processed; An information parsing module, used for parsing the information to be processed by using a semantic analysis model to obtain a plurality of structured data for the information to be processed; as well as A blacklist determination module, configured to determine blacklist data in the plurality of structured data using a preset blacklist search engine; The information to be processed includes multiple pieces of information; and the semantic analysis model is used to parse the information to be processed, including: Identifying the information to be processed to determine at least one piece of information with a predetermined identifier in the information to be processed as the information to be parsed; and Using the semantic analysis model to parse each piece of information in the information to be parsed, obtaining structured data representing each piece of information, wherein there is a dependency relationship between words constituting the structured data, and the dependency relationship enables the structured data to represent each piece of information in the information to be parsed; The device also includes: a module for determining the probability of false positive data, for determining the probability of each blacklist data in the blacklist data being false positive data by using a preset machine learning model; and An erroneous information determination module, used to determine whether each blacklist data belongs to erroneous information according to the probability that each blacklist data belongs to erroneous data and a preset threshold; The preset machine learning model includes at least two models of different types, and the false data probability determination module includes: An input data determination submodule, used to determine input data of each model in the at least two models according to each blacklist data; an output information determination submodule, configured to input the input data into each model, and output output information for each blacklist data, so as to obtain at least two output information for each blacklist data; and The blacklist data determination submodule is used to determine the probability that each blacklist data is erroneous data according to at least two output information for each blacklist data.

8. An electronic device comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors are enabled to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images