A method and system for identifying risk data
By adopting a two-step analysis method of semantic recognition and deep learning models in the aviation system, the interactive data on the server is monitored, and the problem of insufficient risk analysis on the server in the existing technology is solved, and the accuracy and efficiency of risk data identification is improved.
Patent Information
- Application Number
- CN202210048892.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-17
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-01-17
AI Technical Summary
The existing technology mainly conducts risk analysis from the aircraft side, but fails to meet the comprehensive and high-standard requirements of aviation systems for safety, especially the lack of risk analysis on the server side (such as the control side and the customer service side).
Provide a risk data identification method and system, through continuous monitoring of the interactive data in the communication, semantic recognition and deep learning models are used to perform two-step analysis, extract the interactive content sequence and determine whether there is a risk, ensuring the accurate identification of risk data.
When the service is connected to the server and the user, the interactive data is continuously monitored, which improves the recognition efficiency of risk data, reduces the processing volume of non-essential identification, ensures a low omission rate, and realizes accurate analysis of risk data.
Smart Images

Figure CN114444511B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the fields of aviation and information technology, and in particular to a risk data identification method and system. Background Art
[0002] Aviation safety is to ensure that there are no accidents related to aircraft operations, such as casualties and damage to aircraft. Aviation safety mainly includes flight safety, aviation ground safety and air defense safety, including: 1. Flight safety: no accidents such as casualties and damage to aircraft caused by flight or other reasons during aircraft operation. 2. Aviation ground safety: the safety of production activities carried out within the apron and flight area around aircraft operations. Prevent aircraft damage, casualties of passengers and ground personnel, and damage to various ground facilities. It also includes the safety of aircraft maintenance, loading and unloading of cargo and service supplies, aircraft refueling, and the safety of military aircraft weapons and ammunition. 3. Security safety: prevent illegal interference activities that affect the normal operation of aircraft and directly endanger flight safety, and prevent accidental firing of ground weapons.
[0003] After searching, the following prior art related to aviation safety was found:
[0004] Patent document 1 (CN13626616A) discloses an aircraft safety warning method, including: continuously monitoring changes in multiple heterogeneous data sources; for the changed data sources, extracting entities and their relationships in the data sources, and using the relationships to associate the entities to form multiple knowledge graphs corresponding to the data sources; according to the ontology model, fusing the multiple knowledge graphs corresponding to the data sources, wherein the ontology model is modeled according to aviation safety threat events; according to the warning logic rules and the knowledge graph, predicting the danger of the target aircraft; and issuing a danger warning to the ground warning terminal so that the air traffic controller can respond to the danger warning and issue a command to the target aircraft.
[0005] Patent document 2 (CN112668057A) discloses a security system based on aviation big data, including: an aviation big data source, an aviation data processing device, an aviation data storage device and an aviation data application device, wherein the aviation big data source is electrically connected to the aviation data processing device, and the aviation data processing device is electrically connected to the aviation data storage device and the aviation data application device at the same time; the aviation big data source is used to obtain personal identity information, flight record information, credit information, travel trajectory information and social relationship information from an external system, and send the personal identity information, flight record information, credit information, travel trajectory information and external social relationship information to the aviation data processing device; the aviation data processing device is used to standardize the received personal identity information, flight record information, credit information, travel trajectory information and social relationship information, and send the standardized data to the aviation data storage device and the aviation data application device at the same time; the aviation data application device is used to receive the personal identity information, flight record information, credit information, travel trajectory information and social relationship information after data standardization, and detect and judge abnormal information according to the personal identity information, flight record information, credit information, travel trajectory information and social relationship information.
[0006] Patent document 3 (CN112885153A) A general aviation safety monitoring system based on multi-network integration, comprising: an airborne terminal system and a ground terminal system; the airborne terminal system is installed in a general aviation aircraft, including a flight data acquisition equipment group, a central computer, a picture transmission radio and a multi-network integration equipment group; the flight data acquisition equipment group includes one or more sensor devices for collecting flight data of the general aviation aircraft, and the flight data is used to remotely monitor the flight status of the general aviation aircraft; the multi-network integration equipment group includes a public network signal sending device, a Beidou signal sending device and a Shulaya signal sending device; the ground terminal system is installed in a ground monitoring station, including a picture transmission base station, a multi-network integration receiver group and a display device group; the multi-network integration receiver group includes a public network signal receiver, a Beidou signal receiver and a Shulaya signal receiver; the display device group includes one or more from the picture transmission base station, the public network signal receiver The image transmission station, the public network signal sending device, the Beidou signal sending device and the Shulaya signal sending device respectively establish communication with the image transmission base station, the public network signal receiver, the Beidou signal receiver and the Shulaya signal receiver to transmit flight data; the central computer receives the flight data collected by each sensor device in the flight data acquisition device group, determines the network transmission link by analyzing one or more of the ground-air environment in which the general aviation aircraft is located, the type of mission performed, and the type of flight data, and starts one or more of the image transmission station, the public network signal sending device, the Beidou signal sending device and the Shulaya signal sending device according to the network transmission link to establish communication with the corresponding equipment in the ground terminal system to transmit flight data.
[0007] It can be found that the patent documents involved in the above-mentioned prior art basically conduct risk analysis from the aircraft side, but rarely conduct risk analysis on the service side (such as the control side, customer service side, etc.), which cannot meet the aviation system's comprehensive and high-standard requirements for safety. Summary of the invention
[0008] In order to at least solve the technical problems existing in the above-mentioned background technology, the present invention provides a risk data identification method, system, electronic device and storage medium.
[0009] A first aspect of the present invention provides a risk data identification method, comprising the following steps:
[0010] S10, in response to the service connection signal, continuously monitoring the interaction data during the connection process;
[0011] S20, performing semantic recognition on the interaction data to extract an interaction content sequence;
[0012] S30, performing a first calculation on the interactive content sequence, if a first result of the first calculation satisfies a first condition, then go to S40, otherwise go to S10;
[0013] S40, performing a second calculation on the interaction content sequence, and if a second result of the second calculation satisfies a second condition, outputting a prompt message indicating that the interaction data has a risk, otherwise, turning to S10.
[0014] Optionally, in step S20, performing semantic recognition on the interaction data to extract an interaction content sequence includes:
[0015] Converting the interactive data into standard text data, segmenting the standard text data to obtain a text matrix, wherein the text matrix includes a plurality of words;
[0016] Matching and calculating the text matrix with each subject item to obtain a target subject item matrix;
[0017] The text matrix and the target subject item matrix are used as the interactive content sequence.
[0018] Optionally, the matching calculation of the text matrix with each subject item to obtain a target subject item matrix includes:
[0019] Calculating first similarities between each word in the text matrix and each subject item respectively;
[0020] The second similarity between the text matrix and each subject item is calculated based on the first similarity of each word, and the target subject item matrix matching the text matrix is determined based on the second similarity.
[0021] Optionally, the calculating the second similarity between the text matrix and each subject item based on the first similarity of each word includes:
[0022] assigning each word to each topic item based on the first similarity and a first threshold;
[0023] The second similarity between the text matrix and each subject item is calculated by the following formula:
[0024]
[0025] In the formula, z i = j represents assigning the text matrix to topic item j, that is, S(z i =j) represents the second similarity between the text matrix and each subject item, that is, the probability of the text matrix being assigned to the subject item j; n wrepresents the number of specific words assigned to the subject item j in the text matrix, the specific words are pre-set for each subject item and are used to describe the core connotation of the subject item, and δ represents the number of specific words in the text matrix; n d Represents the number of all words in the text matrix that are assigned to subject items, k represents the number of subject items; α and β are preset adjustment parameters, wherein α+β=1, and is determined based on the equivalent serial number of the specific word in the text matrix. The larger the equivalent serial number, the larger α is, and the equivalent serial number can be the mean / weighted mean of the serial numbers of each specific word.
[0026] Optionally, in step S30, performing a first calculation on the interactive content sequence includes:
[0027] Extracting a first risk sequence based on the target subject item matrix;
[0028] Calculating a third similarity between the text matrix and the first risk sequence based on the assigned words in the text matrix, and taking the third similarity as a first result;
[0029] Correspondingly, the first result of the first calculation satisfies the first condition, specifically: the third similarity is greater than the second threshold.
[0030] Optionally, in step S40, performing a second calculation on the interactive content sequence includes:
[0031] Input the text matrix and the first risk sequence into a deep learning model, the deep learning model outputs a fourth similarity, and the fourth similarity is used as the second result;
[0032] Correspondingly, the second result of the second calculation satisfies the second condition, specifically: the fourth similarity is greater than the third threshold.
[0033] Optionally, after outputting the prompt information that the interactive data has a risk, the method further includes:
[0034] monitoring response data, wherein the response data includes positive responses and negative responses;
[0035] A training set is constructed based on the text matrix corresponding to the negative response and the first risk sequence, and the deep learning model is retrained based on the training set.
[0036] A second aspect of the present invention provides a risk data identification system, comprising a processing module, a storage module, and a communication module, wherein the processing module is connected to the storage module and the communication module respectively; wherein:
[0037] The storage module is used to store executable computer program code;
[0038] The communication module is used to continuously monitor the interaction data during the connection process in response to the service connection signal, and transmit the interaction data to the processing module;
[0039] The processing module is used to execute the method as described in any of the preceding items by calling the executable computer program code in the storage module.
[0040] A third aspect of the present invention provides an electronic device, comprising:
[0041] A memory storing executable program code;
[0042] a processor coupled to the memory;
[0043] The processor calls the executable program code stored in the memory to execute the method as described in any of the preceding items.
[0044] A fourth aspect of the present invention provides a computer storage medium having a computer program stored thereon, and the computer program, when executed by a processor, executes any of the methods described above.
[0045] The beneficial effect of the present invention is that when the service end and the user end are connected, the interactive data is continuously monitored, and the interactive content sequence is extracted therefrom. The interactive content sequence is first calculated to preliminarily determine whether there is a risk in the interactive content, and then the interactive content sequence preliminarily determined to have a risk is calculated again to finally determine whether there is a risk, thereby realizing the analysis of risk data. The scheme of the present invention designs a two-step analysis method for risk data, which not only ensures a low omission rate of risk data identification, but also reduces the processing volume of unnecessary identification, thereby improving identification efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.
[0047] Figure 1 It is a flowchart of a risk data identification method disclosed in an embodiment of the present invention;
[0048] Figure 2 It is a structural schematic diagram of a risk data identification system disclosed in an embodiment of the present invention;
[0049] Figure 3 It is a structural schematic diagram of an electronic device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION
[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0051] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.
[0052] In the description of the present invention, it should be noted that if the terms "upper", "lower", "inside", "outside", etc. appear to indicate an orientation or position relationship, they are based on the orientation or position relationship shown in the accompanying drawings, or are the orientation or position relationship in which the product of the invention is usually placed when used. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.
[0053] The terms "first", "second", "third", and "fourth" in the specification and claims of the present invention are used to distinguish different objects rather than to describe a specific order of the objects. For example, the first input, the second input, the third input, and the fourth input are used to distinguish different inputs rather than to describe a specific order of the inputs.
[0054] In the embodiments of the present invention, words such as "exemplarily" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplarily" or "for example" in the embodiments of the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplarily" or "for example" is intended to present related concepts in a specific way.
[0055] In the description of the embodiments of the present invention, unless otherwise specified, “multiple” means two or more than two. For example, multiple processing units refer to two or more processing units; multiple elements refer to two or more elements, etc.
[0056] It should be noted that, in the absence of conflict, the features in the embodiments of the present invention may be combined with each other.
[0057] Embodiment 1
[0058] See also Figure 1 , Figure 1 FIG. 1 is a flow chart of a risk data identification method disclosed in an embodiment of the present invention. Figure 1 As shown, a risk data identification method according to an embodiment of the present invention includes the following steps:
[0059] S10, in response to the service connection signal, continuously monitoring the interaction data during the connection process;
[0060] S20, performing semantic recognition on the interaction data to extract an interaction content sequence;
[0061] S30, performing a first calculation on the interactive content sequence, if a first result of the first calculation satisfies a first condition, then go to S40, otherwise go to S10;
[0062] S40, performing a second calculation on the interaction content sequence, and if a second result of the second calculation satisfies a second condition, outputting a prompt message indicating that the interaction data has a risk, otherwise, turning to S10.
[0063] In the embodiment of the present invention, the user end (such as the aircraft end, the client end, the authorized user end, etc.) will send request information to the server end, for example, to inquire about the weather, aircraft traffic, service information, etc. in the area where the airport is located, request to land, request to jump the queue to land, request to land on runway X, etc. After receiving the request information from the user end, the server end will have a full-time staff (such as a controller) to provide corresponding answers, operations and other services. At this time, the system responds to the service connection signal and continuously monitors the interactive data in the connection process, and extracts the content involved in the interaction through semantic recognition technology, and then determines whether the answers, operations, etc. involved in the interactive data are risky. If there is a risk, a prompt message is output.
[0064] Among them, the present invention sets the identification of risk data in two steps, that is, the extracted interactive content sequence is first calculated, and if the calculation result meets the first condition (has obvious risk characteristics), it is preliminarily identified as risk data, and then the interactive content sequence that meets the first condition is calculated. If it also meets the second condition, it is finally confirmed that there is risk data in the interactive data. By setting a two-step identification method for risk data, the present invention greatly improves the efficiency of risk data identification compared to the indiscriminate risk data identification method.
[0065] It should be noted that the solution of the present invention can be implemented in various aviation control platforms, in which the service connection may be provided by, for example, a controller, and the connection request may be provided by various civil aircraft, helicopters, etc., and the interactive data may be either voice data or text content (such as a short message). The various aviation control platforms in the present invention may be interactive platforms built on servers, and the server should not be simply understood as one or a class of servers, but various possible forms of servers deployed in actual applications to support applications or web pages, such as cloud servers, server clusters, etc.
[0066] Optionally, in step S20, performing semantic recognition on the interaction data to extract an interaction content sequence includes:
[0067] Converting the interactive data into standard text data, segmenting the standard text data to obtain a text matrix, wherein the text matrix includes a plurality of words;
[0068] Matching and calculating the text matrix with each subject item to obtain a target subject item matrix;
[0069] The text matrix and the target subject item matrix are used as the interactive content sequence.
[0070] In an embodiment of the present invention, for interactive data, especially voice-type interactive data, a large number of worthless words will be included, such as various modal particles, repeated content, format content, etc. The existence of these data may affect the subsequent recognition of interactive content. Therefore, the present invention first pre-processes the interactive data to convert it into standard text data. For example, a database is established based on format content, typical modal particles, etc., and such words are deleted by comparison. For repeated content, it can be identified and deleted by similarity comparison. After obtaining relatively "clean" standard text data, it can be segmented into words to obtain a text matrix. Since text segmentation belongs to mature prior art, the present invention will not be repeated here.
[0071] In addition, the matching degree between the text matrix and each subject item is further calculated to determine the target subject item matrix, and the two are used as the interactive content sequence. When implemented specifically, the two can also be integrated, for example, into one matrix. Among them, the present invention pre-sets a plurality of subject items, for example, a landing item, a take-off item, a ground safety item, a meteorological item, a fire-fighting and medical care item, etc. By calculating the similarity (for example, Euclidean distance, Manhattan distance, etc.) between each word in the text matrix and each subject item, each word can be assigned to the corresponding subject item, for example, the word "request to land" is assigned to the subject item "landing item", the word "fire" is assigned to the subject item "fire-fighting and medical care item", etc.
[0072] It needs to be explained that the text matrix generally only includes words with real meanings, such as "ready, ready to taxi out (airplane)", "can taxi out (tower)", "ready to taxi out, request to take off (airplane)", "can take off (tower)", etc. Various conjunctions, such as "of", are also deleted.
[0073] Optionally, the matching calculation of the text matrix with each subject item to obtain a target subject item matrix includes:
[0074] Calculating first similarities between each word in the text matrix and each subject item respectively;
[0075] The second similarity between the text matrix and each subject item is calculated based on the first similarity of each word, and the target subject item matrix matching the text matrix is determined based on the second similarity.
[0076] In an embodiment of the present invention, the first similarity between each word in the text matrix and all the subject items is calculated first, so as to determine the matching status between a single word and each subject item; then, based on the first similarities corresponding to all the words, the second similarity between the text matrix as a whole and each subject item is calculated, so as to finally determine the matching status between the text matrix and each subject item.
[0077] Optionally, the calculating the second similarity between the text matrix and each subject item based on the first similarity of each word includes:
[0078] assigning each word to each topic item based on the first similarity and a first threshold;
[0079] The second similarity between the text matrix and each subject item is calculated by the following formula:
[0080]
[0081] In the formula, z i = j represents assigning the text matrix to topic item j, that is, S(z i =j) represents the second similarity between the text matrix and each subject item, that is, the probability of the text matrix being assigned to the subject item j; n w represents the number of specific words assigned to the subject item j in the text matrix, the specific words are pre-set for each subject item and are used to describe the core connotation of the subject item, and δ represents the number of specific words in the text matrix; n dRepresents the number of all words in the text matrix that are assigned to subject items, k represents the number of subject items; α and β are preset adjustment parameters, wherein α+β=1, and is determined based on the equivalent serial number of the specific word in the text matrix. The larger the equivalent serial number, the larger α is, and the equivalent serial number can be the mean / weighted mean of the serial numbers of each specific word.
[0082] In the embodiment of the present invention, the present invention first allocates each word with a first similarity greater than a first threshold to a corresponding subject project. On this basis, the present invention also designs specific words, which are preset and most accurate words for describing the subject project. For example, the specific words of "weather project" include rain, fog, wind, hail, etc. The more such words in the text matrix, the higher the similarity between the text matrix and the subject project. Therefore, when the second similarity is greater than the threshold, it can be determined that the text matrix matches the subject project, that is, the target subject project is obtained.
[0083] Optionally, in step S30, performing a first calculation on the interactive content sequence includes:
[0084] Extracting a first risk sequence based on the target subject item matrix;
[0085] Calculating a third similarity between the text matrix and the first risk sequence based on the assigned words in the text matrix, and taking the third similarity as a first result;
[0086] Correspondingly, the first result of the first calculation satisfies the first condition, specifically: the third similarity is greater than the second threshold.
[0087] In an embodiment of the present invention, after the interactive content sequence is obtained, the first risk sequence can be extracted based on the target subject project matrix, wherein the first risk sequence is pre-set for each subject project, for example, the first risk sequence of the subject project "weather project" includes "strong wind + take-off", which means that take-off is not allowed during strong winds. Next, the third similarity is calculated between the non-specific words in the text matrix and the first risk sequence. For example, if the text matrix includes the non-specific words "wind force 9" and "take-off allowed", it is determined that the third similarity is high, that is, the first result indicates that the text matrix matches the first risk sequence, and is preliminarily determined to be risky.
[0088] Optionally, in step S40, performing a second calculation on the interactive content sequence includes:
[0089] Input the text matrix and the first risk sequence into a deep learning model, the deep learning model outputs a fourth similarity, and the fourth similarity is used as the second result;
[0090] Correspondingly, the second result of the second calculation satisfies the second condition, specifically: the fourth similarity is greater than the third threshold.
[0091] In an embodiment of the present invention, unlike the first calculation, the second calculation of the present invention is implemented through a deep learning model. The pre-trained deep learning model has higher accuracy and adaptability, and is particularly suitable for the establishment of no clear functional relationship, but the amount of calculation is large and the timeliness is slightly poor. Therefore, the present invention is arranged to use a deep learning model in the final determination link of risk data. For the training of the deep learning model, a number of text matrices and corresponding first risk sequence combination training sets can be obtained from historical data, and input into the deep learning model to train it. When the number of iterations reaches the limit or the system converges, a deep learning model with higher accuracy can be obtained. If the second result output by the deep learning model also indicates the existence of risk data, it can be finally determined.
[0092] Optionally, after outputting the prompt information that the interactive data has a risk, the method further includes:
[0093] monitoring response data, wherein the response data includes positive responses and negative responses;
[0094] A training set is constructed based on the text matrix corresponding to the negative response and the first risk sequence, and the deep learning model is retrained based on the training set.
[0095] In an embodiment of the present invention, in the aviation field, various air traffic control rules will fluctuate under different weather, traffic volume and other actual conditions, and there is no clear and unique correspondence between the words in the text matrix and the first risk sequence, so the deep learning model needs to be continuously retrained to ensure the accuracy of risk data identification. Specifically, after outputting the prompt information that there is a risk in the interactive data, the present invention continues to monitor the response data of the server, such as whether the controller accepts the prompt information, and the monitoring can be achieved by detecting the controller's selection of the operation button corresponding to the prompt information; then, a training set is constructed based on the text matrix corresponding to the negative response and the first risk sequence, and the deep learning model is retrained based on the training set, so that the ear deep learning model obtained by training is more accurate.
[0096] Embodiment 2
[0097] See also Figure 2 , Figure 2 Schematic diagram of the structure of a risk data identification system disclosed in an embodiment of the present invention. Figure 2As shown, a risk data identification system (100) according to an embodiment of the present invention comprises a processing module (101), a storage module (102), and a communication module (103), wherein the processing module (101) is connected to the storage module (102) and the communication module (103) respectively; wherein,
[0098] The storage module (102) is used to store executable computer program code;
[0099] The communication module (103) is used to continuously monitor the interaction data during the connection process in response to the service connection signal, and transmit the interaction data to the processing module (101);
[0100] The processing module (101) is used to execute the method described in the first embodiment by calling the executable computer program code in the storage module (102).
[0101] The specific functions of a risk data identification system in this embodiment refer to the above-mentioned embodiment 1. Since the system in this embodiment adopts all the technical solutions of the above-mentioned embodiments, it at least has all the beneficial effects brought by the technical solutions of the above-mentioned embodiments, which will not be described one by one here.
[0102] Embodiment 3
[0103] See also Figure 3 , Figure 3 An electronic device disclosed in an embodiment of the present invention includes:
[0104] A memory storing executable program code;
[0105] a processor coupled to the memory;
[0106] The processor calls the executable program code stored in the memory to execute the method described in the first embodiment.
[0107] Embodiment 4
[0108] The embodiment of the present invention further discloses a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method described in the first embodiment is executed.
[0109] It should be noted that the storage module (102) in the second embodiment of the present invention, the memory in the third embodiment, and the computer storage medium in the fourth embodiment can all be, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disc-read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, etc.
[0110] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0111] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0112] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0113] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.
Claims
1. A risk data identification method, characterized in that: The steps include: S10, in response to the service connection signal, continuously monitoring the interaction data during the connection process; S20, performing semantic recognition on the interaction data to extract an interaction content sequence; S30, performing a first calculation on the interactive content sequence, if a first result of the first calculation satisfies a first condition, then go to S40, otherwise go to S10; S40, performing a second calculation on the interaction content sequence, and if a second result of the second calculation satisfies a second condition, outputting a prompt message indicating that the interaction data has a risk, otherwise, turning to S10; In step S20, the semantic recognition of the interaction data to extract the interaction content sequence includes: Converting the interactive data into standard text data, segmenting the standard text data to obtain a text matrix, wherein the text matrix includes a plurality of words; Matching and calculating the text matrix with each subject item to obtain a target subject item matrix; Using the text matrix and the target subject item matrix as the interactive content sequence; The matching calculation of the text matrix with each subject item to obtain a target subject item matrix includes: Calculating first similarities between each word in the text matrix and each subject item respectively; Calculating a second similarity between the text matrix and each subject item based on the first similarity of each word, and determining a target subject item matrix that matches the text matrix based on the second similarity; The calculating the second similarity between the text matrix and each subject item based on the first similarity of each word includes: assigning each word to each topic item based on the first similarity and a first threshold; The second similarity between the text matrix and each subject item is calculated by the following formula: In the formula, z i = j represents assigning the text matrix to topic item j, that is, S(z i =j) represents the second similarity between the text matrix and each subject item, that is, the probability of the text matrix being assigned to the subject item j; n w represents the number of specific words assigned to the subject item j in the text matrix, the specific words are pre-set for each subject item and are used to describe the core connotation of the subject item, and δ represents the number of specific words in the text matrix; n d represents the number of all words in the text matrix that are assigned with subject items, and k represents the number of subject items; α and β are preset adjustment parameters, wherein α+β=1, and are determined according to the equivalent sequence number of the specific word in the text matrix. The larger the equivalent sequence number, the larger α is, and the equivalent sequence number is the mean / weighted mean of the sequence numbers of each specific word; In step S30, performing a first calculation on the interactive content sequence includes: Extracting a first risk sequence based on the target subject item matrix; Calculating a third similarity between the text matrix and the first risk sequence based on the assigned words in the text matrix, and taking the third similarity as a first result; Accordingly, the first result of the first calculation satisfies the first condition, specifically: the third similarity is greater than the second threshold; In step S40, performing a second calculation on the interactive content sequence includes: Input the text matrix and the first risk sequence into a deep learning model, the deep learning model outputs a fourth similarity, and the fourth similarity is used as the second result; Correspondingly, the second result of the second calculation satisfies the second condition, specifically: the fourth similarity is greater than the third threshold.
2. A risk data identification method according to claim 1, characterized in that: After outputting the prompt information that the interactive data has risks, the method further includes: monitoring response data, wherein the response data includes positive responses and negative responses; A training set is constructed based on the text matrix corresponding to the negative response and the first risk sequence, and the deep learning model is retrained based on the training set.
3. A risk data identification system, comprising a processing module, a storage module, and a communication module, wherein the processing module is connected to the storage module and the communication module respectively; wherein: The storage module is used to store executable computer program code; The communication module is used to continuously monitor the interaction data during the connection process in response to the service connection signal, and transmit the interaction data to the processing module; It is characterized in that: the processing module is used to execute the method as claimed in claim 1 or 2 by calling the executable computer program code in the storage module.
4. An electronic device, comprising: A memory storing executable program code; a processor coupled to the memory; It is characterized in that: the processor calls the executable program code stored in the memory to execute the method as claimed in claim 1 or 2.
5. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to claim 1 or 2 is performed.
