Emergency detection method, device, electronic device and readable storage medium

Abstract

The emergency event detection method, device, electronic device and readable storage medium provided by the present invention include: obtaining first text data of a current time window, second text data of multiple historical time windows, and third text data of a target sub-time window; determining a plurality of target phrases from the third text data, the number of each target phrase in the first text data, the second text data and the third text data, and the number of times it is transferred to manual processing; for each target phrase, determining the corresponding burst index set according to the number of target phrases, the total number of texts corresponding to the first text data, the second text data and the third text data, the phrase type and the number of times it is transferred to manual processing; if the target burst index in the burst index set is greater than the threshold value corresponding to the target burst index, the target phrase is alarmed. The present invention reduces the detection time granularity, and uses different burst indexes to perform burst detection, so that the detection result is accurate and reliable.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an emergency detection method, device, electronic device and readable storage medium. Background Art

[0002] In intelligent customer service, users ask a large number of questions every day. For some routine questions, such as product functions, operating manuals, etc., there is usually a specific response strategy. However, for an emergency problem, such as system anomalies, policy changes, etc., there is no specific response strategy. If such emergencies can be discovered in time, operations can quickly intervene to develop a treatment plan and reduce the impact of the emergency in time.

[0003] However, existing emergency detection schemes often have delayed false alarms, which affects the accuracy of emergency detection. Therefore, how to provide an accurate emergency detection scheme is a technical problem that needs to be solved. Summary of the invention

[0004] One of the purposes of the present invention is to provide a method, device, electronic device and readable storage medium for detecting an emergency, so as to improve the accuracy of detecting an emergency.

[0005] In a first aspect, the present invention provides a method for detecting an emergency, the method comprising:

[0006] Acquire first text data generated in a current time window and second text data generated in multiple historical time windows, and extract third text data from the first text data according to a target sub-time window; wherein the target sub-time window is located in the current time window;

[0007] Determine a plurality of target phrases from the third text data, and the number of each target phrase in the first text data, the second text data, and the third text data and the number of manual processing times;

[0008] For each of the target phrases, a burst index set corresponding to the target phrase is determined according to the number of the target phrases, the total number of texts corresponding to the first text data, the second text data, and the third text data, the phrase type, and the number of times of switching to manual processing; wherein the burst index set includes a target sub-time window burst index; a manual switching burst index; a current time window burst index or a quantity ratio burst index;

[0009] If the target burst index in the burst index set is greater than a threshold value corresponding to the target burst index, an alarm process is performed on the determined target phrase; wherein the target burst index is any one in the burst index set.

[0010] In a second aspect, the present invention provides an emergency event detection device, comprising:

[0011] An acquisition module, used to acquire first text data generated in a current time window, second text data generated in multiple historical time windows, and extract third text data from the first text data according to a target sub-time window; wherein the target sub-time window is located in the current time window;

[0012] A determination module, used to determine a number of target phrases from the third text data, and the number of each target phrase in the first text data, the second text data and the third text data and the number of manual processing times;

[0013] The determination module is further used to determine, for each of the target phrases, a burst index set corresponding to the target phrase according to the number of the target phrases, the total number of texts corresponding to the first text data, the second text data, and the third text data, the phrase type, and the number of times of manual processing; wherein the burst index set includes a target sub-time window burst index; a manual processing burst index; a current time window burst index or a quantity ratio burst index;

[0014] The detection module is used to perform alarm processing on the determined target phrase if the target burst index in the burst index set is greater than a threshold corresponding to the target burst index; wherein the target burst index is any one in the burst index set.

[0015] In a third aspect, the present invention provides an electronic device, comprising a processor and a memory, wherein the memory stores a computer program executable by the processor, and the processor can execute the computer program to implement the emergency event detection method described in the first aspect.

[0016] In a fourth aspect, the present invention provides a readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the emergency event detection method described in the first aspect.

[0017] The emergency event detection method, device, electronic device and readable storage medium provided by the present invention first obtain first text data generated in the current time window, second text data generated in multiple historical time windows, and third text data extracted from the first text data, and then use the target phrase determined in the third text data, as well as the number of target phrases in the three types of text data and the number of manual processing times, and finally obtain multiple types of emergency indexes corresponding to the phrase based on the obtained number and the number of manual processing times, and compare each type of emergency index with a threshold value. If the emergency index exceeds the threshold value, an alarm is processed. It can be seen that the present application uses the text data generated in the current time window and multiple historical time windows to detect emergencies, avoiding the situation where inaccurate detection occurs due to incomplete information in a single time period, and further reduces the granularity of the detection time window to avoid the hugging phenomenon. In the detection process, a variety of emergency situations are fully considered, and different emergency indexes are used for early warning, and the detection results have certain reliability and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be 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 paying creative work.

[0019] Figure 1 A schematic diagram of an application scenario of the emergency event detection method provided in an embodiment of the present application;

[0020] Figure 2 Firstly, a structural block diagram of an electronic device is provided for the embodiment of the present application;

[0021] Figure 3 A schematic flow chart of an emergency event detection method provided in an embodiment of the present application;

[0022] Figure 4 A schematic flowchart of step S303 provided in an embodiment of the present application;

[0023] Figure 5 A schematic flowchart of step S303-1 provided in an embodiment of the present application;

[0024] Figure 6 A schematic flowchart of step S303-2 provided in an embodiment of the present application;

[0025] Figure 7 A comparison chart of the effects of quantity ratio test and sigma anomaly detection provided in the embodiment of the present application;

[0026] Figure 8 A functional module diagram of an emergency detection device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0028] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.

[0029] 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.

[0030] In the description of the present invention, it should be noted that if the terms "upper", "lower", "inside", "outside", etc. appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the drawings, or is the orientation or position relationship in which the product of the invention is usually placed when used. It is only for the convenience of describing the present invention and simplifying the description, and does 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.

[0031] In addition, the terms “first”, “second”, etc., if used, are merely used to distinguish between the descriptions and should not be understood as indicating or implying relative importance.

[0032] 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.

[0033] Please refer to Figure 1 , Figure 1A schematic diagram of an application scenario of the emergency detection method provided in an embodiment of the present application, wherein the application scenario may include multiple terminal devices 101 (including terminal devices 101-1, ... terminal devices 101-n), and a server 102; the terminal devices 101 and the server 102 are connected via a wireless or wired network. The terminal devices 101 include but are not limited to electronic devices such as desktop computers, mobile phones, mobile computers, tablet computers, media players, smart wearable devices, and smart TVs. The server 102 may be a single server or a server cluster consisting of several servers.

[0034] like Figure 1 As shown, applications may be installed in the terminal device 101, such as applications corresponding to social platforms, BBSs, forums, e-commerce platforms, etc. Users may post questions through the applications in the terminal device 101, and the data corresponding to these questions may be uploaded to the server 102 for storage. The server 102 may obtain the question data uploaded within a certain time period, and then perform emergency detection on these question data. Once an emergency is determined, an alarm message may be sent to relevant operation and maintenance personnel or supervisors.

[0035] Of course, the method provided in the embodiment of the present application is not limited to Figure 1 The application scenarios shown can also be used in other possible application scenarios, and the embodiments of the present application are not limited thereto. Figure 1 The functions that can be implemented by each device in the application scenario shown will be described in the subsequent method embodiments, and will not be described in detail here.

[0036] Continue to see Figure 1 At present, users post a large number of questions every day. For some routine issues, such as product functions, operation manuals, etc., there is usually a corresponding specific response strategy, but for an emergency problem, such as system abnormalities, changes in business activities, etc., there is no specific response strategy. If such emergencies can be discovered in time, operations can quickly intervene to formulate treatment plans, and the impact of emergencies can be reduced in time.

[0037] However, the existing emergency detection schemes often have delayed false alarms. For example, in the relevant detection scheme, emergency detection is performed based on the data within 1 hour before the current moment. Assuming that within this 1 hour, for the abnormal phrases that appeared in the first half hour and have been restored in the second half hour, no emergency warning is required. However, this detection method still calculates the data of the first half hour when comparing the data in the second half hour, and still makes emergency warnings, resulting in false alarms. Furthermore, the relevant technology proposes to shorten the duration of the above 1 hour to 10 minutes or half an hour. This method may cause some channels with a small number of users (such as the problem of anchor channels with relatively few users, the cumulative amount in 10 minutes may be only 2 or 3, and it can accumulate more in one hour) to be unable to report abnormal situations in time, resulting in delays.

[0038] In addition, if the number of questions increases significantly, if the intelligent customer service can solve them well, this can also be considered not to be an emergency (or not an emergency that should be paid attention to). For example, an activity guides users to apply for membership, and some less careful users may ask how to do it. At this time, the intelligent customer service can answer well and solve it. In this case, it can be considered not to be an emergency. However, the current separate proportion test method cannot accurately handle this situation, resulting in underreporting.

[0039] For example: in the current time window, there are a total of 120 sentences, the phrase [network, abnormal] appears 100 times, and [unable to, handle, member] appears 20 times, while in the comparison time window (such as the historical time window), there are a total of 20 sentences, the phrase [network, abnormal] appears 10 times, and [unable to, handle, member] appears 10 times; it can be seen that in the comparison time window, the proportion of [network, abnormal] is 50%, and the proportion of [unable to, handle, member] is 50%; in the current time window, the proportion of [network, abnormal] becomes 83.3%, which exceeds the threshold and can be found to be abnormal, but the proportion of "unable to, handle, member" is only 17.7%, and its proportion has become smaller. If it does not exceed the threshold, there will be no abnormal alarm, but the actual change from 10 to 20 is still an emergency and needs to be processed by alarm, but the existing processing method directly ignores this abnormal situation, which is obviously not in line with the actual situation.

[0040] In order to solve the various problems mentioned above, an embodiment of the present application provides a method for detecting emergencies. When using data in the comparison time, the method uses data within the past week, which is more accurate than the prior art that only uses data from the previous day. At the same time, in the process of determining word combinations, the embodiment of the present application uses data within a preset time period before the current moment, which is more efficient and responsive. At the same time, the embodiment of the present application provides different measurement methods for common word combinations and abnormal word combinations, that is, different emergency indexes are used to solve two situations: the proportion of solving conventional problems remains unchanged, and abnormal problems almost do not occur. Finally, the embodiment of the present application also provides a manual emergency index, which can filter out anomalies that can be solved by intelligent customer service in large quantities by itself. This index is a special evaluation method that may be a real anomaly, and has a good effect in intelligent customer service.

[0041] The emergency event detection method provided in the embodiment of the present application can be applied to Figure 1 Server 102 in , therefore, see Figure 2 , Figure 2 First, a structural block diagram of an electronic device is provided for an embodiment of the present application. The electronic device may be Figure 1 Server 102 in.

[0042] like Figure 2 As shown, the electronic device 200 includes a memory 201, a processor 202 and a communication interface 203, and the memory 201, the processor 202 and the communication interface 203 are directly or indirectly electrically connected to each other to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.

[0043] The memory 201 can be used to store software programs and modules, such as the instructions / modules of the emergency detection device 400 provided in the embodiment of the present invention, which can be stored in the memory 201 in the form of software or firmware or solidified in the operating system (OS) of the electronic device 200. The processor 202 executes various functional applications and data processing by executing the software programs and modules stored in the memory 201. The communication interface 203 can be used to communicate signaling or data with other node devices.

[0044] Among them, the memory 201 can be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), etc.

[0045] The processor 202 may be an integrated circuit chip with signal processing capability. The processor 202 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0046] Understandably, Figure 2 The structure shown is for illustration only. The electronic device 200 may also include Figure 2 More or fewer components as shown, or with Figure 2 Different configurations are shown. Figure 2 The components shown may be implemented in hardware, software or a combination thereof.

[0047] See 3, Figure 3 A schematic flow chart of an emergency event detection method provided in an embodiment of the present application, the execution subject of the method may be Figure 2 In the electronic device 200 shown, the method may include the following steps:

[0048] S301, acquiring first text data generated in a current time window and second text data generated in multiple historical time windows, and extracting third text data from the first text data according to a target sub-time window;

[0049] Among them, the target sub-time window is located within the current time window.

[0050] S302, determining a number of target phrases from the third text data, and the number of each target phrase in the first text data, the second text data and the third text data and the number of times of manual processing.

[0051] S303, for each target phrase, determine a burst index set corresponding to the target phrase according to the number of target phrases, the total number of texts corresponding to the first text data, the second text data and the third text data, the phrase type and the number of manual processing times.

[0052] Among them, the burst index set includes the target sub-time window burst index; the conversion manual burst index; the current time window burst index or the quantity ratio burst index;

[0053] S304, if the target burst index in the burst index set is greater than the threshold corresponding to the target burst index, an alarm process is performed on the determined target phrase;

[0054] The target burst index is any one in the burst index set.

[0055] According to the emergency event detection method provided by the embodiment of the present application, firstly, the first text data generated in the current time window, the second text data generated in multiple historical time windows, and the third text data extracted from the first text data are obtained, and then the target phrase determined in the third text data, and the number of the target phrases in these three text data and the number of manual processing times are used. Finally, based on the obtained number and the number of manual processing times, multiple types of emergency indexes corresponding to the phrase can be obtained, and each emergency index is compared with the threshold value. If the emergency index exceeds the threshold value, an alarm is performed. It can be seen that the embodiment of the present application uses the text data generated in the current time window and multiple historical time windows to detect emergencies, avoiding the situation where inaccurate detection occurs due to incomplete information in a single time period, and further reduces the granularity of the detection time window to avoid the hugging phenomenon. In the detection process, a variety of emergency situations are fully considered, and different emergency indexes are used for early warning. The detection result has certain reliability and accuracy.

[0056] The following is a detailed introduction to each of the above steps.

[0057] In step S301, first text data generated in a current time window and second text data generated in multiple historical time windows are obtained, and third text data are extracted from the first text data according to a target sub-time window.

[0058] In an embodiment of the present application, the current time window refers to the time period from the first preset time length before the current moment to the current moment, and the time length of the current time window is equal to the first preset time length, wherein the first preset time length can be determined according to actual detection needs, and can be 1 hour or 1 day, and the embodiment of the present application is not limited, wherein the embodiment of the present application preferably is 1 hour. For abnormal word combinations, this setting can prevent the problem of being overwhelmed when multiple abnormal emergencies occur at the same time.

[0059] In the embodiment of the present application, the historical time window refers to the time period that is several days apart from the current time window and before the current time window. In order to ensure the accuracy of the predicted data, the time length of the historical time window is usually consistent with the time length of the current time window, and each historical time window does not overlap each other. The number of historical time windows can be multiple. In the embodiment of the present application, 7 historical time windows are preferably used. For example, the current time window is 9:00 to 10:00 in the morning, and the 7 noon to 10:00 corresponding to the current time window are sequentially spaced by 1 day, 2 days, ... 7 days, so that compared with the prior art, only the data of the previous day are used for detection, the accuracy is higher.

[0060] In an embodiment of the present application, the target sub-time window refers to the time period from the second preset time length before the current moment to the current moment. The time length of the target sub-time window is equal to the second preset time length, and the second preset time length is less than the first preset time length. For example, the current time window is from 9:00 to 10:00 in the morning, and the second preset time length is 10 minutes. Then the target sub-time window is from 9:50 to 10:00 in the morning. In an embodiment of the present application, the second preset time length of 10 minutes is preferred, which is more efficient and responsive in the process of determining the target phrase.

[0061] In the embodiment of the present application, the text data is various types of question-type data, for example, why can't I pay, the payment failed, I can't buy anything, etc.

[0062] In an optional implementation, the text data corresponding to the moment when the emergency occurs in the first text data, the second text data, and the third text data can be first eliminated to avoid the influence of such abnormal data on subsequent results. The subsequent processing process can be based on the processed first text data, the second text data, and the third text data.

[0063] In step S302, a number of target phrases are determined from the third text data, as well as the number of each target phrase in the first text data, the second text data and the third text data and the number of manual processing times.

[0064] In specific implementation, all words corresponding to each third text data can be obtained through existing word segmentation technology. For example, for the text "Why can't I pay?", the corresponding words can be: how, can't, pay, and so on. Then, the words whose frequency of appearance is greater than or equal to the preset frequency can be counted, and the word combination formed by these words can be used to obtain the target phrase by using the association analysis algorithm.

[0065] In actual scenarios, [Unable to pay] appears. In some cases, some users choose to solve it through intelligent customer service, but in some cases, users choose manual processing. Therefore, the number of times the system switches to manual processing in the embodiment of the present application refers to the number of times the system switches to manual processing after the phrase appears, and this value can be obtained from the system log. By calculating the manual processing burst index by the number of times the system switches to manual processing, it is possible to filter out anomalies that can be solved by intelligent customer service in large quantities by itself for common word combinations. This index is a special evaluation method that has a good effect in intelligent customer service.

[0066] In step S303, for each target phrase, a burst index set corresponding to the target phrase is determined according to the number of target phrases, the total number of texts corresponding to the first text data, the second text data and the third text data, the phrase type and the number of manual processing times.

[0067] In this embodiment, for each target phrase, multiple burst indexes corresponding to the target phrase may be calculated based on the number of the target phrase in the first text data, the second text data, and the number of manual processing times.

[0068] In the embodiment of the present application, the burst index set includes a target sub-time window burst index; a conversion artificial burst index; a current time window burst index or a quantity ratio burst index. Among them, the target sub-time window burst index refers to the test value of the target phrase as a sudden event in the time period from the second preset time before the current moment to the current moment; the current time window burst index refers to the test value of the target phrase as a sudden event in the time period from the first preset time before the current moment to the current moment; the conversion artificial burst index refers to the test value of the artificial handling of sudden events corresponding to the target phrase in the time period from the second preset time before the current moment to the current moment; the quantity ratio burst index is the test value of the abnormal change in the number of target phrases in the time period from the second preset time before the current moment to the current moment.

[0069] In this embodiment, the phrase type of each target phrase is taken into consideration, and different burst metrics are given based on different phrase types. In an optional implementation, the phrase type can be an abnormal phrase type and a non-abnormal phrase type (i.e., an ordinary phrase type). For abnormal phrase types, the embodiment of the present application sets a quantity ratio burst index, which can avoid the problem of underreporting when the quantity changes are small.

[0070] See also Figure 4 , Figure 4 A schematic flowchart of step S303 provided in an embodiment of the present application, wherein step S303 may include the following steps:

[0071] S303-1, determining a manual processing burst index according to the number and quantity of manual processing of the target phrase in the first text data and the second text data respectively.

[0072] In the embodiment of the present application, in the implementation process of determining the transfer to artificial burst index, it is determined based on the text data corresponding to the current time window and the multiple historical windows. In an optional implementation, the above step S301-1 can be as follows Figure 5 As shown, Figure 5 A schematic flow chart of step S303-1 provided in an embodiment of the present application:

[0073] S303-1-1, determining a first proportion value according to the number of manual transfers corresponding to the target phrase in the first text data and the number of the target phrase in the first text data.

[0074] Assuming that the number of times the target phrase appears in the first text data is the number of the target phrase, denoted as m(1), and the number of times it is manually processed is denoted as n(1), then the first proportion value p(1)=n(1) / m(1).

[0075] S303-1-2, determining a second proportion value according to the number of manual transfers corresponding to the target phrase in the second text data and the number of the target phrase in the second text data.

[0076] Assuming that the number of target phrases in the second text data is m(2), and the number of manual processing times is n(2), then the first proportion p(2)=n(2) / m(2).

[0077] It should be noted here that m(2) is the number of target phrases counted in the second text data corresponding to each historical time window. For example, assuming there are 7 historical time windows, then m(2)=m(2,1)+m(2,2)+…+m(2,7), where m(2,1) represents the number of target phrases in the first historical time window; similarly, n(2) can be obtained in a similar way.

[0078] S303-1-3, obtaining a manual burst index according to the first proportion value and the second proportion value.

[0079] In the embodiment of the present application, the ratio between the first proportion value and the second proportion value can be used as the artificial burst conversion index, that is: p(1) / p(2).

[0080] S303 - 2 : Determine a burst index of a target sub-time window according to the number of target phrases in the first text data and the third text data, and the total number of texts in the first text data and the third text data.

[0081] In the embodiment of the present application, in the implementation process of determining the burst index of the target sub-time window, it is determined based on the text data corresponding to the current time window and the target sub-time window. In an optional implementation, the above step S301-2 can be as follows Figure 6 As shown, Figure 6 A schematic flow chart of step S303-2 provided in an embodiment of the present application:

[0082] S303-2-1, determining a third ratio of the number of target phrases in the third text data to the total number of texts in the third text data.

[0083] Continuing to assume that the number of target phrases in the third text data is m(3), and the total number of texts in the third text data is num(3), then the third proportion is: m(3) / num(3).

[0084] S303-2-2, determining a fourth ratio of the number of target phrases in the first text data to the total number of texts in the first text data.

[0085] Assuming that the number of target phrases in the first text data is m(1), and the total number of texts in the third text data is num(1), then the third proportion is: m(1) / num(1).

[0086] S303-2-3, determining a burst index of the current time window according to the third proportion value and the fourth proportion value.

[0087] In the embodiment of the present application, the ratio between the third proportion value and the fourth proportion value can be used as the artificial burst conversion index, that is, [m(3) / num(3)] / [m(1) / num(1)].

[0088] S303-3, determining whether the target phrase is an abnormal phrase according to a preset abnormal phrase table.

[0089] In an embodiment of the present application, an abnormal phrase table can be pre-generated based on historical business operation experience. This table is used to maintain some abnormal words set by the user, such as: [abnormal], [cannot], [unable to], etc., because abnormal words should theoretically not appear or appear very rarely, and have little to do with the number of visitors, while ordinary words will change according to the change in the number of visitors, so the change type can be distinguished based on this.

[0090] In the implementation process, when determining whether a target phrase is an abnormal phrase, as long as at least one word in the target phrase appears in the abnormal word list, the target phrase is considered to be an abnormal phrase.

[0091] S303-4, if not, determine the burst index of the current time window according to the number of target phrases in the first text data and the second text data respectively, and the total number of texts in the first text data and the second text data respectively, and take the artificial burst index, the target sub-time window burst index and the current time window burst index as the burst index set.

[0092] In the embodiment of the present application, if the word combination contains no abnormal words, the word group is a non-abnormal word group, and whether it is an emergency event can be determined by the burst index of the current time window.

[0093] In an optional implementation, determining the burst index of the current time window may be performed according to the following steps:

[0094] a1, determining a fourth ratio of the number of target phrases in the first text data to the total number of texts in the first text data.

[0095] a2, determining a fifth ratio of the number of the target phrases in the second text data to the total number of texts in the second text data.

[0096] a3. Determine the burst index of the current time window according to the fourth proportion value and the fifth proportion value.

[0097] Assuming that the fourth proportion value corresponding to the target phrase is still m(1) / num(1) mentioned above, the number of target phrases in the second text data is recorded as m(2), and the total number of texts in the third text data is recorded as num(2), then the third proportion value is: m(2) / num(2), then the burst index of the current time window can be expressed as: [m(1) / num(1)] / [m(2) / num(2)].

[0098] S303-5, if yes, determine the quantity ratio burst index according to the quantity of the target phrases in the first text data and the second text data respectively, and take the artificial burst index, the target sub-time window burst index and the quantity ratio burst index as the burst index set.

[0099] In the embodiment of the present application, if the word combination contains abnormal words, which are distinguished from ordinary words, the embodiment of the present application will use the quantity burst index to determine whether there is an emergency event.

[0100] In an optional implementation, determining the quantity ratio burst index may be performed according to the following steps:

[0101] b1, according to the number of target phrases in the second text data, obtain the average value and standard deviation of the number of target phrases;

[0102] b2, determining the quantity ratio burst index according to the quantity, average value and standard deviation of the target phrase in the first text data.

[0103] Assume that the number of target phrases in the corresponding second text data in multiple historical time windows is recorded as [last(n), last(n-1), ..., last(1)], where n is the nth historical time window. According to [last(n), last(n-1), ..., last(1)], the mean of the target phrase in the historical time window can be obtained as mean, and the standard deviation is std. The number of target phrases in the first text data is recorded as cur, then the quantity ratio burst index can be expressed as: [(cur-mean) / (cur+std)] and [std / (std+cur)]. It can be seen from this calculation formula that the quantity ratio burst index can represent the probability that the current quantity deviation is larger than the standard deviation. The mean part is retained in the denominator because when the change is the same, if the cardinality is smaller, the change may be stronger. For example, cur=10, mean=5, std=0 is stronger than cur=55, mean=50, std=0.

[0104] The calculation method for determining the number ratio burst pointed out in the embodiment of the present application increases the sorting capability for a small number of cases compared to the n-sigma anomaly detection, and also prevents the problem that std is 0 and cannot be compared. The comparison results are as follows: Figure 7 , Figure 7 The comparison diagram of the effects of the quantity ratio test and the sigma anomaly detection provided in the embodiment of the present application shows that the quantity ratio test can provide better sorting capabilities for other situations (comparison group 1, comparison group 2) without losing the sorting capability (comparison group 3).

[0105] In step S304, if the target burst index in the burst index set is greater than the threshold corresponding to the target burst index, an alarm process is performed on the determined target phrase; wherein the target burst index is any one in the burst index set.

[0106] In the embodiment of the present application, for each target phrase, if any burst index in the burst index set corresponding to it exceeds a preset value, an alarm may be triggered so that the operator, maintenance personnel or supervisory personnel may formulate a response strategy in a timely manner.

[0107] In an embodiment of the present application, alarm sampling can be performed every n minutes, where n minutes can be set dynamically, mainly based on comprehensive considerations such as system performance, user consultation volume, and alarm real-time performance.

[0108] In order to implement the various steps in the above embodiments to achieve the corresponding technical effects, the emergency event detection method provided in the embodiment of the present application can be executed in a hardware device or in the form of a software module. When the emergency event detection method is implemented in the form of a software module, the embodiment of the present application also provides an emergency event detection device, see Figure 8 , Figure 8 This is a functional module diagram of an emergency event detection device provided in an embodiment of the present application. The emergency event detection device 400 may include:

[0109] The acquisition module 410 is used to acquire the first text data generated in the current time window and the second text data generated in multiple historical time windows, and extract the third text data from the first text data according to the target sub-time window; wherein the target sub-time window is located in the current time window;

[0110] A determination module 420 is used to determine a number of target phrases from the third text data, and the number of each target phrase in the first text data, the second text data and the third text data and the number of times of manual processing;

[0111] The determination module 420 is further used to determine, for each target phrase, a burst index set corresponding to the target phrase according to the number of target phrases, the total number of texts corresponding to the first text data, the second text data, and the third text data, the phrase type, and the number of times of manual processing; wherein the burst index set includes a burst index of a target sub-time window; a manual processing burst index; a burst index of a current time window or a burst index of a quantity ratio;

[0112] The detection module 430 is used to perform alarm processing on the determined target phrase if the target burst index in the burst index set is greater than a threshold corresponding to the target burst index; wherein the target burst index is any one in the burst index set.

[0113] In an optional implementation, the determination module 420 is further configured to execute Figure 4 , Figure 5 The various steps in and the various steps in steps a1 to a3 and steps b1 to b2 are executed to achieve the corresponding technical effects.

[0114] In an optional implementation, the emergency event detection device 400 may further include a removal module, which is used to remove text data corresponding to the moment when the emergency event occurs in the first text data, the second text data, and the third text data.

[0115] The embodiment of the present application also provides a readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method for detecting an emergency event in any of the above-mentioned embodiments is implemented. The computer-readable storage medium can be, but is not limited to, a U disk, a mobile hard disk, a ROM, a RAM, a PROM, an EPROM, an EEPROM, a magnetic disk, or an optical disk, and other media that can store program codes.

[0116] The above are only specific embodiments 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 subject to the protection scope of the claims.

Claims

1. A method for detecting an emergency, characterized in that: The method comprises: Acquire first text data generated in a current time window and second text data generated in multiple historical time windows, and extract third text data from the first text data according to a target sub-time window; wherein the target sub-time window is located in the current time window; Determine a plurality of target phrases from the third text data, and the number of each target phrase in the first text data, the second text data, and the third text data and the number of manual processing times; For each of the target phrases, a burst index set corresponding to the target phrase is determined according to the number of the target phrases, the total number of texts corresponding to the first text data, the second text data, and the third text data, the phrase type, and the number of times of switching to manual processing; wherein the burst index set includes a target sub-time window burst index; a manual switching burst index; a current time window burst index or a quantity ratio burst index; If the target burst index in the burst index set is greater than the threshold value corresponding to the target burst index, an alarm is performed on the determined target phrase; wherein the target burst index is any one of the burst index set; For each of the target phrases, determining a burst index set corresponding to the target phrase according to the number of the target phrases, the total number of texts corresponding to the first text data, the second text data, and the third text data, the phrase type, and the number of times of manual processing, including: Determining the manual processing burst index according to the number and quantity of manual processing of the target phrase in the first text data and the second text data respectively; Determine the target sub-time window burst index according to the number of the target phrases in the first text data and the third text data, respectively, and the total number of texts in the first text data and the third text data; Determining whether the target phrase is an abnormal phrase according to a preset abnormal phrase table; If not, determining the current time window burst index according to the number of the target phrases in the first text data and the second text data, and the total number of texts in the first text data and the second text data, and taking the converted artificial burst index, the target sub-time window burst index and the current time window burst index as the burst index set; If so, the quantity ratio burst index is determined according to the quantity of the target phrases in the first text data and the second text data respectively, and the converted artificial burst index, the target sub-time window burst index and the quantity ratio burst index are used as the burst index set.

2. The method for detecting an emergency according to claim 1, characterized in that: Determining the manual processing burst index according to the number and quantity of manual processing of the target phrase in the first text data and the second text data, respectively, includes: Determine a first proportion value according to the number of manual conversions corresponding to the target phrase in the first text data and the number of the target phrase in the first text data; Determine a second proportion value according to the number of manual transfers corresponding to the target phrase in the second text data and the number of the target phrase in the second text data; The artificial burst conversion index is obtained according to the first proportion value and the second proportion value.

3. The method for detecting an emergency according to claim 1, characterized in that: Determining the target sub-time window burst index according to the number of the target phrases in the first text data and the third text data, respectively, and the total number of texts in the first text data and the third text data, respectively, includes: Determine a third ratio of the number of the target phrases in the third text data to the total number of texts in the third text data; Determining a fourth ratio of the number of the target phrases in the first text data to the total number of texts in the first text data; The target sub-time window burst index is determined according to the third proportion value and the fourth proportion value.

4. The method for detecting an emergency according to claim 1, characterized in that: Determining the current time window burst index according to the number of the target phrases in the first text data and the second text data, respectively, and the total number of texts in the first text data and the second text data, respectively, includes: Determining a fourth ratio of the number of the target phrases in the first text data to the total number of texts in the first text data; Determining a fifth ratio of the number of the target phrases in the second text data to the total number of texts in the second text data; The current time window burst index is determined according to the fourth proportion value and the fifth proportion value.

5. The method for detecting an emergency according to claim 1, characterized in that: Determining the quantity ratio burst index according to the quantity of the target phrase in the first text data and the second text data respectively includes: According to the number of the target phrases in the second text data, obtaining an average value and a standard deviation of the number of the target phrases; The quantity ratio burst index is determined according to the quantity of the target phrase in the first text data, the average value and the standard deviation.

6. The method for detecting an emergency according to claim 1, characterized in that: After acquiring first text data generated in a current time window and second text data generated in a plurality of historical time windows, and extracting third text data from the first text data according to a target sub-time window; wherein the target sub-time window is located within the current time window, the method further includes: The text data corresponding to the moment when the emergency occurs in the first text data, the second text data and the third text data are eliminated.

7. An emergency event detection device, characterized in that: include: An acquisition module, used to acquire first text data generated in a current time window, second text data generated in multiple historical time windows, and extract third text data from the first text data according to a target sub-time window; wherein the target sub-time window is located in the current time window; A determination module, used to determine a number of target phrases from the third text data, and the number of each target phrase in the first text data, the second text data and the third text data and the number of manual processing times; The determination module is further used to determine, for each of the target phrases, a burst index set corresponding to the target phrase according to the number of the target phrases, the total number of texts corresponding to the first text data, the second text data, and the third text data, the phrase type, and the number of times of manual processing; wherein the burst index set includes a target sub-time window burst index; a manual processing burst index; a current time window burst index or a quantity ratio burst index; A detection module, configured to perform alarm processing on the determined target phrase if the target burst index in the burst index set is greater than a threshold value corresponding to the target burst index; wherein the target burst index is any one of the burst index set; The determination module is specifically used to determine the conversion to artificial burst index according to the number of conversion to artificial processing and the number of the target phrase in the first text data and the second text data respectively; determine the target sub-time window burst index according to the number of the target phrase in the first text data and the third text data respectively, and the total number of texts of the first text data and the third text data respectively; determine whether the target phrase is an abnormal phrase according to a preset abnormal phrase table; if not, determine the current time window burst index according to the number of the target phrase in the first text data and the second text data respectively, and the total number of texts of the first text data and the second text data respectively, and use the conversion to artificial burst index, the target sub-time window burst index and the current time window burst index as the burst index set; if yes, determine the quantity ratio burst index according to the number of the target phrase in the first text data and the second text data respectively, and use the conversion to artificial burst index, the target sub-time window burst index and the quantity ratio burst index as the burst index set.

8. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program executable by the processor, and the processor can execute the computer program to implement the emergency event detection method according to any one of claims 1 to 6.

9. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the emergency event detection method according to any one of claims 1 to 6 is implemented.

Images