Method, apparatus, electronic device and storage medium for determining intention burst information

Abstract

The present invention relates to the technical field of data processing, and provides a method, apparatus, electronic device and storage medium for determining intention burst information. The method includes: obtaining a first intention and a second intention, where the first intention is obtained by performing intention recognition on problems within a to-be-detected period, and the second intention is obtained by performing intention recognition on problems within a historical period, and the historical period is associated with the to-be-detected period; determining a first distribution of the first intention; determining a second distribution of the second intention; if the first distribution is inconsistent with the second distribution, determining the burst information of the first intention, so that a user can judge whether there is an unexpected event within the to-be-detected period according to the burst information. The present invention enables a user to timely discover the burst intention within the to-be-detected period according to the burst information, and then can respond to the burst intention in a targeted manner in a timely manner.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to a method, device, electronic device, and storage medium for determining intent burst information. Background Art

[0002] Intelligent customer service has established a fast and effective technical means based on natural language for communication between enterprises and a large number of users. In intelligent customer service, users ask a large number of questions every day. The existing technology can handle routine questions relatively well, but for sudden problems that occur during a specific time period, for example, users suddenly make a large number of inquiries about special problems in the real world within a specific time, such as system anomalies, policy changes, etc., it is impossible to detect and respond in a timely manner. Summary of the Invention

[0003] The purpose of the present invention is to provide a method, device, electronic device, and storage medium for determining intent burst information, which can timely determine intent burst information, so that users can timely discover and respond to emergencies based on the burst information.

[0004] To achieve the above purpose, the technical solutions adopted in the embodiments of the present invention are as follows:

[0005] In a first aspect, an embodiment of the present invention provides a method for determining intent burst information. The method includes: obtaining a first intent and a second intent, where the first intent is obtained by performing intent recognition on questions during a to-be-detected period, and the second intent is obtained by performing intent recognition on questions during a historical period, and the historical period is associated with the to-be-detected period; determining a first distribution of the first intent; determining a second distribution of the second intent; if the first distribution is inconsistent with the second distribution, determining the burst information of the first intent, so that a user can determine whether there is an emergency during the to-be-detected period based on the burst information.

[0006] Optionally, both the first intent and the second intent include multiple intent types, each of the first intent and each of the second intent respectively corresponds to its own intent probability. The first intent includes a first burst intent with an intent probability greater than or equal to a preset burst threshold, and the second intent includes a second burst intent with an intent probability greater than or equal to the preset burst threshold. The burst information includes an intent burst index. The intent burst index of the first intent includes the intent burst index of the first burst intent of each intent type. The step of determining the burst information of the first intent includes:

[0007] Calculating a first quantity of the first burst intent of each intent type and a second quantity of the second burst intent of each intent type;

[0008] Calculate the first total of all the first intentions and the second total of all the second intentions;

[0009] Determine the intention burst index of the first burst intention of each intention type according to the first total, the second total, the first quantity and the second quantity of each intention type.

[0010] Optionally, the burst information further includes keywords and the word burst index of the keywords. Each of the first intentions corresponds to at least one first question, and each of the second intentions corresponds to at least one second question. The first intention includes a to-be-confirmed intention with an intention probability less than the preset burst threshold. The step of determining the burst information of the first intention further includes:

[0011] Obtain the text data of the first question corresponding to the to-be-confirmed intention;

[0012] Extract the keywords in the text data for characterizing the to-be-confirmed intention;

[0013] Determine the word burst intention of the keyword according to the number of times the keyword appears in the first question, the number of times the keyword appears in the second question, the total number of the first questions and the total number of the second questions.

[0014] Optionally, the method further includes:

[0015] Sort the intention burst index and the word burst index from largest to smallest to obtain a burst index sequence;

[0016] Obtain each element in the burst index sequence in turn;

[0017] If the element is an intention burst index, display the intention burst index and the corresponding burst intention;

[0018] If the element is a word burst index, display the word burst index and the corresponding keyword and to-be-confirmed intention.

[0019] Optionally, the first intention includes multiple intention types, and the step of determining the first distribution of the first intention includes:

[0020] Calculate the quantity of the first intention of each intention type;

[0021] Determine the first distribution of the first intention according to the quantity of the first intention of all intention types.

[0022] Optionally, before the step of determining the burst information of the first intent if the first distribution is inconsistent with the second distribution, so that the user can judge whether there is an emergency in the to-be-detected period according to the burst information, the method further includes:

[0023] Calculate the correlation coefficient between the first distribution and the second distribution;

[0024] If the correlation coefficient is less than or equal to a preset correlation value, it is determined that the first distribution is inconsistent with the second distribution;

[0025] If the correlation coefficient is greater than the preset correlation value, it is determined that the first distribution is consistent with the second distribution.

[0026] Optionally, the method further includes:

[0027] If the first distribution is consistent with the second distribution, it is determined that the first intent does not include burst information, so that the user can determine that there is no emergency in the to-be-detected period.

[0028] In a second aspect, an embodiment of the present invention provides an intent burst information determination device, including: an acquisition module, configured to acquire a first intent and a second intent, where the first intent is obtained by performing intent recognition on a problem in a to-be-detected period, and the second intent is obtained by performing intent recognition on a problem in a historical period, and the historical period is associated with the to-be-detected period; a determination module, configured to determine a first distribution of the first intent; the determination module is further configured to determine a second distribution of the second intent; the determination module is further configured to, if the first distribution is inconsistent with the second distribution, determine the burst information of the first intent, so that the user can judge whether there is an emergency in the to-be-detected period according to the burst information.

[0029] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory; the memory is used to store a program; the processor is configured to, when executing the program, implement the intent burst information determination method as described in the first aspect above.

[0030] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the intent burst information determination method as described in the first aspect above is implemented.

[0031] Compared with the prior art, the method, apparatus, electronic device and storage medium for determining intention burst information provided by the embodiments of the present invention first obtain a first intention and a second intention. The first intention is obtained by performing intention recognition on the problems within the to-be-detected period, and the second intention is obtained by performing intention recognition on the problems within the historical period. The historical period is associated with the to-be-detected period. Then, the first distribution of the first intention and the second distribution of the second intention are determined. If the first distribution is inconsistent with the second distribution, the burst information of the first intention is determined, so that the user can judge whether there is an unexpected event within the to-be-detected period according to the burst information. By judging the consistency of the distributions of the first intention within the to-be-detected period and the second intention within the historical period, and when the two are inconsistent, further determining the burst information of the first intention within the to-be-detected period, the user can timely discover the burst intention within the to-be-detected period according to the burst information, and then can respond to the burst intention in a targeted manner in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and thus should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0033] Figure 1 It is a flowchart example of the method for determining intention burst information provided by the embodiments of the present invention Figure 1 。

[0034] Figure 2 It is a flowchart example of the method for determining intention burst information provided by the embodiments of the present invention Figure 2 。

[0035] Figure 3 It is a flowchart example of the method for determining intention burst information provided by the embodiments of the present invention Figure 3 。

[0036] Figure 4 It is a flowchart example of the method for determining intention burst information provided by the embodiments of the present invention Figure 4 。

[0037] Figure 5 It is a flowchart example of the method for determining intention burst information provided by the embodiments of the present invention Figure 5 。

[0038] Figure 6 It is an example flowchart of the method for determining intention burst information provided by the embodiments of the present invention.

[0039] Figure 7The block diagram of the intention burst information determination device provided by the embodiments of the present invention is shown.

[0040] Figure 8 The block diagram of the electronic device provided by the embodiments of the present invention is shown.

[0041] Icons: 10 - electronic device; 11 - processor; 12 - memory; 13 - bus; 100 - intention burst information determination device; 110 - acquisition module; 120 - determination module; 130 - display module. Detailed implementation manners

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. Components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations.

[0043] Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed present invention, but is merely representative of selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0044] It should be noted that like reference numerals and letters denote like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.

[0045] In the description of the present invention, it should be noted that if terms such as "upper", "lower", "inner", "outer", etc. are used to indicate the orientation or positional relationship, it is based on the orientation or positional relationship shown in the drawings or the orientation or positional relationship in which the product of the present invention is usually placed during use. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation of the present invention.

[0046] In addition, if terms such as "first", "second", etc. are used only for distinguishing descriptions, they cannot be understood as indicating or implying relative importance.

[0047] It should be noted that, without conflict, the features in the embodiments of the present invention may be combined with each other.

[0048] Please refer to Figure 1 , Figure 1 which is a process example of the intention burst information determination method provided by the embodiments of the present inventionFigure 1 , the method includes the following steps:

[0049] Step S100, obtain a first intent and a second intent, where the first intent is obtained by performing intent recognition on the problems during the to-be-detected period, and the second intent is obtained by performing intent recognition on the problems during the historical period, and the historical period is associated with the to-be-detected period.

[0050] In this embodiment, intent recognition can be obtained through a specially trained intent recognition model, or a rule template can be used to directly match through a word list or obtained after parsing based on the rule template. The first intent can be obtained by inputting the problems during the to-be-detected period into the trained intent recognition model, and the second intent can be obtained by inputting the problems during the historical period into the trained intent recognition model.

[0051] In this embodiment, the historical period being associated with the to-be-detected period indicates that the business characteristics within the historical period are similar to those within the to-be-detected period. For example, both the historical period and the to-be-detected period are business peak periods, or both are business idle periods. Depending on the specific business scenario, the historical period and the to-be-detected period can have different dates but the same time interval. For example, the to-be-detected period is 10:00 - 11:00 on the current day, and the historical period is 10:00 - 11:00 every weekday within the past week. The duration of the historical period and the to-be-detected period can be the same and can be set as needed. For example, the duration of both the historical period and the to-be-detected period is set to one hour.

[0052] In this embodiment, for the application scenario of intelligent customer service, intelligent customer service usually needs to handle a large number of problems asked by users. For each problem, intent recognition can be performed on it to obtain the corresponding intent. For example, if the problem is: "The web page fails to open", the recognized intent of this problem can be "Network anomaly", etc.

[0053] In this embodiment, whether it is the historical period or the to-be-detected period, there can be multiple problems therein, and the intents of different problems can be the same or different. Therefore, both the first intent and the second intent can be multiple. For example, there are a total of 10 problems during the to-be-detected period, where the intents of problems 1 to 4 are all "Network anomaly", the intents of problems 5 to 8 are all "Unable to purchase membership", and the intents of problems 9 to 10 are "Unable to use discount", then the first intent is: {"Network anomaly", "Unable to purchase membership", "Unable to use discount"}.

[0054] Step S101, determine the first distribution of the first intent.

[0055] In this embodiment, the first distribution can be determined according to the query time of the questions corresponding to each intention in the first intention and / or the number of each intention in the first intention. Alternatively, a distribution model of the first distribution can be preset in advance, and then the model parameters in the distribution model can be determined according to the first intention.

[0056] Step S102, determine the second distribution of the second intention.

[0057] In this embodiment, the determination method of the second distribution is the same as that of the first distribution.

[0058] Step S103, if the first distribution is inconsistent with the second distribution, determine the burst information of the first intention so that the user can judge whether there is an emergency during the period to be detected according to the burst information.

[0059] In this embodiment, if the first distribution is inconsistent with the second distribution, there may be an emergency during the period to be detected. At this time, the burst information of the first intention can be determined so that the user can more timely and accurately confirm whether there is a real emergency according to the burst information, so as to further take response measures for the emergency. The burst information can, but is not limited to, the burst index representing the burst degree, keywords included in the burst problem, etc.

[0060] The above method provided by the embodiment of the present invention judges the consistency of the distributions of the first intention during the period to be detected and the second intention during the historical period. When the two are inconsistent, further determine the burst information of the first intention during the period to be detected, so that the user can timely discover the burst intention during the period to be detected according to the burst information, and then can respond to the burst intention in a targeted manner in a timely manner.

[0061] In this embodiment, both the first intention and the second intention include multiple intention types. Each first intention and each second intention respectively correspond to their own intention probabilities. The first intention includes a first burst intention with an intention probability greater than or equal to a preset burst threshold, and the second intention includes a second burst intention with an intention probability greater than or equal to a preset burst threshold. The burst information includes an intention burst index. On the Figure 1 basis, the embodiment of the present invention also provides a specific implementation manner for determining the burst information of the first intention. This implementation manner is used to determine the intention burst index of the first burst intention in the first intention. Please refer to Figure 2 , Figure 2 which is a flow example of the intention burst information determination method provided by the embodiment of the present invention Figure 2 Step S103 includes the following sub-steps:

[0062] Sub-step S1031, calculate the first quantity of the first burst intention of each intention type and the second quantity of the second burst intention of each intention type.

[0063] In this embodiment, each intention is obtained after intention recognition for the corresponding problem. When there are multiple problems during the period to be detected, there are multiple first intentions. First, the first intentions of the same intention type can be grouped into one category, and then the first quantity of the first intentions of each intention type is counted. For example, the first intentions include 5: intention Figure 1 , intention Figure 2 , intention Figure 2 , intention Figure 1 , intention Figure 3 , intention Figure 1 is intention type 1, intention Figure 2 is intention type 2, intention Figure 3 is intention type 3, the first quantity of intention Figure 1 is 2, the first quantity of intention Figure 2 is 2, the first quantity of intention Figure 3 is 1. The calculation method of the second quantity is the same as that of the first quantity, which will not be elaborated here.

[0064] Sub-step S1032, calculate the first total of all first intentions and the second total of all second intentions.

[0065] In this embodiment, the first total of the first intentions of all intention types is the sum of the quantities of the first intentions of each type. For example, the first intentions include 5 types: type 1 to type 5, and their quantities are: 3, 4, 5, 4, 3 respectively, then the total is: 3 + 4 + 5 + 4 + 3 = 19. The calculation method of the second total is the same as that of the first total, which will not be elaborated here.

[0066] Sub-step S1033, determine the intention burst index of the first burst intention of each intention type according to the first total, the second total, the first quantity and the second quantity of each intention type.

[0067] In this embodiment, for the first burst intention in the first intentions, when the first burst intention includes multiple intention types, the intention burst index of the first burst intention includes the intention burst indices of the first burst intentions of each intention type. For the intention burst index of the first burst intention of any target intention type in the intention type, as a specific implementation: perform a ratio test or a proportion test on the first total, the second total, the first quantity of the first burst intention of the target intention type, and the second quantity of the second burst intention of the target intention type, and use the obtained test result as the intention burst index of the first burst intention of the target intention type. The larger the intention burst index, the higher the burst degree, and the more attention the user needs to give in time. The calculation methods of the intention burst indices of other types of first burst intentions are similar, which will not be elaborated here.

[0068] In the method provided by the embodiment of the present invention, the suddenness degree is characterized by the intention suddenness index, and the intention suddenness index of the first sudden intention of each intention type is calculated, so that the user can further analyze and process each first sudden intention with focus and pertinence according to the intention suddenness index.

[0069] In this embodiment, when using the pre-trained intention recognition model to recognize the intention of the problem in the period to be detected, the intention recognition model will output the intention probability that the problem belongs to various intention types, and the intention type with the highest intention probability is used as the first intention corresponding to the problem. Since the intention suddenness index is obtained after processing the result of the intention recognition model, and the intention recognition model is trained according to the problems in the historical period that have occurred, for unknown sudden events that have never occurred, their intention probability may not be high, and the corresponding intention suddenness index is also not high. In order to increase the attention to such sudden events and enable them to be noticed by users in time, the embodiment of the present invention is based on Figure 1 For the specific implementation method of processing such to-be-confirmed intentions with intention probability less than the preset sudden threshold, please refer to Figure 3 , Figure 3 is a flow example of the intention sudden information determination method provided by the embodiment of the present invention Figure 3 , step S103 further includes the following sub-steps:

[0070] Sub-step S1034, obtaining the text data of the first problem corresponding to the to-be-confirmed intention.

[0071] In this embodiment, if the intention probability of the first intention is greater than or equal to the preset sudden threshold, then the first intention is a sudden intention; if the intention probability of the first intention is less than the preset sudden threshold, then the first intention is a to-be-confirmed intention. Of course, the preset sudden threshold can be set according to the needs of the actual scenario. Although the intention probability of the to-be-confirmed intention is less than the preset sudden threshold, it may be a sudden intention that appears for the first time. Therefore, in order to facilitate the user to further identify it, it is also necessary to further obtain the keywords of such to-be-confirmed intentions here. The keywords also belong to the sudden information. Through the keywords, the user can analyze whether the first intention corresponding to the keyword is a known intention type or an unknown intention type.

[0072] In this embodiment, since the user can input a text question or a voice question during consultation, when the user asks a question by voice, the voice question needs to be converted into the corresponding text data.

[0073] Sub-step S1035, extracting the keywords for characterizing the to-be-confirmed intention from the text data.

[0074] In this embodiment, as a specific implementation, for any problem, the text data of the problem can be segmented into words to obtain one or more words, and then the proportion of each word in the text data of all problems during the detection period can be used. For example, whether a word is a keyword can be determined according to the number of times each word appears in the text data of all problems during the detection period. If the number of times is greater than the preset number of times, it is determined as a keyword. For example, the preset number of times is 10 times. The word "unable" appears 11 times, the word "buy" appears 15 times, and the word "select" appears 5 times. Then, "unable" and "buy" are keywords.

[0075] Sub-step S1036: Determine the word burst intention of the keyword according to the number of times the keyword appears in the first problem, the number of times the keyword appears in the second problem, the total number of the first problems, and the total number of the second problems.

[0076] In this embodiment, the first problems are all problems corresponding to the first intention, and the second problems are all problems corresponding to the second intention. Specifically, the method for determining the word burst intention of the keyword can be: performing a ratio test or a proportion test on the number of times the keyword appears in the first problem, the number of times the keyword appears in the second problem, the total number of the first problems, and the total number of the second problems, and using the obtained test result as the word burst intention of the keyword.

[0077] It should be noted that sub-steps S1031 to S1033 and sub-steps S1034 to S1036 can exist simultaneously, or there can be only sub-steps S1031 to S1033. The execution order of sub-steps S1034 to S1035 for extracting keywords and sub-steps S1031 to S1033 for determining the burst index can be one after another, or they can be executed simultaneously.

[0078] In this embodiment, in order to enable users to more intuitively obtain the intention burst information, after determining the intention burst information, the embodiment of the present invention also provides a way to display the burst information. Please refer to Figure 4 , Figure 4 which is a flow example of the intention burst information determination method provided by the embodiment of the present invention Figure 4 ,and this method further includes the following steps:

[0079] Step S104: Sort the intention burst index and the word burst index from largest to smallest to obtain a burst index sequence.

[0080] In this embodiment, for example, the intention burst index is e1, e2, the word burst index is e3, e4, and e1 > e3 > e4 > e2. Then, the burst index sequence obtained by sorting from largest to smallest is: {e1, e3, e4, e2}.

[0081] Step S105: Obtain each element in the burst index sequence in turn.

[0082] Step S106: If the element is an intention burst index, display the intention burst index and the corresponding burst intention.

[0083] Step S107: If the element is a word burst index, display the word burst index and the corresponding keyword and intention to be confirmed.

[0084] In this embodiment, for example, e1 is an intention burst index with the intention name a, and e2 and e3 are word burst indexes with the intention names b and c respectively, and e1 > e2 > e3. The keywords of e2 and e3 are: unable - to - buy, unable - to - pay. Then the burst information is displayed as follows:

[0085] Intended Name Sudden Information a e1 b Unable to buy, e2 c Unable to pay, e3

[0086] It can be understood that steps S104 to S107 are executed after step S103.

[0087] In Figure 1 On this basis, the present invention further provides a specific implementation manner for determining the first distribution. Please refer to Figure 5 , Figure 5 which is a flow example of the intention burst information determination method provided by the embodiment of the present invention Figure 5 , and step S101 includes the following sub - steps:

[0088] Sub - step S1011: Calculate the number of the first intentions of each intention type.

[0089] Sub - step S1012: Determine the first distribution of the first intentions according to the number of the first intentions of all intention types.

[0090] In this embodiment, for example, the first intentions include type a and type b. The number of type a is 20, and the number of type b is 80. Then the first distribution is: a, b is a 20:80 distribution.

[0091] It should be noted that the determination method of the second distribution is the same as that of the first distribution, which will not be elaborated here. It can be understood that the intention types included in the first distribution and the second distribution may have the same intention types or different intention types.

[0092] For the above - mentioned method provided by the embodiment of the present invention, according to the intention types and the number of the first intentions corresponding to each intention type, the distribution of the first intentions that is more in line with the actual situation and more accurate during the period to be detected can be determined. Finally, the burst information of the first intentions determined according to the first distribution and the second distribution is also more reasonable.

[0093] An embodiment of the present invention also provides a specific implementation method for determining whether the first distribution is consistent with the second distribution, specifically:

[0094] First, calculate the correlation coefficient between the first distribution and the second distribution.

[0095] In this embodiment, the correlation coefficient can be determined by the chi-square test method.

[0096] Secondly, if the correlation coefficient is less than or equal to the preset correlation value, it is determined that the first distribution and the second distribution are inconsistent.

[0097] Finally, if the correlation coefficient is greater than the preset correlation value, it is determined that the first distribution and the second distribution are consistent.

[0098] In this embodiment, in order to avoid the trouble caused by unnecessary false alarms to users, an embodiment of the present invention also provides a specific implementation method:

[0099] If the first distribution is consistent with the second distribution, it is determined that the first intention does not include sudden information, so that the user can determine that there is no sudden event during the period to be detected.

[0100] In this embodiment, if the first distribution is consistent with the second distribution, it means that the first intention does not include sudden information, that is, there is no sudden event during the period to be detected. At this time, it can be not displayed and no hint is given to the user. Of course, it can also prompt the user that there is no sudden event, so as to prevent the user from mistakenly thinking that no analysis of sudden information has been done during the period to be detected.

[0101] To more clearly and completely illustrate the entire method process, an embodiment of the present invention also provides an example of a specific implementation process. Please refer to Figure 6 , Figure 6 which is an example flowchart of the method for determining intention sudden information provided by the embodiment of the present invention. Figure 6 In, this process includes three stages: a training stage, an identification stage, and a sudden information determination stage.

[0102] The training stage is to train the intention recognition model. The specific processing process is as follows:

[0103] S11: Obtain the questions raised by the user within a preset historical period.

[0104] S12: Annotate the questions to obtain a training set.

[0105] S13: Use the training set to train the preset intention recognition model to obtain a trained intention recognition model. The intention recognition model can be any model that supports classification recognition, such as a bert-based pre-trained model, an xgboost model, etc.

[0106] S14: Save the trained intent recognition model.

[0107] In the recognition stage, the trained intent recognition model is used to recognize the corresponding intent for each user question, and the recognized intent is saved. The specific processing process is as follows:

[0108] S21: Obtain each user question.

[0109] S22: For each user question, use the trained intent recognition model to perform intent recognition to obtain the recognition result.

[0110] S23: If the maximum value of the intent probability in the recognition result is greater than or equal to the preset threshold, determine the intent corresponding to the intent probability as the sudden intent.

[0111] S24: If the maximum value of the intent probability in the recognition result is less than the preset threshold, determine the intent corresponding to the intent probability as the intent to be confirmed.

[0112] S25: Save each user question and the corresponding intent.

[0113] In the sudden information determination stage, the intent sudden information is determined according to each user question and the corresponding intent obtained in the recognition stage. The specific processing process is as follows:

[0114] S31: Obtain the first intent and the second intent of the questions in the period to be detected and the historical period.

[0115] S32: Calculate the total number of the first intent and the total number of the second intent respectively.

[0116] S33: Use the correlation test to calculate the first distribution of the first intent and the second distribution of the second intent respectively.

[0117] S34: If the first distribution and the second distribution are consistent, determine that there is no sudden event and the process ends normally.

[0118] S35: If the first distribution and the second distribution are inconsistent, calculate the intent sudden index of the first sudden intent in the first intent.

[0119] S36: If the first distribution and the second distribution are inconsistent, determine the keywords of the intent to be confirmed in the first intent and the corresponding word sudden index.

[0120] S37: Display the sudden index and the corresponding keywords in descending order. The sudden index includes the intent sudden index and the word sudden index. For the intent sudden index, display the sudden intent in the corresponding first intent. For the word sudden index, display the keywords corresponding to the intent to be confirmed in the corresponding first intent and the word sudden index.

[0121] To execute the corresponding steps in the above embodiments and each possible implementation manner, the following presents an implementation manner of an intention burst information determination device 100. Please refer to Figure 7 , Figure 7 FIG. Figure 7 shows a block diagram of an intention burst information determination device 100 provided by an embodiment of the present invention. It should be noted that the basic principle and the technical effects generated by the intention burst information determination device 100 provided in this embodiment are the same as those in the above embodiments. For the sake of brief description, some parts of this embodiment are not mentioned.

[0122] The intention burst information determination device 100 includes an acquisition module 110, a determination module 120, and a display module 130.

[0123] The acquisition module 110 is configured to acquire a first intention and a second intention, where the first intention is obtained by performing intention recognition on problems during a to-be-detected period, and the second intention is obtained by performing intention recognition on problems during a historical period, and the historical period is associated with the to-be-detected period.

[0124] The determination module 120 is configured to determine a first distribution of the first intention.

[0125] The determination module 120 is further configured to determine a second distribution of the second intention.

[0126] The determination module 120 is further configured to, if the first distribution is inconsistent with the second distribution, determine the burst information of the first intention, so that a user can determine whether there is an unexpected event during the to-be-detected period according to the burst information.

[0127] Optionally, both the first intention and the second intention include multiple intention types, each first intention and each second intention respectively correspond to their own intention probabilities. The first intention includes a first burst intention with an intention probability greater than or equal to a preset burst threshold, and the second intention includes a second burst intention with an intention probability greater than or equal to a preset burst threshold. The burst information includes an intention burst index. The intention burst index of the first intention includes the intention burst index of each first burst intention of each intention type. The determination module 120 is specifically configured to: calculate a first quantity of each first burst intention of each intention type and a second quantity of each second burst intention of each intention type; calculate a first total of all first intentions and a second total of all second intentions; determine the intention burst index of each first burst intention of each intention type according to the first total, the second total, the first quantity, and the second quantity of each intention type.

[0128] Optionally, the burst information further includes keywords and the word burst index of the keywords. Each first intention corresponds to at least one first question, and each second intention corresponds to at least one second question. The first intention includes a to-be-confirmed intention with an intention probability less than a preset burst threshold. Specifically, the determination module 120 is further configured to: obtain the text data of the first question corresponding to the to-be-confirmed intention; extract the keywords used to characterize the to-be-confirmed intention from the text data; and determine the word burst intention of the keywords according to the number of times the keywords appear in the first question, the number of times the keywords appear in the second question, the total number of the first questions, and the total number of the second questions.

[0129] Optionally, the first intention includes multiple intention types. Specifically, the determination module 120 is configured to: calculate the number of the first intentions of each intention type; and determine the first distribution of the first intentions according to the number of the first intentions of all intention types.

[0130] Optionally, before determining the burst information of the first intention when the first distribution is inconsistent with the second distribution, so that the user can judge whether there is an emergency during the to-be-detected period according to the burst information, the determination module 120 is further configured to: calculate the correlation coefficient between the first distribution and the second distribution; if the correlation coefficient is less than or equal to a preset correlation value, determine that the first distribution is inconsistent with the second distribution; if the correlation coefficient is greater than the preset correlation value, determine that the first distribution is consistent with the second distribution.

[0131] Optionally, the determination module 120 is further configured to: if the first distribution is consistent with the second distribution, determine that the first intention does not include burst information, so that the user determines that there is no emergency during the to-be-detected period.

[0132] The display module 130 is configured to: sort the intention burst index and the word burst index from large to small to obtain a burst index sequence; sequentially obtain each element in the burst index sequence; if the element is an intention burst index, display the intention burst index and the corresponding burst intention; if the element is a word burst index, display the word burst index and the corresponding keyword and the to-be-confirmed intention.

[0133] The embodiment of the present invention further provides a block diagram of the electronic device 10. The method in the foregoing embodiment is applied to the electronic device 10. Please refer to Figure 8 , Figure 8 which shows a block diagram of the electronic device 10 provided by the embodiment of the present invention. The electronic device 10 includes a processor 11, a memory 12, and a bus 13, and the processor 11 and the memory 12 are connected through the bus 13.

[0134] Processor 11 can be an integrated circuit chip with signal processing capabilities. During implementation, each step of the method for determining intentional burst information can be completed by hardware integrated logic circuits or software instructions in processor 11. The aforementioned processor 11 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can 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.

[0135] The memory 12 is used to store programs, such as Figure 7 The apparatus for determining intended sudden information includes at least one software functional module that can be stored in the form of software or firmware in the memory 12 or embedded in the operating system (OS) of the electronic device 10. Upon receiving the execution instruction, the processor 11 executes the program to implement the method for determining intended sudden information disclosed in the above embodiment.

[0136] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for determining the intended burst information as described above is implemented.

[0137] In summary, the embodiments of the present invention provide a method, device, electronic device, and storage medium for determining intentional emergency information, the method comprising: obtaining a first intention and a second intention, wherein the first intention is obtained by performing intention identification on a problem within a time period to be detected, and the second intention is obtained by performing intention identification on a problem within a historical time period, the historical time period being associated with the time period to be detected; determining a first distribution of the first intention; determining a second distribution of the second intention; if the first distribution is inconsistent with the second distribution, determining emergency information of the first intention, so that the user can judge whether there is an emergency event within the time period to be detected based on the emergency information. Compared with the prior art, the embodiments of the present invention judge the consistency of the distribution of the first intention within the time period to be detected and the second intention within the historical time period, and when the two are inconsistent, further determining the emergency information of the first intention within the time period to be detected, thereby enabling the user to promptly discover the emergency intention within the time period to be detected based on the emergency information, and then to promptly respond to the emergency intention in a targeted manner.

[0138] As described above, it is only the specific implementation manner 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 those skilled in the art within the technical scope disclosed by the present invention should be covered within 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 described above.

Claims

1. A method for determining intention burst information, characterized in that, The method includes: Obtaining a first intention and a second intention, where the first intention is obtained by performing intention recognition on problems within a to-be-detected period, the second intention is obtained by performing intention recognition on problems within a historical period, and the historical period is associated with the to-be-detected period; Determining a first distribution of the first intention; Determining a second distribution of the second intention; If the first distribution is inconsistent with the second distribution, determining burst information of the first intention so that a user can judge whether there is an unexpected event within the to-be-detected period according to the burst information. The first intention includes a first burst intention with an intention probability greater than or equal to a preset burst threshold and a to-be-confirmed intention with an intention probability less than the preset burst threshold. The burst information includes an intention burst index of the first burst intention and keywords of the to-be-confirmed intention and a word burst index of the keywords.

2. The method for determining the intention burst information according to claim 1, wherein Both the first intention and the second intention include multiple intention types, and each first intention and each second intention respectively correspond to their own intention probabilities. The second intention includes a second burst intention with an intention probability greater than or equal to the preset burst threshold. The intention burst index of the first intention includes the intention burst index of the first burst intention of each intention type. The step of determining the burst information of the first intention includes: Calculating a first quantity of the first burst intention of each intention type and a second quantity of the second burst intention of each intention type; Calculating a first total of all the first intentions and a second total of all the second intentions; Determining the intention burst index of the first burst intention of each intention type according to the first total, the second total, the first quantity and the second quantity of each intention type.

3. The method for determining the intention burst information according to claim 2, wherein, Each first intention corresponds to at least one first problem, and each second intention corresponds to at least one second problem. The step of determining the burst information of the first intention further includes: Obtaining text data of the first problems corresponding to the to-be-confirmed intention; Extracting keywords for characterizing the to-be-confirmed intention from the text data; Determining a word burst intention of the keywords according to the number of times the keywords appear in the first problems, the number of times the keywords appear in the second problems, the total number of the first problems and the total number of the second problems.

4. The method for determining the intention burst information according to claim 3, characterized in that The method further includes: Sorting the intention burst index and the word burst index from largest to smallest to obtain a burst index sequence; Sequentially obtaining each element in the burst index sequence; If the element is an intention burst index, displaying the intention burst index and the corresponding burst intention; If the element is a word burst index, displaying the word burst index and the corresponding keywords and the to-be-confirmed intention.

5. The method for determining intention burst information according to claim 1, wherein The first intention includes multiple intention types. The step of determining the first distribution of the first intention includes: Calculating the quantity of the first intention of each intention type; Determining the first distribution of the first intention according to the quantities of the first intention of all intention types.

6. The method for determining the intention burst information according to claim 1, wherein Before the step of determining the burst information of the first intention if the first distribution is inconsistent with the second distribution, so that the user can judge whether there is an emergency in the to-be-detected period according to the burst information, the method further includes: Calculating a correlation coefficient between the first distribution and the second distribution; If the correlation coefficient is less than or equal to a preset correlation value, it is determined that the first distribution is inconsistent with the second distribution; If the correlation coefficient is greater than the preset correlation value, it is determined that the first distribution is consistent with the second distribution.

7. The method for determining the intention burst information according to claim 1, wherein The method further includes: If the first distribution is consistent with the second distribution, it is determined that the first intention does not include burst information, so that the user can determine that there is no emergency in the to-be-detected period.

8. An apparatus for determining intention burst information, characterized in that The device includes: An acquisition module, configured to acquire a first intention and a second intention, where the first intention is obtained by performing intention recognition on a problem in a to-be-detected period, the second intention is obtained by performing intention recognition on a problem in a historical period, and the historical period is associated with the to-be-detected period; A determination module, configured to determine a first distribution of the first intention; The determination module is further configured to determine a second distribution of the second intention; The determination module is further configured to, if the first distribution is inconsistent with the second distribution, determine the burst information of the first intention, so that the user can judge whether there is an emergency in the to-be-detected period according to the burst information. The first intention includes a first burst intention with an intention probability greater than or equal to a preset burst threshold and a to-be-confirmed intention with an intention probability less than the preset burst threshold. The burst information includes the intention burst index of the first burst intention, the keyword of the to-be-confirmed intention, and the word burst index of the keyword.

9. An electronic device, characterized in that, Including a processor and a memory; the memory is used to store a program; the processor is used to implement the intention burst information determination method according to any one of claims 1-7 when executing the program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the intention burst information determination method according to any one of claims 1-7.

Images