Event processing method and apparatus, computer device, and storage medium

Abstract

The present disclosure discloses a processing method and device of an event, a computer device and a storage medium, and the implementation scheme is as follows: a first vector corresponding to a to-be-processed event and a second vector corresponding to each reference event in a preset event graph are determined; each associated event related to the to-be-processed event is determined from the event graph according to the similarity between the first vector and the second vector; each transition probability between the to-be-processed event and each associated event is determined according to the similarity between the first vector and the second vector corresponding to each associated event; and a type label corresponding to the to-be-processed event is determined according to each transition probability and a label vector corresponding to each associated event. Thus, the type label corresponding to the to-be-processed event is determined by combining the semantic information of the to-be-processed event, each transition probability between the to-be-processed event and each associated event, and the label vector corresponding to each associated event, thereby improving the rationality and reliability of the event handling suggestion and reducing the event processing cost.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to the field of artificial intelligence technology such as natural language processing and deep learning, specifically to an event processing method, apparatus, computer device, and storage medium. Background Technology

[0002] In life, various unexpected events often occur, such as accidents at urban transportation hubs. Without efficient emergency response, the harm caused by these accidents can escalate over time. To adapt to the development of emergency information technology, pre-established contingency plans can greatly reduce losses in the event of potential or possible emergencies.

[0003] In related technologies, contingency plans are typically obtained by predicting the category of an event and then querying a manually pre-defined category-disposal suggestion dictionary. However, manually pre-defined dictionaries and contingency plans are usually cumbersome, making adaptive iteration difficult. Furthermore, existing contingency plan models often suffer from pattern rigidity and rarely utilize the semantic information of the event itself, leading to unreasonable contingency plans that fail to provide optimal disposal suggestions. Summary of the Invention

[0004] This disclosure provides an event processing method, apparatus, computer device, and storage medium.

[0005] According to a first aspect of this disclosure, an event handling method is provided, comprising:

[0006] Determine the first vector corresponding to the event to be processed and the second vector corresponding to each reference event in the preset event graph;

[0007] Based on the similarity between the first vector and each of the second vectors, each associated event related to the event to be processed is determined from the event graph;

[0008] Based on the similarity between the first vector and the second vector corresponding to each of the associated events, the transition probability between the event to be processed and each of the associated events is determined;

[0009] Based on each of the transition probabilities and the label vector corresponding to each of the associated events, the type label corresponding to the event to be processed is determined.

[0010] According to a second aspect of this disclosure, an event processing apparatus is provided, comprising:

[0011] The first determining module is used to determine the first vector corresponding to the event to be processed and the second vector corresponding to each reference event in the preset event graph;

[0012] The second determining module is used to determine, based on the similarity between the first vector and each of the second vectors, each associated event related to the event to be processed from the event graph;

[0013] The third determining module is used to determine the transition probabilities between the event to be processed and each of the associated events based on the similarity between the first vector and the second vector corresponding to each of the associated events.

[0014] The fourth determining module is used to determine the type label corresponding to the event to be processed based on the transition probability and the label vector corresponding to each associated event.

[0015] According to a third aspect of this disclosure, an electronic device is provided, comprising:

[0016] At least one processor; and

[0017] A memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described in the first aspect.

[0019] According to a fourth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing the computer to perform the method as described in the first aspect.

[0020] According to a fifth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method as described in the first aspect.

[0021] The event processing methods, apparatus, devices, and storage media disclosed herein have at least the following beneficial effects:

[0022] First, a first vector corresponding to the event to be processed and a second vector corresponding to each reference event in a pre-defined event graph are determined. Then, based on the similarity between the first vector and each second vector, related events associated with the event to be processed are identified from the event graph. Next, based on the similarity between the first vector and the second vector corresponding to each related event, the transition probabilities between the event to be processed and each related event are determined. Finally, based on the transition probabilities and the label vectors corresponding to each related event, the type label corresponding to the event to be processed is determined. Thus, by combining the semantic information of the event to be processed, the transition probabilities with each related event, and the label vectors corresponding to each related event, the type label corresponding to the event to be processed is automatically determined. This not only improves the rationality and reliability of event handling suggestions but also reduces event processing costs.

[0023] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0024] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0025] Figure 1 This is a schematic flowchart of an event processing method according to an embodiment of the present disclosure;

[0026] Figure 2 This is a schematic flowchart of an event processing method according to another embodiment of the present disclosure;

[0027] Figure 3 This is a flowchart illustrating an event processing method according to yet another embodiment of the present disclosure;

[0028] Figure 4 This is a structural block diagram of an event processing apparatus according to an embodiment of the present disclosure;

[0029] Figure 5 This is a block diagram of an electronic device used to implement the event processing method of the embodiments of this disclosure; Detailed Implementation

[0030] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0031] The event processing method provided in this disclosure can be executed by the event processing device provided in this disclosure or by the electronic device provided in this disclosure. The electronic device may include, but is not limited to, terminal devices such as desktop computers and tablet computers, or servers. The following describes the execution of an event processing method provided in this disclosure by the event processing device provided in this disclosure, without limiting the disclosure. The device is referred to as "device".

[0032] The following is a detailed description of an event handling method provided in this disclosure, with reference to the accompanying drawings.

[0033] Figure 1 This is a flowchart illustrating an event processing method provided in an embodiment of the present disclosure.

[0034] like Figure 1 As shown, the method for handling this event can include the following steps:

[0035] Step 101: Determine the first vector corresponding to the event to be processed and the second vector corresponding to each reference event in the preset event graph.

[0036] In this disclosure, the vector corresponding to the event to be processed can be determined based on any method that can vectorize text. For example, a word set model can be used to determine the vector corresponding to the event to be processed; or, a bag-of-words model can be used to determine the vector corresponding to the event to be processed; or, the bag-of-words model and the Inverse Document Frequency (IDF) value can be used to determine the vector corresponding to the event to be processed; or, the Smooth Inverse Frequency (SIF) value can be used to determine the vector corresponding to the event to be processed, etc. This disclosure does not limit this. The first vector can be the vector corresponding to the event to be processed, and the second vector can be the vector corresponding to each reference event.

[0037] Step 102: Based on the similarity between the first vector and each second vector, identify the related events associated with the event to be processed from the event graph.

[0038] In this disclosure, in order to intuitively display the relationship between various events, an event graph can be constructed, with the event to be processed and each reference event as nodes, and there are lines between the nodes. The device can use the similarity between the event to be processed and each reference event as the weight of the corresponding line.

[0039] Specifically, the device can calculate the similarity between the first vector corresponding to the event to be processed and the second vectors corresponding to each reference event, and then determine the related events of the event to be processed from the event graph based on the similarity.

[0040] As one possible approach, after vectorizing the text, this disclosure can assess their similarity by calculating the cosine of the angle between the first vector and each of the second vectors, without limitation.

[0041] Furthermore, the device can identify reference events whose similarity to the event to be processed is greater than a specified threshold as associated events. By comparing the similarity between the event to be processed and each reference event, the device can determine the associated events related to the event to be processed from the event graph.

[0042] For example, if calculating events similarity It can Events as nodes That is, the event The weights of the edges between these two nodes are used to calculate the similarity between the event to be processed and each reference event using cosine similarity, as shown in the following formula:

[0043]

[0044] in, For reference events The corresponding vector, For reference The corresponding vector,

[0045] vector The corresponding modulus, vector The corresponding module length.

[0046] Understandably, the event to be processed corresponds to a first vector, and each reference event corresponds to a second vector. Therefore, the device can determine the similarity between the event to be processed and each reference event by calculating the similarity between the first vector and each second vector. Then, reference events with a similarity greater than a threshold to the event to be processed are identified as associated events related to the event to be processed, while reference events with a similarity less than a threshold are identified as unrelated events with a weaker relationship to the event to be processed, without any restrictions on this.

[0047] Step 103: Determine the transition probabilities between the event to be processed and each associated event based on the similarity between the first vector and the second vector corresponding to each associated event.

[0048] As we can understand, transition probability is used to characterize the probability of two events transforming into each other. For example, the probability of a pending event transforming into a related event. The higher the transition probability of a pending event transforming into a related event, the easier it is for the pending event to transform into that related event.

[0049] Specifically, the device can first calculate the sum of similarities between the event to be processed and each associated event, and then determine the transition probability from the event to each associated event based on the ratio of the similarity between the event to be processed and each associated event to the sum of the similarities between the event to be processed and each associated event.

[0050] Optionally, after determining the transition probabilities between the event to be processed and each associated event, this disclosure may also update the preset event graph based on the event to be processed and the transition probabilities between the event to be processed and each associated event.

[0051] Step 104: Determine the type label corresponding to the event to be processed based on each transition probability and the label vector corresponding to each associated event.

[0052] In this disclosure, the type label can be a label used to describe the type of the event, such as crowd gathering, traffic congestion, littering, etc., or it can be a label with corresponding handling suggestions for the event, etc. This disclosure does not limit it in this way.

[0053] Optionally, the device can determine the type label vector corresponding to the event to be processed based on each transition probability and the label vector corresponding to each associated event, and then determine the type label corresponding to the event to be processed based on the value of each element in the type label vector and the type label corresponding to each element in the type label vector.

[0054] For ease of explanation, the events to be processed will be denoted as... by Indicates related events A one-hot column vector of labels. If associated events... When the k-th tag is used, the device can determine the associated event under the k-th tag. Record it as 1; otherwise, it can be determined that the associated event is under the k-th label. Record it as 0.

[0055] Furthermore, the device can be based on each transition probability. and the tag vector corresponding to each associated event Identify pending events Corresponding type label vector The formula is as follows:

[0056]

[0057] The device can then calculate the value of each element in the type label vector and use the type labels corresponding to the top g elements with the largest values ​​in the type label vector as the type labels of the events to be processed, without any restrictions.

[0058] It should be noted that, if using type label vectors As a score for the device's handling suggestions for events to be handled, the type label corresponding to the element with the higher handling suggestion score can be used as the type label for the event to be handled. The event to be handled can also have multiple type labels, thus providing the user with several better handling suggestions to choose from.

[0059] In this embodiment, the device first determines a first vector corresponding to the event to be processed and a second vector corresponding to each reference event in a preset event graph. Then, based on the similarity between the first vector and each second vector, it identifies each associated event related to the event to be processed from the event graph. Next, based on the similarity between the first vector and the second vector corresponding to each associated event, it determines each transition probability between the event to be processed and each associated event. Finally, based on each transition probability and the label vector corresponding to each associated event, it determines the type label corresponding to the event to be processed. Thus, by combining the semantic information of the event to be processed, the transition probabilities between it and each associated event, and the label vectors corresponding to each associated event, the type label corresponding to the event to be processed is automatically determined. This not only improves the rationality and reliability of event handling suggestions but also reduces event processing costs.

[0060] Figure 2 This is a flowchart illustrating another event processing method provided in an embodiment of the present disclosure.

[0061] like Figure 2 As shown, the method for handling this event can include the following steps:

[0062] Step 201: Obtain the first description text set, wherein the first description text set includes multiple first description texts, and each first description text corresponds to a historical event.

[0063] The first descriptive text can be the descriptive text corresponding to the historical event, that is, the text to be segmented.

[0064] Step 202: Perform word segmentation on multiple first description texts in the first description text set to obtain the word segmentation set and the frequency of each first word in the word segmentation set appearing in the first description text set. The word segmentation set includes each first word that appears in multiple historical first description texts.

[0065] It should be noted that the device can preset a historical event library, in which there can be multiple historical first description texts. For example, the first description text corresponding to the event one month before the current event to be processed. The scope of historical events is not limited in the present disclosure.

[0066] In the present disclosure, after the device obtains the first description text set, it can determine each first description text in the first description text set, and then perform word segmentation processing on the first description text. Specifically, it can perform word segmentation and stop word removal operations on the first description text. Among them, the word segmentation operation can be to split the continuous first description text to obtain a word sequence, and the stop word removal operation can be to delete the meaningless words in the word sequence, and no limitation is imposed on this.

[0067] For example, if the first description text is "There is a lot of construction waste piled up on the east side of Chengxin Community. Please clear the waste in time.", through the word segmentation operation, the description text can be split into "Chengxin Community", "of", "east side", "piled up", "a lot of", "construction", "waste", "please", "clear in time", "waste". Through the stop word removal operation, the word sequence obtained after the above word segmentation operation can be filtered to obtain "Chengxin Community", "east side", "piled up", "construction", "waste", "clear in time", "waste".

[0068] It should be noted that the above example is only for illustrative purposes, and the present disclosure does not limit the specific operations of the word segmentation processing.

[0069] Specifically, when the device performs word segmentation processing, it uses a word segmentation dictionary. In order to improve the accuracy of the word segmentation processing, the device can insert a geographical entity list into the word segmentation dictionary. The geographical entity list usually includes a list of road names, such as "Fuqiang Road", "Minzhu Road", "Wenming Road", "Hexie Road", etc., or a list of community names, such as "Chengxin Community", "Youshan Community", etc., and no limitation is imposed on this.

[0070] It should be noted that the geographical entity list is not limited to the list of road names and the list of community names, and can also be other Point of Interest (POI) lists, such as a list of building names, and the present disclosure does not limit this here.

[0071] After performing word segmentation processing on the first description text corresponding to each historical event, the device can determine the first description text set corresponding to each first word segmentation. Among them, each first description text set contains the frequency of each first word segmentation appearing in the historical event.

[0072] For example, the first descriptive text set is Q, where the first segmentation word contained in the first descriptive text set can be "timely", "clearance", "garbage", and the frequency of the first segmentation word is [timely - 5 times], [clearance - 4 times], [garbage - 1 time], etc. This disclosure does not impose any restrictions.

[0073] Step 203: Determine the word probability dictionary based on the frequency of each first word appearing in the first description text set.

[0074] Specifically, the device can statistically analyze the frequency of each first word in the first descriptive text set to form a word probability dictionary. The word probability dictionary includes each word and its probability pair corresponding to its occurrence in the event.

[0075] Step 204: Traverse the word probability dictionary to determine the first vector corresponding to the event to be processed and the second vector corresponding to each reference event.

[0076] As one possible implementation, the device can also traverse the word probability dictionary, then determine the word probability of each second word in the second descriptive text corresponding to the event to be processed, and the word probability of each third word in the third descriptive text corresponding to each reference event. Then, based on the word probability of each second word, the word vector of each second word, and the number of historical events, the first vector corresponding to the event to be processed is determined. Finally, based on the word probability of each third word, the word vector of each third word, and the number of historical events, the second vector corresponding to the reference event is determined.

[0077] Understandably, in a word probability dictionary, a word segment with a higher probability of occurrence in a given event indicates a higher probability of its occurrence across various events, and consequently, a smaller impact on each event. Conversely, a word segment with a lower probability of occurrence in a given event indicates a lower probability of its occurrence in that event, and consequently, a larger impact on that event.

[0078] Optionally, the word vectors corresponding to each second word can be weighted and averaged according to the probability of each second word to determine the first vector corresponding to the event to be processed; or, the word vectors corresponding to each second word can be adjusted according to the probability of each second word, and then the adjusted word vectors can be concatenated to determine the first vector corresponding to the event to be processed.

[0079] For ease of explanation, let s denote the number of historical events, where s is not 0, and let the first vector be denoted as... The word vector corresponding to the word segmentation is denoted as Then the first vector It can be calculated in the following ways:

[0080]

[0081] in, For smoothing parameters, For words The probability of appearing in the word segmentation results, where U is the set of segmented words. for The singular vector corresponding to the largest singular value.

[0082] It should be noted that word vectors The second vector can be obtained from open-source word vector files or trained based on an event corpus. There are no restrictions on this. The specific calculation process of the second vector can refer to the specific calculation process of the first vector mentioned above, and will not be repeated here.

[0083] Step 205: Based on the first vector corresponding to the event to be processed and the second vector corresponding to each reference event, determine the related events associated with the event to be processed.

[0084] It should be noted that the specific implementation process of step 205 can be found in the detailed description of the above embodiments, and will not be repeated here.

[0085] Step 206: Determine the sum of similarities between the first vector and each second vector corresponding to each associated event.

[0086] After obtaining the similarity between the first vector and each second vector corresponding to each associated event, the device can sum the similarity scores between the first vector and each second vector. If we consider events... As an event to be processed, As the number of related events of the event to be processed, and k as the number of related events, then we can use... This represents the sum of the similarities between the first vector and each second vector corresponding to each associated event.

[0087] Step 207: Determine the transition probability from the event to be processed to each associated event based on the similarity between the event to be processed and each associated event, and the sum of the similarities.

[0088] Specifically, the device can calculate the event to be processed using the following formula. Transition probability to each associated event:

[0089]

[0090] in, The similarity between the event to be processed and related events. Pending events The sum of similarities with each associated event, For nodes The neighboring nodes, k is the number of associated events.

[0091] It should be noted that the transition probability represents the ease with which a pending event can be redirected to a related event. For example, if... If it is larger, it indicates that the event It is relatively easy to transfer.

[0092] Step 208: Determine the type label corresponding to the event to be processed based on each transition probability and the label vector corresponding to each associated event.

[0093] It should be noted that step 208 can be implemented with reference to the specific implementation of the above embodiments, and this disclosure does not limit it.

[0094] In this embodiment of the present disclosure, the device first acquires a first descriptive text set, which includes multiple first descriptive texts, each corresponding to a historical event. Then, it performs word segmentation on each of the multiple first descriptive texts in the first descriptive text set to obtain a word segmentation set and the frequency of each first word in the word segmentation set appearing in the first descriptive text set. The word segmentation set includes each first word appearing in multiple historical first descriptive texts. Then, based on the frequency of each first word appearing in the first descriptive text set, a word probability dictionary is determined. The word probability dictionary is then traversed to determine a first vector corresponding to the event to be processed and a second vector corresponding to each reference event. Based on the first vector corresponding to the event to be processed and the second vector corresponding to each reference event, the similarity between the first vector and each second vector corresponding to each associated event is determined. Then, based on the similarity between the event to be processed and each associated event, and the sum of the similarities, the transition probability from the event to each associated event is determined. Finally, based on the transition probabilities and the tag vectors corresponding to each associated event, the type tag corresponding to the event to be processed is determined. Therefore, by combining historical events, the vector of the event to be processed is determined, and the type label corresponding to the event to be processed is determined based on the transition probability between the event to be processed and each associated event and the label vector corresponding to each associated event. This makes the determined type label corresponding to the event to be processed more reasonable and provides conditions for providing the best handling suggestions.

[0095] Figure 3 This is a flowchart illustrating yet another event handling method provided in an embodiment of the present disclosure.

[0096] like Figure 3 As shown, the method for handling this event can include the following steps:

[0097] Step 301: Determine the first vector corresponding to the event to be processed and the second vector corresponding to each reference event in the preset event graph.

[0098] Step 302: Based on the first vector corresponding to the event to be processed and the second vector corresponding to each reference event, determine the related events associated with the event to be processed.

[0099] Step 303: Based on the similarity between the first vector and the second vector corresponding to each associated event, determine the transition probabilities between the event to be processed and each associated event.

[0100] It should be noted that steps 301, 302, and 303 described above can be implemented with reference to the specific implementation methods of the above embodiments, and will not be elaborated upon here.

[0101] Step 304: Obtain the event label matrix, where each row vector in the label matrix is ​​a label vector corresponding to a reference event, and each column vector corresponds to a type of label.

[0102] In this disclosure, the event tag matrix can be determined in a variety of ways. The following disclosure will describe one possible implementation method.

[0103] Optionally, the device first acquires multiple historical events, including multiple first historical events with labeled type tags and multiple second historical events without labeled type tags.

[0104] For ease of explanation, this disclosure denotes the number of historical events obtained as Q, the number of first historical events with labeled type as h, and the number of remaining second historical events without labeled type as Qh.

[0105] Then, the device can determine an initial label matrix based on the label type label corresponding to each first historical event and the preset initialization type label. The initial label matrix includes S rows and C columns, where S is the number of multiple historical events and C is the sum of the number of label type labels and the number of initialization type labels.

[0106] The preset initialization type label can be one or more. For second historical events without a labeled type label, the device can preset an initialization type label for it. In the initial label matrix, the number of rows can represent the number of historical events, and the number of columns can represent the sum of the number of labeled type labels and the number of initialization type labels. For example, if the initial label matrix includes 5 rows and 4 columns, it means that the number of historical events is 5, and the sum of the number of labeled type labels and the number of initialization type labels is 4.

[0107] Then, based on the similarity between the multiple third vectors corresponding to multiple historical events, the transition matrix corresponding to multiple historical events can be determined. The transition matrix includes S rows and S columns, and the sum of the elements in each row of the transition matrix is ​​1. Then, based on the transition matrix, the initial label matrix is ​​updated to obtain the first label matrix.

[0108] It should be noted that the transition matrix is ​​denoted as R, which is a matrix with the same number of rows and columns. Each row of the transition matrix corresponds to a historical event, and the sum of the transition probabilities of each type of label in each row is 1. Each historical event corresponds to a third vector. The device can determine the transition probability between the event to be processed and the historical events by calculating the similarity between them.

[0109] In this disclosure, for the initial label matrix The device can propagate labels according to the Label Propagation Algorithm (LPA), specifically, it can propagate labels based on the transition matrix P and the initial label matrix. Calculate the new label matrix L ’ In this disclosure, the new label matrix is ​​denoted as the first label matrix, which is the matrix obtained by updating the initial label matrix, as shown in the following formula:

[0110]

[0111] Furthermore, if the type label represented by the row vector corresponding to each first historical event in the first label matrix is ​​consistent with the annotation type label, and the type label represented by the row vector corresponding to each second historical event label is consistent with the initial type label, then the device determines that the first label matrix is ​​an event label matrix. Correspondingly, if the type label represented by the row vector corresponding to any first historical event in the updated first label matrix is ​​inconsistent with the annotation type label, then the corresponding row vector in the first label matrix is ​​replaced with the annotation type label vector corresponding to any first historical event to generate the second label matrix.

[0112] Furthermore, based on the transition matrix, the device can repeatedly perform the operation of updating the second label matrix until the updated label matrix is ​​the same as the original label matrix.

[0113] It is understandable that by repeatedly updating the label matrix, the updated label matrix can gradually reach the convergence condition, that is, each label in the updated label matrix can accurately represent the label corresponding to each event.

[0114] Alternatively, the device can also determine the updated tag matrix as the event tag matrix after the tag matrix has reached a preset number of updates.

[0115] For example, if the preset number of updates is 5, the device can determine the tag matrix as an event tag matrix after the tag matrix has been updated 5 times, without any limitation.

[0116] Step 305: Determine the tag vector corresponding to each associated event based on the correspondence between each row vector in the event tag matrix and the reference event.

[0117] For ease of explanation, the events to be processed will be denoted as... by Indicates related events A one-hot column vector of labels. If associated events... When the k-th tag is used, the device can determine the associated event under the k-th tag. Record it as 1; otherwise, it can be determined that the associated event is under the k-th label. Record it as 0.

[0118] Step 306: Determine the type label corresponding to the event to be processed based on each transition probability and the label vector corresponding to each associated event.

[0119] It should be noted that step 306 above can be implemented with reference to the specific implementation of the above embodiments, and will not be described in detail here.

[0120] Step 307: Output the type label corresponding to the event to be processed.

[0121] Specifically, after determining the type label corresponding to the event to be processed, the type label corresponding to the event to be processed can be output to the display end of the device for the user to select, without any limitation.

[0122] Step 308: Upon receiving a type label modification instruction, update the type label of the event to be processed according to the type label in the modification instruction.

[0123] It is understandable that when a user selects or enters the type label corresponding to the event to be processed, that is, when the handling suggestion corresponding to the event to be processed is obtained, the type label modification instruction can be triggered. Upon receiving the type label modification instruction, the device can update the type label of the event to be processed according to the type label in the modification instruction, thereby enabling the device to achieve adaptive learning rather than relying on fixed rules, thus improving the flexibility of event processing.

[0124] Step 309: Update the event label matrix according to the type labels corresponding to the updated events to be processed.

[0125] It should be noted that certain update conditions must be met before updating the event tag matrix. The device can determine whether to update the event tag matrix by judging whether the type tag corresponding to the updated event meets the update conditions.

[0126] Specifically, the update conditions can be freely set according to the actual situation. For example, the device can update the event tag matrix at certain time intervals, such as updating it every day, or updating it when the number of pending events reaches a certain number.

[0127] Alternatively, updates can be performed when the event type label is inaccurate or when the number of inaccurate type labels reaches a specified threshold, when the proportion of incorrect labels reaches a preset threshold, or when the difference between the predicted label and the modified label reaches a certain level.

[0128] It should be noted that the above update conditions are merely illustrative of this disclosure and there may be various situations, which are not limited here.

[0129] In this embodiment, the device first determines a first vector corresponding to the event to be processed and a second vector corresponding to each reference event in a preset event graph. Then, based on the first vector and the second vector of each reference event, it determines each associated event related to the event to be processed. Next, based on the similarity between the first vector and the second vector of each associated event, it determines each transition probability between the event to be processed and each associated event. Then, it obtains an event label matrix, where each row vector in the label matrix is ​​a label vector corresponding to a reference event, and each column vector corresponds to a type label. Then, based on the correspondence between each row vector and the event in the event label matrix, it determines the label vector corresponding to each associated event. Next, based on each transition probability and the label vector corresponding to each associated event, it determines the type label corresponding to the event to be processed. Finally, it outputs the type label corresponding to the event to be processed. Upon receiving a type label modification instruction, it updates the type label of the event to be processed according to the type label in the modification instruction, and updates the event label matrix according to the updated type label corresponding to the event to be processed. Thus, it achieves adaptive iterative optimization of the event label matrix, providing a basis and conditions for accurately determining the type label corresponding to the event, thereby improving the reliability of event processing.

[0130] To implement the above embodiments, this disclosure also proposes an event processing apparatus. Figure 4 This is a schematic diagram of the structure of an event processing device provided in an embodiment of the present disclosure.

[0131] like Figure 4 As shown, the event processing device includes: a first determining module 410, a second determining module 420, a third determining module 430, and a fourth determining module 440.

[0132] The first determining module 410 is used to determine the first vector corresponding to the event to be processed and the second vector corresponding to each reference event in the preset event graph.

[0133] The second determining module 420 is used to determine, based on the similarity between the first vector and each of the second vectors, each associated event related to the event to be processed from the event graph;

[0134] The third determining module 430 is used to determine the transition probabilities between the event to be processed and each of the associated events based on the similarity between the first vector and the second vector corresponding to each of the associated events.

[0135] The fourth determining module 440 is used to determine the type label corresponding to the event to be processed based on the transition probability and the label vector corresponding to each associated event.

[0136] Optionally, the first determining module includes:

[0137] The first acquisition unit is used to acquire a first descriptive text set, wherein the first descriptive text set includes a plurality of first descriptive texts, wherein each first descriptive text corresponds to a historical event;

[0138] The second acquisition unit is used to perform word segmentation processing on multiple first description texts in the first description text set to obtain a word segmentation set and the frequency of each first word in the word segmentation set appearing in the first description text set, wherein the word segmentation set includes each first word that appears in the multiple historical first description texts.

[0139] The first determining unit is configured to determine a word probability dictionary based on the frequency of each first word in the first descriptive text set;

[0140] The second determining unit is used to determine, based on the word probability dictionary, the first vector corresponding to the event to be processed and the second vector corresponding to each of the reference events.

[0141] Optionally, the second determining unit is specifically used for:

[0142] The word probability dictionary is traversed to determine the word probability corresponding to each second word in the second description text corresponding to the event to be processed, and the word probability corresponding to each third word in the third description text corresponding to each reference event;

[0143] Based on the word probability corresponding to each second word segment, the word vector corresponding to each second word segment, and the number of historical events, determine the first vector corresponding to the event to be processed;

[0144] The second vector corresponding to the reference event is determined based on the word probability corresponding to each third word, the word vector corresponding to each third word, and the number of historical events.

[0145] Optionally, the second determining module is specifically used for:

[0146] Reference events whose similarity to the event to be processed is greater than a threshold are identified as related events to the event to be processed.

[0147] Optionally, the third determining module is specifically used for:

[0148] Determine the sum of the similarities between the first vector and each of the second vectors corresponding to each of the associated events;

[0149] The transition probability from the event to be processed to each of the associated events is determined based on the similarity between the event to be processed and each of the associated events, and the sum of the similarities.

[0150] Optionally, the fourth determining module includes:

[0151] The third acquisition unit is used to acquire an event tag matrix, wherein each row vector in the tag matrix is ​​a tag vector corresponding to a reference event, and each column vector corresponds to a type of tag;

[0152] The third determining unit is used to determine the tag vector corresponding to each associated event based on the correspondence between each row vector in the tag matrix and the reference event.

[0153] Optionally, the third acquisition unit is specifically used for:

[0154] Acquire multiple historical events, wherein the multiple historical events include multiple first historical events with labeled type tags and multiple second historical events without labeled type tags;

[0155] An initial label matrix is ​​determined based on the label type label corresponding to each first historical event and the preset initialization type label, wherein the initial label matrix includes S rows and C columns, where S is the number of the plurality of historical events, and C is the sum of the number of label type labels and the number of initialization type labels;

[0156] Based on the similarity between the multiple third vectors corresponding to the multiple historical events, a transition matrix corresponding to the multiple historical events is determined, wherein the transition matrix includes S rows and S columns, and the sum of the elements in each row of the transition matrix is ​​1;

[0157] Based on the transition matrix, the initial label matrix is ​​updated to obtain the first label matrix;

[0158] If the type label represented by the row vector corresponding to each first historical event in the first label matrix is ​​consistent with the annotation type label, and the type label represented by the row vector corresponding to each second historical event label is consistent with the initialization type label, then the first label matrix is ​​determined to be the event label matrix.

[0159] Optionally, the third acquisition unit is further configured to:

[0160] If the type label represented by the row vector corresponding to any first historical event in the first label matrix is ​​inconsistent with the annotation type label, the corresponding row vector in the first label matrix is ​​replaced by the annotation type label vector corresponding to any first historical event to generate a second label matrix.

[0161] Based on the transition matrix, the operation of updating the tag matrix is ​​repeated until the updated tag matrix is ​​the same as the original tag matrix, or until a preset number of updates is reached, and the updated tag matrix is ​​determined as the event tag matrix.

[0162] Optionally, the third determining unit includes:

[0163] The fourth determining unit is used to determine the type label vector corresponding to the event to be processed based on each transition probability and the label vector corresponding to each associated event;

[0164] The fifth determining unit is used to determine the type label corresponding to the event to be processed based on the value of each element in the type label vector and the type label corresponding to each element in the type label vector.

[0165] Optionally, the fifth determining unit is further configured to:

[0166] Output the type label corresponding to the event to be processed;

[0167] Upon receiving a type label modification instruction, the type label of the event to be processed is updated according to the type label in the modification instruction;

[0168] Update the event label matrix according to the updated type labels corresponding to the events to be processed.

[0169] Optionally, the third determining module is further configured to:

[0170] The preset event graph is updated based on the events to be processed and the transition probabilities between the events to be processed and each of the associated events.

[0171] In this embodiment, the device first determines a first vector corresponding to the event to be processed and a second vector corresponding to each reference event in a preset event graph. Then, based on the similarity between the first vector and each second vector, it identifies each associated event related to the event to be processed from the event graph. Next, based on the similarity between the first vector and the second vector corresponding to each associated event, it determines each transition probability between the event to be processed and each associated event. Finally, based on each transition probability and the label vector corresponding to each associated event, it determines the type label corresponding to the event to be processed. Thus, by combining the semantic information of the event to be processed, the transition probabilities between it and each associated event, and the label vectors corresponding to each associated event, the type label corresponding to the event to be processed is automatically determined. This not only improves the rationality and reliability of event handling suggestions but also reduces event processing costs.

[0172] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0173] Figure 5 A schematic block diagram of an example electronic device 500 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0174] like Figure 5 As shown, device 500 includes a computing unit 501, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 502 or a computer program loaded from storage unit 508 into random access memory (RAM) 503. RAM 503 may also store various programs and data required for the operation of device 500. The computing unit 501, ROM 502, and RAM 503 are interconnected via bus 504. Input / output (I / O) interface 505 is also connected to bus 504.

[0175] Multiple components in device 500 are connected to I / O interface 505, including: input unit 506, such as keyboard, mouse, etc.; output unit 507, such as various types of monitors, speakers, etc.; storage unit 508, such as disk, optical disk, etc.; and communication unit 509, such as network card, modem, wireless transceiver, etc. Communication unit 509 allows device 500 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0176] The computing unit 501 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above, such as event processing methods. For example, in some embodiments, the event processing methods may be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program may be loaded and / or installed on device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by the computing unit 501, one or more steps of the event processing methods described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform event processing methods by any other suitable means (e.g., by means of firmware).

[0177] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0178] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0179] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0180] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0181] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), the Internet, and blockchain networks.

[0182] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. A server can be a cloud server, also known as a cloud computing server or cloud host, a hosting product within the cloud computing service system that addresses the shortcomings of traditional physical hosts and VPS (Virtual Private Server) services, such as high management difficulty and weak business scalability. Servers can also be servers for distributed systems or servers incorporating blockchain technology.

[0183] In this embodiment, the device first determines a first vector corresponding to the event to be processed and a second vector corresponding to each reference event in a preset event graph. Then, based on the similarity between the first vector and each second vector, it identifies each associated event related to the event to be processed from the event graph. Next, based on the similarity between the first vector and the second vector corresponding to each associated event, it determines each transition probability between the event to be processed and each associated event. Finally, based on each transition probability and the label vector corresponding to each associated event, it determines the type label corresponding to the event to be processed. Thus, by combining the semantic information of the event to be processed, the transition probabilities between it and each associated event, and the label vectors corresponding to each associated event, the type label corresponding to the event to be processed is automatically determined. This not only improves the rationality and reliability of event handling suggestions but also reduces event processing costs.

[0184] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0185] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method of processing an event, characterized by, The method comprises the following steps: acquiring a first description text set, wherein the first description text set comprises a plurality of first description texts, each of which corresponds to a historical event; performing word segmentation processing on the plurality of first description texts respectively to acquire a segmentation set and a frequency of each first segmentation in the segmentation set in the first description text set, wherein the segmentation set comprises each first segmentation appearing in the plurality of historical first description texts; determining a word probability dictionary according to the frequency of each first segmentation in the first description text set; traversing the word probability dictionary to determine a word probability of each second segmentation in a second description text corresponding to a to-be-processed event and a word probability of each third segmentation in a third description text corresponding to each reference event in a preset event graph; determining a first vector corresponding to the to-be-processed event according to the word probability of each second segmentation, a word vector corresponding to each second segmentation and a number of the historical events; and determining a second vector corresponding to each reference event according to the word probability of each third segmentation, a word vector corresponding to each third segmentation and the number of the historical events; determining each associated event related to the to-be-processed event from the event graph according to a similarity between the first vector and each second vector; determining a transition probability between the to-be-processed event and each associated event according to a similarity between the first vector and a second vector corresponding to each associated event respectively; determining a type label corresponding to the to-be-processed event according to each transition probability and a label vector corresponding to each associated event.

2. The method of claim 1, wherein, The determining each associated event related to the to-be-processed event from the event graph according to a similarity between the first vector and each second vector comprises: determining a reference event having a similarity greater than a threshold value with the to-be-processed event as an associated event related to the to-be-processed event.

3. The method of claim 1, wherein, The determining a transition probability between the to-be-processed event and each associated event according to a similarity between the first vector and a second vector corresponding to each associated event respectively comprises: determining a sum of the similarities between the first vector and the second vector corresponding to each associated event; and determining a transition probability from the to-be-processed event to each associated event according to the similarity between the to-be-processed event and each associated event and the sum of the similarities.

4. The method according to any one of claims 1 to 3, characterized in that, Before the determining a type label corresponding to the to-be-processed event according to each transition probability and a label vector corresponding to each associated event, the method further comprises: acquiring an event label matrix, wherein each row vector in the label matrix is a label vector corresponding to a reference event, and each column vector corresponds to a type label; determining a label vector corresponding to each associated event according to a corresponding relationship between each row vector in the label matrix and a reference event.

5. The method of claim 4, wherein, The acquiring an event label matrix comprises: acquiring a plurality of historical events, wherein the plurality of historical events comprise a plurality of first historical events having labeled type labels and a plurality of second historical events having no labeled type labels; According to the annotation type label corresponding to each first historical event and the preset initialization type label, an initial label matrix is determined, wherein the initial label matrix includes S rows and C columns, S is the number of the plurality of historical events, and C is the number of the annotation type label and the number of the initialization type label; According to the similarity between the plurality of third vectors corresponding to the plurality of historical events, a transition matrix corresponding to the plurality of historical events is determined, wherein the transition matrix includes S rows and S columns, and the sum of the elements in each row of the transition matrix is 1; Based on the transition matrix, the initial label matrix is updated to obtain a first label matrix; In a case where the type label represented by the row vector corresponding to each first historical event in the first label matrix is consistent with the annotation type label, and the type label represented by the row vector corresponding to each second historical event label is consistent with the initialization type label, the first label matrix is determined as the event label matrix.

6. The method of claim 5, wherein, After the first label matrix is obtained, the method further includes: In a case where the type label represented by the row vector corresponding to any first historical event in the first label matrix is inconsistent with the annotation type label, the annotation type label vector corresponding to the any first historical event is used to replace the corresponding row vector in the first label matrix to generate a second label matrix; Based on the transition matrix, the operation of updating the second label matrix is repeatedly performed until the updated label matrix is the same as the label matrix before updating, or a preset number of updates is reached, and the updated label matrix is determined as the event label matrix.

7. The method of claim 4, wherein, The determining of the type label corresponding to the to-be-processed event according to each transition probability and the label vector corresponding to each associated event includes: According to each transition probability and the label vector corresponding to each associated event, a type label vector corresponding to the to-be-processed event is determined; According to the value of each element in the type label vector and the type label corresponding to each element in the type label vector, the type label corresponding to the to-be-processed event is determined.

8. The method of claim 4, wherein, After the type label corresponding to the to-be-processed event is determined, the method further includes: Outputting the type label corresponding to the to-be-processed event; In a case where a type label modification instruction is received, updating the type label of the to-be-processed event according to the type label in the modification instruction; Updating the event label matrix according to the updated type label corresponding to the to-be-processed event.

9. The method of any one of claims 1-3, wherein, After the transition probability between the to-be-processed event and each associated event is determined, the method further includes: Updating the preset event graph based on the to-be-processed event and the transition probability between the to-be-processed event and each associated event.

10. An event processing apparatus, characterized by comprising: The method includes: The first determining module is configured to: obtain a first description text set, the first description text set including a plurality of first description texts, each first description text corresponding to a historical event; perform word segmentation processing on the plurality of first description texts respectively to obtain a word segmentation set and a frequency of each first word segmentation in the word segmentation set in the first description text set, the word segmentation set including each first word segmentation appearing in the plurality of historical first description texts; and determine a word probability dictionary according to the frequency of each first word segmentation in the first description text set; The first determining module is configured to: obtain a first description text set, the first description text set including a plurality of first description texts, each first description text corresponding to a historical event; perform word segmentation processing on the plurality of first description texts respectively to obtain a word segmentation set and a frequency of each first word segmentation in the word segmentation set in the first description text set, the word segmentation set including each first word segmentation appearing in the plurality of historical first description texts; and determine a word probability dictionary according to the frequency of each first word segmentation in the first description text set; The second determining module is configured to determine, according to a similarity between the first vector and each second vector, each associated event related to the to-be-processed event from the event graph. The third determining module is configured to determine, according to a similarity between the first vector and each second vector corresponding to each associated event, each transition probability between the to-be-processed event and each associated event. The fourth determining module is configured to determine a type label corresponding to the to-be-processed event according to the transition probability and a label vector corresponding to each associated event.

11. The apparatus of claim 10, wherein, The second determining module is specifically configured to: determine, as an associated event related to the to-be-processed event, a reference event having a similarity greater than a threshold value with the to-be-processed event.

12. The apparatus of claim 10, wherein, The third determining module is specifically configured to: determine a sum of each similarity between the first vector and each second vector corresponding to each associated event; and determine a transition probability from the to-be-processed event to each associated event according to the similarity between the to-be-processed event and each associated event and the sum of the similarities.

13. The apparatus of any one of claims 10-12, wherein, The fourth determining module includes: The third obtaining unit is configured to obtain an event label matrix, wherein each row vector in the label matrix is a label vector corresponding to a reference event, and each column vector corresponds to a type label. The third determining unit is configured to determine a label vector corresponding to each associated event according to a correspondence between each row vector in the label matrix and a reference event.

14. The apparatus of claim 13, wherein, The third obtaining unit is specifically configured to: obtain a plurality of historical events, wherein the plurality of historical events include a plurality of first historical events having labeled type labels and a plurality of second historical events having no labeled type labels. determine an initial label matrix according to the annotation type label corresponding to each of the first historical events and the preset initialization type label, wherein the initial label matrix includes S rows and C columns, S is a number of the plurality of historical events, and C is a number of the annotation type label and a number of the initialization type label; determine a transition matrix corresponding to the plurality of historical events according to similarities between a plurality of third vectors corresponding to the plurality of historical events, wherein the transition matrix includes S rows and S columns, and a sum of elements of each row in the transition matrix is 1; update the initial label matrix based on the transition matrix to obtain a first label matrix; in a case where type labels represented by a row vector corresponding to each of the first historical events in the first label matrix are consistent with the annotation type label, and type labels represented by a row vector corresponding to each of the second historical events in the first label matrix are consistent with the initialization type label, determine the first label matrix as the event label matrix.

15. The apparatus of claim 14, wherein, The third obtaining unit is further configured to: in a case where type labels represented by a row vector corresponding to any of the first historical events in the first label matrix are inconsistent with the annotation type label, replace the corresponding row vector in the first label matrix with an annotation type label vector corresponding to the any of the first historical events to generate a second label matrix; based on the transition matrix, repeatedly perform an operation of updating the second label matrix until an updated label matrix is the same as a label matrix before being updated, or a preset number of updates is reached, and determine the updated label matrix as the event label matrix.

16. The apparatus of claim 13, wherein, The third determining unit includes: a fourth determining unit configured to determine a type label vector corresponding to the to-be-processed event according to each transition probability and a label vector corresponding to each of the associated events; a fifth determining unit configured to determine a type label corresponding to the to-be-processed event according to a value of each element in the type label vector and a type label corresponding to each element in the type label vector.

17. The apparatus of claim 16, wherein, The fifth determining unit is further configured to: output the type label corresponding to the to-be-processed event; in a case where a type label modification instruction is received, update the type label of the to-be-processed event according to a type label in the modification instruction; update the event label matrix according to the updated type label corresponding to the to-be-processed event.

18. The apparatus of any one of claims 10-12, wherein, The third determining module is further configured to: update the preset event graph based on the to-be-processed event and a transition probability between the to-be-processed event and each of the associated events.

19. An electronic device, comprising: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method in any one of claims 1-9.

20. A non-transitory computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to enable the computer to perform the method in any one of claims 1-9.

21. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-9.

Images