Fault work order clustering method and apparatus

Abstract

This application relates to the field of natural language processing and provides a method and apparatus for clustering fault work orders. The method includes: obtaining N adjectives from the fault description text of at least one fault work order in a target system; arranging and combining M entity names and N adjectives in the target system to obtain M*N keyword groups; determining the fault work order vector of a single fault work order based on the word vectors of the words to be processed in the single fault work order; sequentially matching the fault description text of the single fault work order with the M*N keyword groups, and obtaining keyword groups to be clustered based on the successful matching; clustering the word vectors of the keyword groups to be clustered in each fault work order to obtain X classes of keyword vectors; and matching the keyword vectors with the fault work order vectors in each of the X classes, clustering the successfully matched fault work order vectors. The fault work order clustering method provided in this application can improve the accuracy of fault classification, thereby improving the accuracy of fault work order clustering.

Description

Technical Field

[0001] This application relates to the field of natural language processing technology, specifically to a fault work order clustering method and apparatus. Background Technology

[0002] Currently, there are relatively few intelligent methods for handling system failure work orders, but the demand for intelligent analysis and processing of these work orders is increasing daily. Intelligent analysis and processing of system failure work orders can significantly reduce the workload of maintenance personnel and improve work efficiency; it can quickly find solutions to similar system failures and resolve customer issues; and it can identify frequently occurring system failures, allowing for prevention or proactive resolution and avoiding resource waste.

[0003] Most existing clustering algorithms for text data such as work orders are based on TF-IDF (Term Frequency – Inverse Document Frequency) and cosine similarity for direct calculation. This algorithm matches each word after segmentation with the fault work order individually. Since there are different combinations of multiple words, the fault description under each combination may be completely different. Therefore, a single word cannot accurately describe the actual fault situation. Matching a single word with the fault work order will result in a low accuracy rate for fault work order clustering. Summary of the Invention

[0004] This application provides a fault work order clustering method and apparatus to solve the technical problem of low fault work order clustering accuracy caused by matching individual words with fault work orders.

[0005] In a first aspect, embodiments of this application provide a fault work order clustering method, including:

[0006] Obtain N adjectives from the fault description text of at least one fault work order in the target system, where N is an integer greater than or equal to 1;

[0007] By arranging and combining the M entity names and the N adjectives of the target system, M*N keyword groups are obtained, where M is an integer greater than or equal to 1;

[0008] The fault work order vector of a single fault work order is determined based on the word vector of the words to be processed in the single fault work order.

[0009] The fault description text of a single fault work order is matched sequentially with the M*N keyword groups, and the keyword groups to be clustered are obtained based on the successful matching.

[0010] Cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1;

[0011] In each of the X classes, the keyword vector is matched with the fault work order vector, and the successfully matched fault work order vectors are clustered.

[0012] In one embodiment, obtaining N adjectives from the fault description text of at least one fault work order of the target system includes:

[0013] The fault description text is segmented to obtain multiple words. The parts of speech of each word are labeled, and N adjectives are obtained from the multiple words.

[0014] In one embodiment, before determining the fault order vector of a single fault order based on the word vector of the words to be processed in the single fault order, the following steps are included:

[0015] The TF-IDF algorithm is used to weight each word after word segmentation in a single fault work order. Based on the weight of each word, Y words are randomly discarded in the single fault work order, and the remaining words are used as words to be processed in the single fault work order, where Y is an integer greater than or equal to 1.

[0016] In one embodiment, determining the fault order vector of a single fault order based on the word vector of the words to be processed in the single fault order includes:

[0017] Vectorize the words to be processed in a single fault work order to obtain the word vectors of the words to be processed in the single fault work order. Calculate the average value of the word vectors of the words to be processed in the single fault work order to obtain the fault work order vector of the single fault work order.

[0018] In one embodiment, the step of sequentially matching the fault description text of a single fault work order with the M*N keyword groups, and obtaining the keyword groups to be clustered based on the successful matching results, includes:

[0019] The fault description text of a single fault work order is matched sequentially with the M*N keyword groups. If there is only one successfully matched keyword group, the successfully matched keyword group is taken as the keyword group to be clustered. If there is more than one successfully matched keyword group, the first successfully matched keyword group is taken as the keyword group to be clustered.

[0020] In one embodiment, the step of clustering the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors includes:

[0021] The keyword group to be clustered for each fault work order is vectorized to obtain the keyword vector to be clustered. The keyword vector to be clustered is then clustered in parallel to obtain the keyword vector of class X.

[0022] In one embodiment, the permutation and combination of the M entity names and the N adjectives of the target system includes:

[0023] The M entity names and N adjectives are arranged and combined according to the order in which they were acquired.

[0024] In one embodiment, the step of sequentially matching the fault description text of a single fault work order with the M*N keyword groups includes:

[0025] If the number of successfully matched keyword groups is 0, then the fault work order will be treated as a separate category.

[0026] Secondly, embodiments of this application provide a fault work order clustering device, comprising:

[0027] The adjective acquisition module is used to acquire N adjectives from the fault description text of at least one fault work order of the target system, where N is an integer greater than or equal to 1;

[0028] The keyword group acquisition module is used to arrange and combine the M entity names and N adjectives of the target system to obtain M*N keyword groups, where M is an integer greater than or equal to 1;

[0029] The fault work order vector acquisition module is used to determine the fault work order vector of a single fault work order based on the word vector of the words to be processed in the single fault work order.

[0030] Keyword group matching module: used to match the fault description text of a single fault work order with the M*N keyword groups in sequence, and obtain the keyword group to be clustered based on the successful matching.

[0031] The keyword group clustering module is used to cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1;

[0032] The fault work order clustering module is used to match the keyword vector with the fault work order vector in each of the X classes, and cluster the successfully matched fault work order vectors.

[0033] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the steps of the fault work order clustering method described in the first aspect.

[0034] Fourthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the fault work order clustering method described in the first aspect.

[0035] The fault work order clustering method and apparatus provided in this application embodiment combines the entity name corresponding to the target system with the adjectives in the fault work order to form keyword groups, then sequentially matches these keyword groups in the fault work order, clusters the successfully matched keyword groups, and finally clusters the corresponding fault work orders in each cluster. This application embodiment uses keyword groups combining the entity name corresponding to the target system with the adjectives in the fault work order to match fault work orders. Since the combination of "entity name + adjective" (e.g., [memory, increase], [CPU utilization, abnormal], [thread, blocking]) is more accurate in describing fault conditions than single words (e.g., [memory], [abnormal], [thread]), it has a higher classification accuracy in practice. Therefore, the matching method in this application embodiment can improve the accuracy of fault classification, thereby improving the accuracy of fault work order clustering. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is one of the flowcharts illustrating the fault work order clustering method provided in the embodiments of this application;

[0038] Figure 2 This is a comparison chart of the classification accuracy of keyword groups for fault work orders in the fault work order clustering method provided in this application embodiment, and the classification accuracy of other types of word groups for fault work orders.

[0039] Figure 3 This is the second flowchart illustrating the fault work order clustering method provided in the embodiments of this application;

[0040] Figure 4 This is a schematic diagram of the fault work order clustering device provided in the embodiments of this application;

[0041] Figure 5 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] Figure 1 This is one of the flowcharts illustrating the fault work order clustering method provided in this application embodiment. (Refer to...) Figure 1 This application provides a fault work order clustering method, which may include:

[0044] 101. Obtain N adjectives from the fault description text of at least one fault work order in the target system;

[0045] Where N is an integer greater than or equal to 1, that is, at least one adjective is obtained. In this embodiment, the number of fault work orders can be extracted as needed. It can be one or all fault work orders of the target system. The more fault work orders extracted, the more comprehensively and accurately the fault situation of the target system can be reflected.

[0046] 102. By arranging and combining the M entity names and N adjectives of the target system, we can obtain M*N keyword groups;

[0047] Where M is an integer greater than or equal to 1, meaning that the target system has at least one entity name, which can be memory, CPU utilization, or IP address such as 1.1.1.101.

[0048] By arranging and combining M entities and N adjectives, a total of M*N keyword groups are generated, such as: [memory, increase], [CPU utilization, anomaly], [ip1.1.1.101, unreachable], etc. These keyword groups are stored to prepare for subsequent work order grouping.

[0049] 103. Determine the fault work order vector of a single fault work order based on the word vector of the words to be processed in the single fault work order;

[0050] After this step is completed, the corresponding fault work order can be represented by a fault work order vector.

[0051] 104. Match the fault description text of a single fault work order with M*N keyword groups in sequence, and obtain the keyword groups to be clustered based on the successful matching results;

[0052] The fault description text of a single fault work order is matched with each of the M*N keyword groups. The comparison is made to see if any one or more of the M*N keyword groups exist in the fault description text of the fault work order. When a match is successful (that is, if any one or more of the M*N keyword groups exist in the fault description text of the fault work order), the keyword group to be clustered for the corresponding fault work order can be obtained.

[0053] 105. Cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain the keyword vectors of class X;

[0054] Where X is an integer greater than or equal to 1, that is, the word vectors of the keyword groups to be clustered are at least clustered into 1 class. In this embodiment, X can be 10, then the fault work order is equivalent to being divided into 10 buckets, and there is another bucket used to place fault work orders that do not match the keyword group.

[0055] It should be noted that the method for clustering the word vectors of the keyword groups to be clustered for each fault work order is not limited here. In this embodiment, the k-means algorithm is used to cluster the word vectors of the keyword groups to be clustered for each fault work order.

[0056] The k-means clustering algorithm is an iterative clustering analysis algorithm. Its steps are as follows: First, the data is divided into K groups. Then, K objects are randomly selected as initial cluster centers. Next, the distance between each object and each seed cluster center is calculated, and each object is assigned to the nearest cluster center. The cluster centers and the objects assigned to them represent a cluster. Each time a sample is assigned, the cluster centers are recalculated based on the existing objects in the cluster. This process is repeated until a termination condition is met. The termination condition may be that no (or a minimum number) objects are reassigned to different clusters, no (or a minimum number) cluster centers change, or the sum of squared errors reaches a local minimum.

[0057] 106. In each class of X, match the keyword vector with the fault work order vector, and cluster the successfully matched fault work order vectors.

[0058] The execution order of step 103 is not strictly limited, as long as it is completed before this step. In step 103, the fault work order vector for each fault work order has been calculated. Since the keyword vector in each class (i.e., each bucket) of class X is derived from the keyword group that was successfully matched with the fault work order, there is a correspondence between the fault work order vector and the keyword vector. Matching the keyword vector with the fault work order vector, if a match is successful, the fault work order vector corresponding to the keyword vector is found. Clustering the fault work order vectors corresponding to the keyword vectors in each class (i.e., each bucket) completes the clustering of the fault work orders corresponding to the fault work order vectors.

[0059] It should be noted that the method for clustering the fault work order vectors corresponding to the keyword vectors in each class (i.e., each bucket) is not limited here. In this embodiment, a hierarchical clustering algorithm is used to cluster the fault work order vectors corresponding to the keyword vectors in each class (i.e., each bucket).

[0060] Hierarchical clustering algorithms calculate the similarity between nodes using a similarity measure and sort them from high to low similarity, gradually reconnecting the nodes. In this embodiment, the hierarchical clustering algorithm can control the number of clusters by setting a similarity threshold. Assuming that in 10 buckets, the fault work orders corresponding to buckets 1 through 10 are clustered into k1, k2, k3, k4, k5, k6, k7, k8, k9, and k10 respectively, then the final fault work orders are clustered into... Each class.

[0061] In actual work order analysis, under otherwise unchanged conditions, the combination of "entity name + adjective" has a high classification accuracy. Analysis of a large number of historical work order fault causes revealed that most causes contained keyword phrases such as "[memory, increase]", "[CPU usage, anomaly]", and "[thread, blocking]". Compared with other methods, this combination method in this embodiment has a higher accuracy. The comparison results are as follows: Figure 2 As shown in the figure, the classification accuracy of the "entity + adjective" combination (i.e., the "entity name + adjective" combination) is 92%, which is higher than the 85% of "verb + adjective", the 80% of "entity + entity" (i.e., "entity name + entity name") and the 68% of "adjective + adjective".

[0062] The fault ticket clustering method provided in this embodiment combines the entity name corresponding to the target system with the adjectives in the fault ticket to form keyword groups. These keyword groups are then matched sequentially in the fault tickets, and the successfully matched keyword groups are clustered. Finally, the corresponding fault tickets are clustered within each cluster category. This embodiment uses keyword groups combining the entity name corresponding to the target system with the adjectives in the fault ticket to match fault tickets. Because the combination of "entity name + adjective" (e.g., [memory, increase], [CPU usage, anomaly], [thread, blocking]) is more accurate in describing fault conditions than single words (e.g., [memory], [anomaly], [thread]), it has a higher classification accuracy in practice. Therefore, the matching method in this embodiment can improve the accuracy of fault classification, thereby improving the accuracy of fault ticket clustering.

[0063] In one embodiment, obtaining N adjectives from the fault description text of at least one fault work order in the target system can be achieved using the following method:

[0064] The fault description text is segmented into words to obtain multiple words. The parts of speech of each word are labeled, and N adjectives are obtained from the multiple words.

[0065] It should be noted that the method of text segmentation is not limited here. In this embodiment, jieba segmentation is used. First, the domain lexicon is loaded, and then the processed text data is segmented to make the segmentation more accurate. After segmentation, each word is labeled with its part of speech, and words with adjective parts of speech are highlighted. The obtained adjectives include: increase, decrease, abnormal, not passable, long duration, etc.

[0066] This embodiment can easily and quickly obtain the adjectives in the fault description text by segmenting the text into words and tagging its parts of speech.

[0067] In one embodiment, before determining the fault work order vector of a single fault work order based on the word vectors of the words to be processed in the single fault work order, the word frequency-inverse text frequency (TF-IDF) algorithm can be used to weight each word after word segmentation in the single fault work order. Based on the weight of each word, Y words are randomly discarded in the single fault work order, and the remaining words are used as the words to be processed in the single fault work order, where Y is an integer greater than or equal to 1, that is, at least one word must be discarded.

[0068] It should be noted that the method for weighting each word after word segmentation in a single fault work order is not limited here. In this example, the TF-IDF algorithm is used to weight each word after word segmentation in a single fault work order.

[0069] The TF-IDF algorithm is a statistical method used to evaluate the importance of a word to a document within a document set or corpus. A word's importance increases proportionally to its frequency of occurrence in a document, but decreases inversely proportionally to its frequency of occurrence in the corpus. Various weighted forms of TF-IDF are commonly used by search engines as a measure or ranking of the relevance between documents and user queries.

[0070] In this embodiment, if the weight of the j-th word after word segmentation in a single fault work order is w j The probability calculation process for the word being discarded is as follows:

[0071] 1. Standardize the weight of each word after word segmentation in a single fault work order using the following formula, converting it into a value between 0 and 1:

[0072]

[0073] Among them, w j For the weight of the j-th word, z j For w j The standardized value, w, is the set of weights of all words after word segmentation in a single fault work order.

[0074] 2. Calculate the probability of the j-th word being discarded using the following formula:

[0075]

[0076] Where, p j Let ∑z represent the probability that the j-th word is discarded, and let ∑z represent the sum of the weights of each word after word segmentation in a single fault work order.

[0077] From the above formula, we can see that w j The larger z is, the better. j The larger z j The larger p is j The smaller the weight, the greater the weight of each word after word segmentation in a single fault work order, and the lower the probability of it being discarded. Conversely, the smaller the weight of each word after word segmentation in a single fault work order, the higher the probability of it being discarded. In this embodiment, the number of words to be discarded can be controlled by setting a threshold Y. Y can be set to 5, that is, 5 words are discarded in a single fault work order.

[0078] It should be noted that the above discussion does not mean that the discarded words are the five words with the lowest weights, but only that words with lower weights are more likely to be discarded. In fact, since this embodiment uses a random discarding method, it is possible that words with higher weights are discarded while words with lower weights are retained. However, this random method retains the basic principle that words with higher weights are more likely to be retained, while increasing the randomness of discarding words, making the discarding and retention of words more realistic. The word vectors calculated accordingly are more representative.

[0079] This embodiment weights each word in a single fault work order after word segmentation and randomly discards some words according to the weights. This makes it more likely that the more important words in a single fault work order will be saved and used as the basis for subsequent calculation of word vectors and fault work order vectors. Since the retained words are relatively important in the fault work order, the fault work order vector obtained from this will be more representative.

[0080] In one embodiment, the fault order vector of a single fault order can be determined based on the word vectors of the words to be processed in the single fault order as follows:

[0081] Vectorize the words to be processed in a single fault work order to obtain the word vectors of the words to be processed in the single fault work order. Calculate the average value of the word vectors of the words to be processed in the single fault work order to obtain the fault work order vector of the single fault work order.

[0082] It should be noted that the method for vectorizing the words to be processed in a single fault work order is not limited here. In this embodiment, the word2vec model is used to vectorize the words to be processed in a single fault work order. Word2vec can quickly and effectively express a word into a vector form based on a given corpus and an optimized training model.

[0083] This embodiment uses the word2vec model to vectorize the words to be processed in a single fault work order, and calculates the fault work order vector of a single fault work order based on the average value of the word vectors of the words to be processed. This can accurately represent a single fault work order while balancing the words in the single fault work order.

[0084] In one embodiment, the fault description text of a single fault work order is matched sequentially with M*N keyword groups, and the keyword groups to be clustered are classified and explained based on the successful matching.

[0085] Figure 3 This is the second flowchart illustrating the fault work order clustering method provided in this application embodiment. (Refer to...) Figure 3 In one embodiment, the fault description text of a single fault work order is matched sequentially with M*N keyword groups. Based on the successful matching, the keyword groups to be clustered are obtained, which may include:

[0086] 301. Match the fault description text of a single fault work order with M*N keyword groups sequentially;

[0087] 302. If there is only one successfully matched keyword group, then the successfully matched keyword group will be used as the keyword group to be clustered.

[0088] 303. If the number of successfully matched keyword groups is greater than 1, then the first successfully matched keyword group is taken as the keyword group to be clustered.

[0089] For example, if the keyword groups [memory, increase] and [CPU utilization, abnormal] are successfully matched in the fault description text of a single fault work order, then [memory, increase] will be taken as the keyword group to be clustered for that fault work order.

[0090] 304. If the number of successfully matched keyword groups is 0, then the fault work order will be treated as a separate category.

[0091] At this point, the fault situation of the fault work order is not within the scope described by these M*N keyword groups, and its fault situation needs to be handled as a separate category (equivalent to being placed in a separate bucket).

[0092] This embodiment classifies the keywords based on the specific situation of matching M*N keyword groups in the fault description text of a single fault work order, selects the keyword groups to be clustered, and can easily cluster similar keyword groups in the subsequent clustering algorithm.

[0093] In one embodiment, clustering the word vectors of the keyword groups to be clustered in each fault work order to obtain X-class keyword vectors can be achieved through the following method:

[0094] Vectorize the keyword group to be clustered for each fault work order to obtain the keyword vector to be clustered. Cluster the keyword vector to be clustered in parallel to obtain the keyword vector of class X.

[0095] It should be noted that the method for vectorizing the keyword group to be clustered for each fault work order is not limited here. In this embodiment, the word2vec model is used to vectorize the keyword group to be clustered for each fault work order.

[0096] Meanwhile, a parallel approach is adopted to cluster the keyword vectors to be clustered. That is, when clustering the keyword vectors to be clustered, several keyword vectors to be clustered are clustered at the same time, rather than one after another. This parallel approach can greatly reduce the time spent on clustering.

[0097] This embodiment utilizes the word2vec model to vectorize the keyword groups to be clustered for each fault work order, and performs parallel clustering of the keyword vectors to be clustered, which greatly reduces the time spent on clustering. Under the pressure of processing large amounts of data, it can greatly improve the efficiency of word vector clustering, thereby greatly improving the effect of fault work order clustering and reducing the pressure of data processing.

[0098] In one embodiment, the M entity names and N adjectives can be arranged and combined according to the order in which they were acquired.

[0099] In this embodiment, the entity name is obtained from the CMDB (Configuration Management Database) and then saved.

[0100] For example, if memory usage is detected before CPU usage, and elevated usage is detected before abnormal usage, then the permutation and combination order would be [memory, elevated], [memory, abnormal], [CPU usage, elevated], [CPU usage, abnormal]. Subsequently, when matching M*N keyword groups sequentially in the fault description text of a single fault work order, the matching would be performed in the order of [memory, elevated], [memory, abnormal], [CPU usage, elevated], [CPU usage, abnormal].

[0101] In this embodiment, the M entity names and N adjectives are arranged and combined in the order of their acquisition time to set a reasonable order for matching keyword groups in subsequent fault work orders.

[0102] In one embodiment, the fault description text also needs to be cleaned before word segmentation.

[0103] It should be noted that the method of text cleaning is not limited here. In this embodiment, regular expressions are used to clean the fault description text.

[0104] Regular expressions are logical formulas for manipulating strings (including ordinary characters (such as letters from a to z) and special characters (called "metacharacters")). They use predefined specific characters and combinations of these characters to form a "rule string," which expresses a filtering logic for the string. A regular expression is a text pattern that describes one or more strings to match when searching text.

[0105] By using regular expressions, special characters such as punctuation marks and spaces are removed from text data. At the same time, numbers, dates, or long non-Chinese strings in the text are replaced with special words (such as number, date, ordernumber, or jobnumber, etc.).

[0106] The fault work order clustering device provided in the embodiments of this application is described below. The fault work order clustering device described below can be referred to in correspondence with the fault work order clustering method described above.

[0107] Figure 4 A schematic diagram of the fault work order clustering device provided in this application embodiment. (Refer to...) Figure 4 This application provides a fault work order clustering device, which may include:

[0108] The adjective acquisition module 401 is used to acquire N adjectives from the fault description text of at least one fault work order of the target system, where N is an integer greater than or equal to 1;

[0109] The keyword group acquisition module 402 is used to arrange and combine M entity names and N adjectives of the target system to obtain M*N keyword groups, where M is an integer greater than or equal to 1;

[0110] The fault work order vector acquisition module 403 is used to determine the fault work order vector of a single fault work order based on the word vector of the words to be processed in the single fault work order.

[0111] Keyword group matching module 404: It is used to match the fault description text of a single fault work order with M*N keyword groups in sequence, and obtain the keyword groups to be clustered based on the successful matching.

[0112] The keyword group clustering module 405 is used to cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1;

[0113] The fault work order clustering module 406 is used to match the keyword vector with the fault work order vector in each class of X, and to cluster the successfully matched fault work order vectors.

[0114] The fault work order clustering device provided in this embodiment combines the entity name corresponding to the system with the adjectives in the fault work order to form keyword groups. These keyword groups are then matched sequentially in the fault work order. The successfully matched keyword groups are clustered, and finally, the corresponding fault work orders are clustered within each cluster category. This embodiment uses keyword groups combining the entity name corresponding to the system with the adjectives in the fault work order to match fault work orders. Because the combination of "entity name + adjective" (e.g., [memory, increase], [CPU usage, anomaly], [thread, blocking]) is more accurate in describing fault conditions than single words (e.g., [memory], [anomaly], [thread]), it has a higher classification accuracy in practice. Therefore, the matching method in this embodiment can improve the accuracy of fault classification, thereby improving the accuracy of fault work order clustering.

[0115] In one embodiment, the adjective acquisition module 401 is specifically used for:

[0116] The fault description text is segmented into words to obtain multiple words. The parts of speech of each word are labeled, and N adjectives are obtained from the multiple words.

[0117] In one embodiment, a weighted discard module (not shown) is further included for:

[0118] The TF-IDF algorithm is used to weight each word in a single fault work order after word segmentation. Based on the weight of each word, Y words are randomly discarded in a single fault work order, and the remaining words are used as words to be processed in the single fault work order, where Y is an integer greater than or equal to 1.

[0119] In one embodiment, the fault work order vector acquisition module 403 is specifically used for:

[0120] Vectorize the words to be processed in a single fault work order to obtain the word vectors of the words to be processed in the single fault work order. Calculate the average value of the word vectors of the words to be processed in the single fault work order to obtain the fault work order vector of the single fault work order.

[0121] In one embodiment, the keyword group matching module 404 is specifically used for:

[0122] The fault description text of a single fault work order is matched sequentially with M*N keyword groups. If there is only one successfully matched keyword group, it is taken as the keyword group to be clustered. If there is more than one successfully matched keyword group, the first successfully matched keyword group is taken as the keyword group to be clustered.

[0123] In one embodiment, the keyword group clustering module 405 is specifically used for:

[0124] Vectorize the keyword group to be clustered for each fault work order to obtain the keyword vector to be clustered. Cluster the keyword vector to be clustered in parallel to obtain the keyword vector of class X.

[0125] In one embodiment, the keyword group acquisition module 402 is specifically used for:

[0126] Arrange the M entity names and N adjectives in the order in which they were obtained.

[0127] In one embodiment, a keyword group matching failure module (not shown in the figure) is also included, for:

[0128] If the number of successfully matched keyword groups is 0, then the fault ticket will be treated as a separate category.

[0129] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 can call a computer program in the memory 530 to execute the steps of the fault work order clustering method, such as including:

[0130] Obtain N adjectives from the fault description text of at least one fault work order in the target system, where N is an integer greater than or equal to 1;

[0131] By arranging and combining the M entity names and the N adjectives of the target system, M*N keyword groups are obtained, where M is an integer greater than or equal to 1;

[0132] The fault work order vector of a single fault work order is determined based on the word vector of the words to be processed in the single fault work order.

[0133] The fault description text of a single fault work order is matched sequentially with the M*N keyword groups, and the keyword groups to be clustered are obtained based on the successful matching.

[0134] Cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1;

[0135] In each of the X classes, the keyword vector is matched with the fault work order vector, and the successfully matched fault work order vectors are clustered.

[0136] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0137] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the steps of the fault work order clustering method provided in the above embodiments, such as including:

[0138] Obtain N adjectives from the fault description text of at least one fault work order in the target system, where N is an integer greater than or equal to 1;

[0139] By arranging and combining the M entity names and the N adjectives of the target system, M*N keyword groups are obtained, where M is an integer greater than or equal to 1;

[0140] The fault work order vector of a single fault work order is determined based on the word vector of the words to be processed in the single fault work order.

[0141] The fault description text of a single fault work order is matched sequentially with the M*N keyword groups, and the keyword groups to be clustered are obtained based on the successful matching.

[0142] Cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1;

[0143] In each of the X classes, the keyword vector is matched with the fault work order vector, and the successfully matched fault work order vectors are clustered.

[0144] On the other hand, embodiments of this application also provide a processor-readable storage medium storing a computer program for causing a processor to perform the steps of the methods provided in the above embodiments, such as including:

[0145] Obtain N adjectives from the fault description text of at least one fault work order in the target system, where N is an integer greater than or equal to 1;

[0146] By arranging and combining the M entity names and the N adjectives of the target system, M*N keyword groups are obtained, where M is an integer greater than or equal to 1;

[0147] The fault work order vector of a single fault work order is determined based on the word vector of the words to be processed in the single fault work order.

[0148] The fault description text of a single fault work order is matched sequentially with the M*N keyword groups, and the keyword groups to be clustered are obtained based on the successful matching.

[0149] Cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1;

[0150] In each of the X classes, the keyword vector is matched with the fault work order vector, and the successfully matched fault work order vectors are clustered.

[0151] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).

[0152] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0153] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A fault work order clustering method, characterized in that, include: Obtain N adjectives from the fault description text of at least one fault work order in the target system, where N is an integer greater than or equal to 1; By arranging and combining the M entity names and N adjectives of the target system, M*N keyword groups are obtained, including: The M entity names and N adjectives are arranged and combined according to the order in which they were acquired; where M is an integer greater than or equal to 1. Based on the word vectors of the words to be processed in a single fault work order, the fault work order vector of the single fault work order is determined, including: Vectorize the words to be processed in a single fault work order to obtain the word vectors of the words to be processed in the single fault work order, calculate the average value of the word vectors of the words to be processed in the single fault work order, and obtain the fault work order vector of the single fault work order. The fault description text of a single fault work order is matched sequentially with the M*N keyword groups. Based on the successful matching, the keyword groups to be clustered are obtained, including: The fault description text of a single fault work order is matched sequentially with the M*N keyword groups. If the number of successfully matched keyword groups is 1, the successfully matched keyword group is taken as the keyword group to be clustered; if the number of successfully matched keyword groups is greater than 1, the first successfully matched keyword group is taken as the keyword group to be clustered. Cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1; In each of the X classes, the keyword vector is matched with the fault work order vector, and the successfully matched fault work order vectors are clustered.

2. The fault work order clustering method according to claim 1, characterized in that, The step involves obtaining N adjectives from the fault description text of at least one fault work order in the target system, including: The fault description text is segmented to obtain multiple words. The parts of speech of each word are labeled, and N adjectives are obtained from the multiple words.

3. The fault work order clustering method according to claim 2, characterized in that, Before determining the fault order vector of a single fault order based on the word vector of the words to be processed in the single fault order, the following steps are included: The TF-IDF algorithm is used to weight each word after word segmentation in a single fault work order. Based on the weight of each word, Y words are randomly discarded in the single fault work order, and the remaining words are used as words to be processed in the single fault work order, where Y is an integer greater than or equal to 1.

4. The fault work order clustering method according to claim 1, characterized in that, The step of clustering the word vectors of the keyword groups to be clustered in each fault work order to obtain X-class keyword vectors includes: The keyword group to be clustered for each fault work order is vectorized to obtain the keyword vector to be clustered. The keyword vector to be clustered is then clustered in parallel to obtain the keyword vector of class X.

5. The fault work order clustering method according to claim 1, characterized in that, The process of matching the fault description text of a single fault work order with the M*N keyword groups sequentially includes: If the number of successfully matched keyword groups is 0, then the fault work order will be treated as a separate category.

6. A fault work order clustering device, characterized in that, The method for performing fault work order clustering as described in claim 1 includes: The adjective acquisition module is used to acquire N adjectives from the fault description text of at least one fault work order of the target system, where N is an integer greater than or equal to 1; The keyword group acquisition module is used to arrange and combine the M entity names and N adjectives of the target system to obtain M*N keyword groups, where M is an integer greater than or equal to 1; The fault work order vector acquisition module is used to determine the fault work order vector of a single fault work order based on the word vectors of the words to be processed in the single fault work order, including: Vectorize the words to be processed in a single fault work order to obtain the word vectors of the words to be processed in the single fault work order, calculate the average value of the word vectors of the words to be processed in the single fault work order, and obtain the fault work order vector of the single fault work order. Keyword group matching module: used to match the fault description text of a single fault work order with the M*N keyword groups in sequence, and obtain the keyword group to be clustered based on the successful matching. The keyword group clustering module is used to cluster the word vectors of the keyword groups to be clustered in each fault work order to obtain X types of keyword vectors, where X is an integer greater than or equal to 1; The fault work order clustering module is used to match the keyword vector with the fault work order vector in each of the X classes, and cluster the successfully matched fault work order vectors.

7. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the fault work order clustering method according to any one of claims 1 to 5.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the fault work order clustering method according to any one of claims 1 to 5.

Images