CN111950573A - Method and device for clustering abnormal problems - Google Patents

Abstract

The present disclosure relates to a method and device for clustering abnormal problems, which are used to improve the clustering effect of abnormal stacks and facilitate more accurate discovery of abnormal problems. The method includes: extracting exception description information from a plurality of exception stacks respectively; calculating similarity of exception description information between exception stacks and exception stacks to obtain similarity; and clustering exception stacks according to the similarity , and determine the class center.

Description

Method and device for clustering abnormal problems

technical field

The present disclosure relates to the fields of communication and computer processing, and in particular, to a method and apparatus for clustering abnormal problems.

Background technique

In the related art, a large number of abnormal problems may occur in the mobile terminal system every day. Information about these exceptions is stored in special stacks called exception stacks. The exception stack is clustered to classify the abnormal problem, which is helpful for the follow-up analysis and other processing of the abnormal problem. Therefore, how to perform more effective clustering processing is a problem that has been researched in the industry.

SUMMARY OF THE INVENTION

In order to overcome the problems existing in the related art, the present disclosure provides a method and apparatus for clustering abnormal problems.

According to a first aspect of the embodiments of the present disclosure, there is provided a method for clustering abnormal problems, including:

Extract the exception description information from multiple exception stacks respectively;

Calculate the similarity of the exception description information between the exception stack and the exception stack to obtain the similarity;

The abnormal stack is clustered according to the similarity, and the cluster center is determined.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: In this embodiment, the exception stack is clustered according to the exception description information in the exception stack. Since the exception description information is more accurate in describing the abnormal problem, the clustering effect Also better.

In one embodiment, the exception description information includes at least one of the following: exception language description information, exception flags, exception values, exception code identifiers, library information and class information of functions related to exceptions.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: the embodiments provide various abnormal description information, can perform clustering processing more effectively, and are suitable for various application scenarios.

In one embodiment, calculating the similarity of the exception description information between the exception stack and the exception to obtain the similarity, including:

The edit distance algorithm is used to calculate the similarity of the abnormal language description information between the exception stack and the exception, the library information and class information of the functions related to the exception, and obtain the corresponding sub-similarity of each exception description information;

The exact matching algorithm is used to calculate the similarity between the exception stack and the exception label, exception value, and exception code identifier respectively, and obtain the corresponding sub-similarity of each exception description information.

According to the sub-similarity of each exception description information, the similarity of the exception stack is obtained.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: this embodiment adopts corresponding similarity calculation methods for different abnormal description information, and the obtained similarity is more accurate, which helps to improve the accuracy of subsequent clustering.

In one embodiment, the edit distance algorithm is used to calculate the similarity of the library information and class information of the exception-related functions between the exception stack and the exception, respectively, to obtain the corresponding sub-similarity of each item of exception description information, include:

For each layer in the library information and the class information, calculate the sub-similarity respectively;

Average multiple sub-similarities of the same level;

A weighted sum is performed on the sub-similarities of multiple levels, wherein the sub-similarity with a lower stack level corresponds to a larger weight.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: this embodiment provides a more detailed similarity calculation method for the hierarchical characteristics of the library information and class information of functions related to exceptions, and the obtained similarity is more accurate .

In one embodiment, before calculating the similarity of the exception description information between the exception stack and the exception stack, and obtaining the similarity, the method further includes:

Uniquely encode the exception description information in each exception stack to obtain a unique encoding;

The exception stack is deduplicated according to the unique code.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: this embodiment performs de-duplication processing in advance, which can reduce the calculation process of similarity, and is more efficient.

In one embodiment, the similarity of the exception description information between the exception stack and the exception stack is calculated to obtain the similarity, including:

Perform vector encoding on the exception description information in each exception stack to obtain the vector encoding result;

Calculate the similarity of the vector encoding results between the exception stack and the exception stack to obtain the similarity.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: this embodiment performs vector encoding on the abnormality description information, which facilitates faster calculation of similarity.

In one embodiment, performing clustering processing on the exception stack according to the similarity, and determining the cluster center, includes:

Count the number of the same exception stack;

In order of the number from high to low, cluster processing is performed on the abnormal stacks according to the similarity, and the cluster center is determined.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: This embodiment performs clustering processing according to the descending order of the number of identical exception stacks, so that categories with higher coverage rates can be obtained more quickly.

In one embodiment, according to the order of the number from high to low, clustering processing is performed on the exception stacks according to the similarity, and the cluster center is determined, including:

The obtained similarity is stored in a preset storage structure; the storage structure includes: a two-layer hash map or an array;

The storage structures are traversed in descending order of the number, and cluster processing is performed, and the cluster center is determined.

The technical solutions provided by the embodiments of the present disclosure may include the following beneficial effects: In this embodiment, the similarity is stored and traversed through a preset storage structure, thereby improving the processing efficiency of clustering.

According to a second aspect of the embodiments of the present disclosure, there is provided an apparatus for clustering abnormal problems, including:

The extraction module is used to extract the exception description information from multiple exception stacks respectively;

The calculation module is used to calculate the similarity of the exception description information between the exception stack and the exception stack, and obtain the similarity;

The clustering module is used for clustering the abnormal stack according to the similarity, and determining the cluster center.

In one embodiment, the exception description information includes at least one of the following: exception language description information, exception flags, exception values, exception code identifiers, library information and class information of functions related to exceptions.

In one embodiment, the computing module includes:

The first calculation submodule is used to calculate the similarity of the abnormal language description information between the exception stack and the exception, the library information and class information of the function related to the exception, and obtain the corresponding abnormal description information for each item by using the edit distance algorithm. the sub-similarity of ;

The second calculation sub-module is used to calculate the similarity between the exception stack and the exception label, exception value, and exception code identifier by using the exact matching algorithm, and obtain the corresponding sub-similarity of each item of exception description information.

The statistical sub-module is used to obtain the similarity of the exception stack according to the sub-similarity of each exception description information.

In one embodiment, the first calculation sub-module calculates the sub-similarity separately for each layer in the library information and the class information; averages multiple sub-similarities at the same level; The similarity is weighted and summed, wherein the sub-similarity with a lower stack level corresponds to a greater weight.

In one embodiment, the apparatus further includes:

The unique encoding module is used to uniquely encode the exception description information in each exception stack to obtain a unique encoding;

A deduplication module, configured to perform deduplication processing on the exception stack according to the unique code.

In one embodiment, the computing module includes:

The vector coding submodule is used for vector coding the exception description information in each exception stack to obtain the vector coding result;

The third calculation sub-module is used to calculate the similarity between the exception stack and the exception stack, the result of vector encoding, to obtain the similarity.

In one embodiment, the clustering module includes:

The number sub-module is used to count the number of the same exception stack;

The clustering sub-module is used for clustering the abnormal stacks according to the similarity according to the order of the number from high to low, and determining the cluster center.

In one embodiment, the clustering sub-module stores the obtained similarity in a preset storage structure; the storage structure includes: a two-layer hash map or an array; sequence, traverse the storage structure, perform clustering processing, and determine the cluster center.

According to a third aspect of the embodiments of the present disclosure, an apparatus for clustering abnormal problems is provided, including:

processor;

memory for storing processor-executable instructions;

wherein the processor is configured to:

Extract the exception description information from multiple exception stacks respectively;

Calculate the similarity of the exception description information between the exception stack and the exception stack to obtain the similarity;

The abnormal stack is clustered according to the similarity, and the cluster center is determined.

According to a fourth aspect of the embodiments of the present disclosure, there is provided a computer-readable storage medium on which computer instructions are stored, characterized in that, when the instructions are executed by a processor, the above-mentioned method for clustering abnormal problems is implemented.

It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

Description of drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description serve to explain the principles of the disclosure.

Fig. 1 is a flow chart of a method for clustering abnormal problems according to an exemplary embodiment.

Fig. 2 is a flow chart of a method for clustering abnormal problems according to an exemplary embodiment.

Fig. 3 is a flow chart of a method for clustering abnormal problems according to an exemplary embodiment.

Fig. 4 is a block diagram of an apparatus for clustering abnormal problems according to an exemplary embodiment.

Fig. 5 is a block diagram of a computing module according to an exemplary embodiment.

Fig. 6 is a block diagram of an apparatus for clustering abnormal problems according to an exemplary embodiment.

Fig. 7 is a block diagram of a computing module according to an exemplary embodiment.

Fig. 8 is a block diagram of a clustering module according to an exemplary embodiment.

Fig. 9 is a block diagram of an apparatus according to an exemplary embodiment.

Detailed ways

Exemplary embodiments will be described in detail herein, examples of which are illustrated in the accompanying drawings. Where the following description refers to the drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the illustrative examples below are not intended to represent all implementations consistent with this disclosure. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present disclosure as recited in the appended claims.

In the related art, when clustering the exception stacks, a possible solution is to judge the similarity of the functions between the exception stacks, so as to perform the clustering processing. However, the function names in the exception stack may not reflect the exception problem, so clustering based on the function name is not very effective.

In order to solve the above problem, in this embodiment, the exception stack is clustered according to the exception description information in the exception stack. The exception description information is an intuitive description of the abnormal problem, so the clustering effect is better.

Fig. 1 is a flow chart of a method for clustering abnormal problems according to an exemplary embodiment. As shown in Fig. 1, the method can be implemented by a computer device and includes the following steps:

In step 101, exception description information is extracted from multiple exception stacks, respectively.

In step 102, the similarity of the exception description information between the exception stack and the exception stack is calculated to obtain the similarity.

In step 103, clustering processing is performed on the abnormal stack according to the similarity, and a cluster center is determined.

The exception stack in this embodiment refers to a stack that stores information about exception issues. The exception description information in this embodiment is information related to exception problems other than the functions in the exception stack. Since the exception description information describes the exception problem more accurately and comprehensively, the result of clustering the exception stack according to the exception description information in this embodiment is also more accurate. Among them, to determine the class center is to determine the class obtained after clustering.

This embodiment is applicable to the abnormal problem of native crash (local crash) in the mobile terminal system. The abnormal problem is a multiple problem, the number of problems in one day has reached one million level, and there is no marked data. In this embodiment, the clustering processing of these abnormal problems is realized according to the abnormal description information of these abnormal problems, and the processing result is relatively accurate.

In one embodiment, the exception description information includes at least one of the following: exception language description information, exception flags, exception values, exception code identifiers, library information and class information of functions related to exceptions.

In this embodiment, there may be various types of exception description information. For example, the exception language description information may be the information stored in the abortmessage (abort message) field, which is similar to natural language, and is equivalent to a language description of the exception problem. The exception tag can be the information stored in the abi field, which stores various tags, and several types of tags represent various abnormal problems, so the abnormal problems can be determined according to the tags stored in this field. The abnormal value can be the information stored in the signal (value) field, and the signal stores various values, several of which represent various abnormal problems, so the abnormal problems can be determined according to the values stored in this field. The exception code identifier can be the information stored in the code (code) field. The code stores the identification information of various exception codes. The exception code reflects the abnormal problem, so the exception code identifier is a marker for various abnormal problems, so it can be The exception problem can be determined according to the exception code identifier stored in this field. The library information and class information of the function related to the exception can be the library information and class information of the backtrace, which is a kind of information similar to natural language, which is equivalent to the language description of the exception problem. Backtrace is a function that calls the stack. The language description information of the stack can be determined according to the library information and class information of the backtrace. For the exception stack, the library information and class information of the backtrace describe the exception problem.

This embodiment provides a variety of exception description information, reflects the exception problem from various angles, can more accurately cluster the exception stack, and is suitable for a variety of application scenarios.

In one embodiment, the step 102 includes: step A1-step A3.

In step A1, the edit distance algorithm is used to calculate the similarity of the abnormal language description information between the exception stack and the exception, the library information and class information of the functions related to the exception, and obtain the corresponding sub-similarity of each exception description information. Spend.

In step A2, the exact matching algorithm is used to calculate the similarity between the exception stack and the exception label, exception value, and exception code identifier, respectively, to obtain the corresponding sub-similarity of each item of exception description information.

In step A3, the similarity of the exception stack is obtained according to the sub-similarity of each item of exception description information.

In this embodiment, step A1 and step A2 may be performed synchronously.

In this embodiment, the abnormal language description information, the library information and class information of the functions related to the exception are description information similar to natural language, so the edit distance algorithm is used to calculate the sub-similarity of the abnormal language description information between different exception stacks , and calculate the sub-similarity of library information and class information of exception-related functions between different exception stacks. Edit distance algorithm is an algorithm that calculates the similarity between two strings. There may be multiple edit distance algorithms, all of which are applicable to this embodiment.

The strings identified by the abnormal mark, abnormal value and abnormal code are short and have no semantics, so they are suitable for the exact matching algorithm, so the obtained sub-similarity is 1 (that is, the similarity is 100%) or 0 (that is, completely dissimilar). ).

In this embodiment, the sub-similarity of each item of exception description information between the two exception stacks is obtained through step A1 and step A2. Then, through step A3, the sub-similarity of each item of exception description information between the two exception stacks is aggregated, and the similarity between the exception stack and the exception stack is obtained. The similarity between the exception stack and the exception stack can be obtained by the summation of each sub-similarity, or the weighted summation. The weight of the sub-similarity of each anomaly description information can be preconfigured.

In one embodiment, the step A1 includes: step A11-step A13.

In step A11, the sub-similarity is calculated separately for each layer in the library information and the class information.

In step A12, the multiple sub-similarities of the same level are averaged.

In step A13, weighted summation is performed on the sub-similarities of multiple levels, wherein the sub-similarity with a lower stack level corresponds to a larger weight.

In this embodiment, the library information and the class information may have multiple layers, that is, the exception stack may have multiple calling relationships. There can be multiple items in one layer of library information and class information, that is, there can be multiple exception stacks in one layer.

If there is only one layer of library information and class information, step A12 can be omitted. If there is only one item of library information or class information in one layer of library information and class information, step A13 may be omitted.

When comparing the sub-similarity of the library information and class information of the two exception stacks, a layer-by-layer comparison is performed, and an item-by-item comparison is performed on the same layer, and then summarized. This embodiment considers that each item of library information and class information in the same layer has the same degree of importance, so the sub-similarities of the layer are obtained by averaging multiple sub-similarities of the same layer. For multi-layer library information and class information, it is considered that the closer to the bottom layer, the higher the importance, so the weighted summation of the sub-similarities of multiple levels is performed, wherein the sub-similarity of the lower stack level corresponds to the greater the weight.

This embodiment provides a more detailed similarity calculation method for the hierarchical characteristics of library information and class information, and the obtained similarity is more accurate.

In one embodiment, before calculating the similarity of the exception description information between the exception stack and the exception stack, and obtaining the similarity, the method further includes: step B1-step B2.

In step B1, the exception description information in each exception stack is uniquely encoded to obtain a unique encoding.

In step B2, the exception stack is deduplicated according to the unique code.

In this embodiment, before calculating the similarity, the overall coding of each item of exception description information in the exception stack may be performed, and a unique coding method may be used for coding. If the exception description information of another exception stack is not exactly the same, the obtained coding result is also different, so that the exact same exception stack can be determined. Then, deduplication processing is performed, so that when the similarity is compared later, the identical exception stacks do not need to be compared, which can reduce the amount of calculation when calculating the similarity. In addition, the deduplication process is also a preliminary clustering process, that is, the same exception stacks are grouped into one class.

Through the unique encoding, the same exception stack can be compared quickly, and the comparison process is faster. There are many unique encoding methods, such as MD5 algorithm (digest 5 algorithm) and so on.

In one embodiment, the step 102 includes: step C1-step C2.

In step C1, vector coding is performed on the exception description information in each exception stack to obtain a vector coding result.

In step C2, the similarity of the vector coding result between the exception stack and the exception stack is calculated to obtain the similarity.

In this embodiment, vector encoding is performed on each item of exception description information in the exception stack, which facilitates quick comparison of similarity. The vector coding method may be a numerical coding method, which encodes various abnormal description information into corresponding numerical values. Moreover, when performing clustering, a clustering algorithm with statistical information of the original data can be used to further improve the accuracy of the clustering.

In one embodiment, the step 103 includes: step D1-step D2.

In step D1, the number of the same exception stack is counted. This step can be performed synchronously with step B2, and both deduplication and counting are the process of determining the same exception stack.

In step D2, according to the sequence of the number from high to low, cluster processing is performed on the abnormal stack according to the similarity, and the cluster center is determined.

In this embodiment, the number of the same exception stacks is first counted, which is equivalent to performing preliminary clustering according to 100% similarity, and determining the number of exception stacks in each category after the preliminary clustering. Then perform secondary clustering in descending order of the number, which is equivalent to prioritizing the category with higher coverage, that is, the category with more abnormal problems and more serious problems. In practical applications, we are more concerned about the categories with more serious exception problems, so it is not necessary to cluster all exception stacks. If the coverage rate is higher, the clustering process can be ended, which also improves the processing efficiency of the clustering. The clustering results obtained by the method of this embodiment have high and accurate coverage.

Among them, in the clustering process, a plurality of abnormal stacks whose similarity exceeds a preset threshold are classified into one category. After one clustering process, the accuracy of the clustering can be checked by sampling. If the accuracy does not meet the expectation, the threshold of similarity can be adjusted and clustering is performed again until the accuracy reaches the expectation.

In one embodiment, the step D2 includes: step D21-step D22.

In step D21, the obtained similarity is stored in a preset storage structure; the storage structure includes: a two-layer hash map or an array.

In step D22, the storage structures are traversed in descending order of the number, and cluster processing is performed, and a cluster center is determined.

This embodiment uses a storage structure such as a two-layer hash map or an array to store the respective similarities obtained before, which is convenient to traverse the storage structures in order of the number from high to low, which can improve the processing of clustering. efficiency.

The implementation process is described in detail below through several embodiments.

Fig. 2 is a flow chart of a method for clustering abnormal problems according to an exemplary embodiment. As shown in Fig. 2, the method can be implemented by a computer device and includes the following steps:

In step 201, each item of exception description information is extracted from a plurality of exception stacks, respectively.

In step 202, the entirety of all the exception description information in each exception stack is uniquely encoded to obtain a unique encoding.

In step 203, the exception stack is deduplicated according to the unique code.

In step 204, vector coding is performed on each item of exception description information in each exception stack to obtain a vector coding result.

In step 205, the edit distance algorithm is used to calculate the similarity between the abnormal language description information between the exception stack and the exception, the library information of the function related to the exception, and the vector coding results of the class information, respectively, to obtain each corresponding item. The sub-similarity of anomaly description information.

In step 206, the exact matching algorithm is used to calculate the similarity of the respective vector encoding results of the exception flag, the exception value, and the exception code identifier between the exception stack and the exception, and obtain the corresponding sub-similarity of each item of exception description information. .

In step 207, the similarity of the exception stack is obtained according to the sub-similarity of each item of exception description information.

In step 208, clustering processing is performed on the abnormal stack according to the similarity, and a cluster center is determined.

FIG. 3 is a flowchart of a method for clustering abnormal problems according to an exemplary embodiment. As shown in FIG. 3 , the method can be implemented by a computer device and includes the following steps:

In step 301, exception description information is extracted from multiple exception stacks, respectively.

In step 302, the exception description information in each exception stack is uniquely encoded to obtain a unique encoding.

In step 303, the exception stack is deduplicated according to the unique code, and the number of the same exception stack is counted.

In step 304, the similarity of the exception description information between the exception stack and the exception stack is calculated to obtain the similarity.

In step 305, the obtained similarity is stored in a preset storage structure; the storage structure includes: a two-layer hash map or an array.

In step 306, the storage structures are traversed in descending order of the number, and cluster processing is performed, and the cluster center is determined.

The above embodiments can be combined in various ways according to actual needs.

Through the above introduction, we understand the realization process of abnormal problem clustering, which is realized by computer. The following introduces the internal structure and function of the device.

Fig. 4 is a schematic diagram of an apparatus for clustering abnormal problems according to an exemplary embodiment. Referring to FIG. 4 , the apparatus includes: an extraction module 401 , a calculation module 402 and a clustering module 403 .

The extraction module 401 is configured to extract the exception description information from the multiple exception stacks respectively.

The calculation module 402 is configured to calculate the similarity of the exception description information between the exception stack and the exception stack to obtain the similarity.

The clustering module 403 is configured to perform clustering processing on the abnormal stack according to the similarity, and determine the cluster center.

In one embodiment, the exception description information includes at least one of the following: exception language description information, exception flags, exception values, exception code identifiers, library information and class information of functions related to exceptions.

In one embodiment, as shown in FIG. 5 , the calculation module 402 includes: a first calculation sub-module 501 , a second calculation sub-module 502 and a statistics sub-module 503 .

The first calculation submodule 501 is used for adopting an edit distance algorithm to calculate the similarity of the abnormal language description information between the exception stack and the exception, the library information and class information of the function related to the exception, and obtain the corresponding description of each exception. Sub-similarity of information.

The second calculation sub-module 502 is configured to use an exact matching algorithm to calculate the similarity between the exception stack and the exception label, exception value, and exception code identifier, respectively, to obtain the corresponding sub-similarity of each item of exception description information.

The statistics sub-module 503 is configured to obtain the similarity of the exception stack according to the sub-similarity of each item of exception description information.

In one embodiment, the first calculation sub-module 501 calculates the sub-similarity separately for each layer in the library information and the class information; averages multiple sub-similarities at the same level; The sub-similarities are weighted and summed, wherein the sub-similarity with a lower stack level corresponds to a larger weight.

In one embodiment, as shown in FIG. 6 , the apparatus further includes: a unique encoding module 601 and a deduplication module 602 .

The unique encoding module 601 is configured to uniquely encode the exception description information in each exception stack to obtain the unique encoding.

A deduplication module 602, configured to perform deduplication processing on the exception stack according to the unique code.

In one embodiment, as shown in FIG. 7 , the calculation module 402 includes: a vector encoding sub-module 701 and a third calculation sub-module 702 .

The vector coding sub-module 701 is configured to perform vector coding on the exception description information in each exception stack to obtain a vector coding result.

The third calculation sub-module 702 is configured to calculate the similarity between the exception stack and the vector coding result to obtain the similarity.

In one embodiment, as shown in FIG. 8 , the clustering module 403 includes: a number sub-module 801 and a clustering sub-module 802 .

The number sub-module 801 is used to count the number of the same exception stack.

The clustering sub-module 802 is configured to perform clustering processing on the abnormal stacks according to the similarity according to the order of the number from high to low, and determine the cluster center.

In one embodiment, the clustering sub-module 802 stores the obtained similarity in a preset storage structure; the storage structure includes: a two-layer hash map or an array; according to the number from high to low order, traverse the storage structure, perform clustering processing, and determine the cluster center.

Regarding the apparatus in the above-mentioned embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be described in detail here.

FIG. 9 is a block diagram of an apparatus 900 for clustering abnormal problems according to an exemplary embodiment. For example, apparatus 900 may be provided as a computer. 9, apparatus 900 includes a processing component 922, which further includes one or more processors, and a memory resource, represented by memory 932, for storing instructions executable by processing component 922, such as an application program. An application program stored in memory 932 may include one or more modules, each corresponding to a set of instructions. Additionally, the processing component 922 is configured to execute instructions to perform the method exception problem clustering described above.

Device 900 may also include a power supply assembly 926 configured to perform power management of device 900 , a wired or wireless network interface 950 configured to connect device 900 to a network, and an input output (I/O) interface 958 . Device 900 may operate based on an operating system stored in memory 932, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™ or the like.

A device for clustering abnormal problems, comprising:

processor;

memory for storing processor-executable instructions;

wherein the processor is configured to:

Extract the exception description information from multiple exception stacks respectively;

Calculate the similarity of the exception description information between the exception stack and the exception stack to obtain the similarity;

The abnormal stack is clustered according to the similarity, and the cluster center is determined.

The processor may also be configured to:

The exception description information includes at least one of the following: exception language description information, exception flag, exception value, exception code identifier, library information and class information of the function related to the exception.

The processor may also be configured to:

The similarity between the exception stack and the exception, the exception description information is calculated, and the similarity is obtained, including:

The edit distance algorithm is used to calculate the similarity of the abnormal language description information between the exception stack and the exception, the library information and class information of the functions related to the exception, and obtain the corresponding sub-similarity of each exception description information;

Using the exact matching algorithm, the similarity between the exception stack and the exception label, exception value, and exception code identifier is calculated respectively, and the corresponding sub-similarity of each exception description information is obtained;

According to the sub-similarity of each exception description information, the similarity of the exception stack is obtained.

The processor may also be configured to:

The edit distance algorithm is used to calculate the similarity of the library information and class information of the function related to the exception between the exception stack and the exception, and obtain the corresponding sub-similarity of each item of exception description information, including:

For each layer in the library information and the class information, calculate the sub-similarity respectively;

Average multiple sub-similarities at the same level;

A weighted sum is performed on the sub-similarities of multiple levels, wherein the sub-similarity with a lower stack level corresponds to a larger weight.

The processor may also be configured to:

Before calculating the similarity of the exception description information between the exception stack and the exception stack, and obtaining the similarity, the method further includes:

Uniquely encode the exception description information in each exception stack to obtain a unique encoding;

The exception stack is deduplicated according to the unique code.

The processor may also be configured to:

Calculate the similarity of the exception description information between the exception stack and the exception stack to obtain the similarity, including:

Perform vector encoding on the exception description information in each exception stack to obtain the vector encoding result;

Calculate the similarity of the vector encoding results between the exception stack and the exception stack to obtain the similarity.

The processor may also be configured to:

The clustering process is performed on the abnormal stack according to the similarity, and the cluster center is determined, including:

Count the number of the same exception stack;

In order of the number from high to low, cluster processing is performed on the abnormal stacks according to the similarity, and the cluster center is determined.

The processor may also be configured to:

The abnormal stacks are clustered according to the similarity according to the order of the number from high to low, and the cluster center is determined, including:

The obtained similarity is stored in a preset storage structure; the storage structure includes: a two-layer hash map or an array;

The storage structures are traversed in descending order of the number, and cluster processing is performed, and the cluster center is determined.

A computer-readable storage medium, when an instruction in the storage medium is executed by a processor of a mobile terminal, the mobile terminal can execute a method for clustering abnormal problems, the method comprising:

Extract the exception description information from multiple exception stacks respectively;

Calculate the similarity of the exception description information between the exception stack and the exception stack to obtain the similarity;

The abnormal stack is clustered according to the similarity, and the cluster center is determined.

The instructions in the storage medium may also include:

The exception description information includes at least one of the following: exception language description information, exception flag, exception value, exception code identifier, library information and class information of the function related to the exception.

The instructions in the storage medium may also include:

The similarity between the exception stack and the exception, the exception description information is calculated, and the similarity is obtained, including:

The edit distance algorithm is used to calculate the similarity of the abnormal language description information between the exception stack and the exception, the library information and class information of the functions related to the exception, and obtain the corresponding sub-similarity of each exception description information;

Using the exact matching algorithm, the similarity between the exception stack and the exception label, exception value, and exception code identifier is calculated respectively, and the corresponding sub-similarity of each exception description information is obtained;

According to the sub-similarity of each exception description information, the similarity of the exception stack is obtained.

The instructions in the storage medium may also include:

The edit distance algorithm is used to calculate the similarity of the library information and class information of the function related to the exception between the exception stack and the exception, and obtain the corresponding sub-similarity of each item of exception description information, including:

For each layer in the library information and the class information, calculate the sub-similarity respectively;

Average multiple sub-similarities at the same level;

A weighted sum is performed on the sub-similarities of multiple levels, wherein the sub-similarity with a lower stack level corresponds to a larger weight.

The instructions in the storage medium may also include:

Before calculating the similarity of the exception description information between the exception stack and the exception stack, and obtaining the similarity, the method further includes:

Uniquely encode the exception description information in each exception stack to obtain a unique encoding;

The exception stack is deduplicated according to the unique code.

The instructions in the storage medium may also include:

Calculate the similarity of the exception description information between the exception stack and the exception stack to obtain the similarity, including:

Perform vector encoding on the exception description information in each exception stack to obtain the vector encoding result;

Calculate the similarity of the vector encoding results between the exception stack and the exception stack to obtain the similarity.

The instructions in the storage medium may also include:

The clustering process is performed on the abnormal stack according to the similarity, and the cluster center is determined, including:

Count the number of the same exception stack;

In order of the number from high to low, cluster processing is performed on the abnormal stacks according to the similarity, and the cluster center is determined.

The instructions in the storage medium may also include:

The abnormal stacks are clustered according to the similarity according to the order of the number from high to low, and the cluster center is determined, including:

The obtained similarity is stored in a preset storage structure; the storage structure includes: a two-layer hash map or an array;

The storage structures are traversed in descending order of the number, and cluster processing is performed, and the cluster center is determined.

Other embodiments of the present disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or techniques in the technical field not disclosed by the present disclosure . The specification and examples are to be regarded as exemplary only, with the true scope and spirit of the disclosure being indicated by the following claims.

It is to be understood that the present disclosure is not limited to the precise structures described above and illustrated in the accompanying drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims (18)

1. A method for clustering abnormal problems, comprising:

extracting the abnormal description information from a plurality of abnormal stacks respectively;

calculating the similarity of the abnormal description information between the abnormal stack and the abnormal stack to obtain the similarity;

and clustering the abnormal stacks according to the similarity, and determining a class center.

2. The method for abnormal problem clustering according to claim 1, wherein the abnormal description information includes at least one of: exception language description information, exception tags, exception values, exception code identification, library information for functions associated with the exception, and class information.

3. The method for clustering abnormal problems according to claim 2, wherein the calculating the similarity between the abnormal stack and the abnormal description information to obtain the similarity comprises:

respectively calculating similarity of the abnormal language description information between the abnormal stack and the abnormality, the library information of the function related to the abnormality and the class information by adopting an edit distance algorithm to obtain the sub-similarity of each corresponding item of abnormal description information;

respectively calculating the similarity of the abnormal marks, the abnormal values and the abnormal code identifications between the abnormal stack and the abnormality by adopting an accurate matching algorithm to obtain the sub-similarity of each corresponding item of abnormal description information;

and obtaining the similarity of the exception stack according to the sub-similarity of each item of exception description information.

4. The method for clustering abnormal problems according to claim 3, wherein the calculating the similarity of the library information and the class information of the function related to the abnormality between the abnormality stack and the abnormality by using the edit distance algorithm to obtain the sub-similarity of each corresponding abnormality description information comprises:

calculating sub-similarity respectively for each layer in the library information and the class information;

averaging the similarity of a plurality of children at the same level;

and carrying out weighted summation on the sub-similarities of the multiple levels, wherein the lower sub-similarity of the stack level corresponds to the higher weight.

5. The method for clustering abnormal problems according to claim 1, wherein before the similarity is obtained by calculating the similarity between the abnormal stack and the abnormal description information, the method further comprises:

uniquely encoding the abnormal description information in each abnormal stack to obtain unique codes;

and performing duplicate removal processing on the abnormal stack according to the unique code.

6. The method for clustering abnormal problems according to claim 1, wherein the step of calculating the similarity between the abnormal stack and the similarity between the abnormal description information to obtain the similarity comprises:

carrying out vector coding on the abnormal description information in each abnormal stack to obtain a vector coding result;

and calculating the similarity of the vector coding results between the abnormal stack and the abnormal stack to obtain the similarity.

7. The method for clustering abnormal problems according to claim 1, wherein the clustering abnormal stacks according to the similarity and determining class centers comprises:

counting the number of the same abnormal stacks;

and according to the sequence of the number from high to low, clustering the abnormal stacks according to the similarity, and determining the class centers.

8. The method for clustering abnormal problems according to claim 7, wherein the clustering abnormal stacks according to the similarity and determining class centers according to the sequence of the number from high to low comprises:

storing the obtained similarity into a preset storage structure; the storage structure includes: two levels of Hash diagrams or arrays;

and traversing the storage structures according to the sequence of the number from high to low, clustering, and determining the class center.

9. An apparatus for clustering abnormal problems, comprising:

the extracting module is used for respectively extracting the abnormal description information from the abnormal stacks;

the calculation module is used for calculating the similarity of the abnormal description information between the abnormal stack and the abnormal stack to obtain the similarity;

and the clustering module is used for clustering the abnormal stacks according to the similarity and determining the class centers.

10. The apparatus for clustering abnormal problems according to claim 9, wherein the abnormal description information includes at least one of: exception language description information, exception tags, exception values, exception code identification, library information for functions associated with the exception, and class information.

11. The apparatus for anomaly clustering according to claim 10, wherein said calculation module comprises:

the first calculation submodule is used for respectively calculating the similarity of the abnormal language description information between the abnormal stack and the abnormality, the library information of the function related to the abnormality and the class information by adopting an edit distance algorithm to obtain the sub-similarity of each corresponding abnormal description information;

the second calculation submodule is used for respectively calculating the similarity of the abnormal marks, the abnormal values and the abnormal code identifications between the abnormal stack and the abnormality by adopting an accurate matching algorithm to obtain the sub-similarity of each corresponding item of abnormal description information;

and the statistic submodule is used for obtaining the similarity of the abnormal stack according to the sub-similarity of each item of abnormal description information.

12. The apparatus for abnormal problem clustering according to claim 11, wherein the first computation submodule separately computes a degree of sub-similarity for each layer of the library information and the class information; averaging the similarity of a plurality of children at the same level; and carrying out weighted summation on the sub-similarities of the multiple levels, wherein the lower sub-similarity of the stack level corresponds to the higher weight.

13. The apparatus for abnormal problem clustering according to claim 9, wherein the apparatus further comprises:

the unique coding module is used for uniquely coding the abnormal description information in each abnormal stack to obtain a unique code;

and the duplicate removal module is used for carrying out duplicate removal processing on the abnormal stack according to the unique code.

14. The apparatus for anomaly clustering according to claim 9, wherein the calculation module comprises:

the vector coding submodule is used for carrying out vector coding on the abnormal description information in each abnormal stack to obtain a vector coding result;

and the third calculation submodule is used for calculating the similarity between the abnormal stack and the vector coding result to obtain the similarity.

15. The apparatus for abnormal problem clustering according to claim 9, wherein the clustering module comprises:

the number submodule is used for counting the number of the same abnormal stacks;

and the clustering submodule is used for clustering the abnormal stacks according to the similarity and determining the class centers according to the sequence of the number from high to low.

16. The apparatus for abnormal problem clustering according to claim 15, wherein the clustering submodule stores the obtained similarity into a preset storage structure; the storage structure includes: two levels of Hash diagrams or arrays; and traversing the storage structures according to the sequence of the number from high to low, clustering, and determining the class center.

17. An apparatus for clustering abnormal problems, comprising:

a processor;

a memory for storing processor-executable instructions;

wherein the processor is configured to:

extracting the abnormal description information from a plurality of abnormal stacks respectively;

calculating the similarity of the abnormal description information between the abnormal stack and the abnormal stack to obtain the similarity;

and clustering the abnormal stacks according to the similarity, and determining a class center.

18. A computer-readable storage medium having computer instructions stored thereon, wherein the instructions, when executed by a processor, implement the method of claims 1 to 8.

Images