CN112069069A - Defect automatic location analysis method, device and readable storage medium - Google Patents

Abstract

The invention relates to an automatic testing tool, and provides a defect automatic positioning analysis method, equipment and medium. In the present invention, the defect data to be analyzed is firstly matched based on the text similarity algorithm, so that when facing the common defect problems in the defect knowledge base, the cause of the defect and the repair plan can be quickly matched; When the defect data to be analyzed is invalid, further fuzzy matching is performed, so that when an accurate result cannot be found, a best matching result that can be found in the current knowledge base is given first; In the repair scheme, the knowledge base is updated and the defect data to be verified is calibrated based on this, so that the defect analysis results of the defect data only through fuzzy matching can be improved, and the knowledge base is adaptively optimized to further improve the efficiency of defect location analysis. In addition, the present invention also relates to the blockchain technology, and the defect data to be verified can be stored in the blockchain.

Description

Defect automatic location analysis method, device and readable storage medium

technical field

The present invention relates to the technical field of software testing, and in particular, to a method, device and computer-readable storage medium for automatic defect location analysis.

Background technique

Whether you are a software developer or a software tester, your day-to-day work involves software defects. Testers need to find program defects or functional problems as much as possible before the program project is officially launched, and fix them one by one by the developers, so as to ensure the high-quality and stable delivery of the program project according to the actual needs of users. This process often costs A lot of time, therefore, how to quickly and accurately investigate and locate the cause of the defect, and effectively repair it in a targeted manner, is a long-term problem that developers and testers need to face.

At present, the inspection and location of software defects in the industry are mainly carried out manually by querying logs, packet capture analysis and other methods. The efficiency of inspection and location depends on testing and the personal experience of developers. The location process is relatively blind and difficult, which often results in failure to timely detect defects. be repaired, thereby blocking the development and testing progress, thus causing the technical problem of inefficiency of the existing software defect location analysis methods.

SUMMARY OF THE INVENTION

The main purpose of the present invention is to provide an automatic defect location analysis method, device and computer-readable storage medium, aiming at solving the technical problem of low efficiency of the existing software defect location analysis method.

In order to achieve the above object, the present invention provides a method for automatic positioning and analysis of defects, and the method for automatic positioning and analysis of defects comprises the following steps:

Acquire defect data to be analyzed, and extract keywords in the defect data to be analyzed, so as to accurately match the keywords in a preset defect knowledge base based on a preset text similarity algorithm, wherein the defect knowledge The library contains multi-category defect sample data, and defect causes and repair schemes corresponding to the multi-category defect sample data;

When it is detected that the current exact match fails, identify the defect category to which the defect data to be analyzed belongs, and search the defect knowledge base for the target defect cause and target repair scheme corresponding to the defect category, so as to complete the Fuzzy matching of defect data to be analyzed;

The defect data to be analyzed after the fuzzy matching is used as the defect data to be verified, and when the accurate defect cause and accurate repair scheme of the defect data to be verified are obtained, the accurate defect cause and accurate repair scheme are updated to the above. In the defect knowledge base, to calibrate the defect data to be calibrated.

Optionally, when detecting that the current exact match fails, identify the defect category to which the defect data to be analyzed belongs, and search the defect knowledge base for the target defect cause and target repair scheme corresponding to the defect category. , the steps of completing the fuzzy matching of the defect data to be analyzed include:

When detecting that the current exact match fails, use a preset entity recognition model to identify the entity information of the defect data to be analyzed, and obtain the problem template based on the entity information;

Perform multi-level semantic analysis on the defect data to be analyzed to obtain the multi-level semantics of the defect data to be analyzed;

Using a preset probability graph model and combining the problem template and the multi-level semantics, predict that the defect data to be analyzed corresponds to the defect category in the defect knowledge base;

According to the defect category and the entity information, the defect data to be analyzed is converted into a structured query of the defect knowledge base, and the target defect cause and the target repair plan are obtained through the query, so as to complete the analysis of the defect to be analyzed. Fuzzy matching of data.

Optionally, the fuzzy matching defect data to be analyzed is used as the defect data to be verified, and when the accurate defect cause and the accurate repair plan of the defect data to be verified are obtained, the accurate defect cause and the accurate repair plan are obtained. After the step of updating the accurate repair scheme to the defect knowledge base to calibrate the defect data to be calibrated, the method further includes:

Marking the calibrated defect data to be calibrated as characteristic defect data;

When it is detected that fuzzy matching is currently performed, the feature defect data is preferentially selected for matching.

Optionally, before the step of acquiring defect data to be analyzed and extracting keywords in the defect data to be analyzed, the method further includes:

Obtaining defective sample data, and performing pre-screening and format conversion on the defective sample data to obtain target sample data;

Classifying the target sample data based on a preset classification algorithm to obtain multi-category defect sample data corresponding to multiple defect categories;

Extracting and screening the multi-category defect sample data to obtain the defect cause and repair scheme corresponding to each defect category, and establishing a mapping relationship between the defect category, defect cause and repair scheme information;

When it is detected that the data amount of the multi-category defect sample data reaches a preset data amount threshold, the defect knowledge base is constructed to perform automatic defect location analysis based on the defect knowledge base.

Optionally, the text similarity algorithm includes a cosine distance algorithm,

The step of acquiring defect data to be analyzed, and extracting keywords in the defect data to be analyzed, so as to accurately match the keywords in a preset defect knowledge base based on a preset text similarity algorithm includes:

Using natural language processing technology to perform word segmentation processing on the defect data to be analyzed to extract keywords in the defect data to be analyzed;

generating a first word frequency vector set of the keyword in the defect data to be analyzed, and a second word frequency vector set of the keyword in the defect knowledge base;

A cosine similarity set between the first word frequency vector and the second word frequency vector is obtained by using a cosine distance algorithm, so as to accurately match the keyword in the defect knowledge base based on the cosine similarity set.

Optionally, using a cosine distance algorithm to obtain a cosine similarity set between the first word frequency vector and the second word frequency vector, so as to place the keyword in the defect knowledge based on the cosine similarity set. After the exact matching steps in the library, it also includes:

Judging whether there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set;

If so, it is determined that the current exact match is successful;

If not, it is determined that the current exact match fails.

Optionally, acquiring the defect data to be analyzed, and extracting keywords in the defect data to be analyzed, so as to accurately match the keywords in a preset defect knowledge base based on a preset text similarity algorithm. After the steps, also include:

When detecting that there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set, acquiring matching defect sample data corresponding to the target cosine similarity in the defect knowledge base;

Obtain the matching problem cause and matching repair scheme corresponding to the matching defect sample data, and display the matching problem cause and matching repair scheme in association.

Optionally, the fuzzy matching defect data to be analyzed is used as the defect data to be verified, and when the accurate defect cause and the accurate repair plan of the defect data to be verified are obtained, the accurate defect cause and the accurate repair plan are obtained. After the step of updating the accurate repair scheme to the defect knowledge base to calibrate the defect data to be calibrated, the method further includes:

According to preset time intervals, regularly screen out the common defect data whose actual recurrence times exceed the preset times threshold from the defect knowledge base, and determine the common defect category, common defect cause and common repair scheme corresponding to the common defect data , to generate a visual common defect statistics table.

In addition, in order to achieve the above purpose, the present invention also provides a defect automatic positioning analysis device, the defect automatic positioning analysis device includes:

The exact matching module is used to obtain the defect data to be analyzed, and extract the keywords in the defect data to be analyzed, so as to accurately match the keywords in the preset defect knowledge base based on the preset text similarity algorithm, Wherein, the defect knowledge base includes multi-category defect sample data, and defect causes and repair solutions corresponding to the multi-category defect sample data;

The fuzzy matching module is used for identifying the defect category to which the defect data to be analyzed belongs when the current exact matching failure is detected, and searching the defect knowledge base for the target defect cause and the target repair scheme corresponding to the defect category , to complete the fuzzy matching of the defect data to be analyzed;

The defect calibration module is used to use the fuzzy matching defect data to be analyzed as the defect data to be verified, and when the accurate defect cause and accurate repair plan of the defect data to be verified are obtained, the accurate defect cause and the accurate repair plan are obtained. The accurate repair scheme is updated into the defect knowledge base to calibrate the defect data to be calibrated.

Optionally, the fuzzy matching module includes:

a problem module acquiring unit, configured to use a preset entity recognition model to identify the entity information of the defect data to be analyzed, and obtain the problem template based on the entity information when it is detected that the current exact match fails;

a multi-level semantic parsing unit, configured to perform multi-level semantic parsing on the defect data to be analyzed to obtain the multi-level semantics of the defect data to be analyzed;

A defect category prediction unit, configured to use a preset probability graph model and combine the problem template and the multi-level semantics to predict the defect category in the defect knowledge base corresponding to the defect data to be analyzed;

The fuzzy matching completion unit is used to convert the defect data to be analyzed into a structured query of the defect knowledge base according to the defect category and the entity information, and obtain the target defect cause and target repair plan by querying Complete the fuzzy matching on the defect data to be analyzed.

Optionally, the automatic defect location analysis device further includes:

The feature marking module is used for feature marking the calibrated defect data to be calibrated as feature defect data;

The preferential matching module is configured to preferentially select the feature defect data for matching when it is detected that fuzzy matching is currently being performed.

Optionally, the automatic defect location analysis device further includes:

A screening and conversion module, used for acquiring defective sample data, and performing pre-screening and format conversion on the defective sample data to obtain target sample data;

a data classification module, configured to classify the target sample data based on a preset classification algorithm to obtain multi-category defect sample data corresponding to a plurality of defect categories;

The mapping establishment module is used for extracting and screening the multi-category defect sample data to obtain the defect cause and repair scheme corresponding to each defect category, and establish the mapping relationship between the defect category, defect cause and repair scheme information ;

The knowledge base building module is configured to build the defect knowledge base when it is detected that the data volume of the multi-category defect sample data reaches a preset data volume threshold, so as to perform automatic defect location analysis based on the defect knowledge base.

Optionally, the text similarity algorithm includes a cosine distance algorithm,

The exact matching module includes:

A key extraction unit, used for performing word segmentation processing on the defect data to be analyzed by using natural language processing technology, so as to extract keywords in the defect data to be analyzed;

A word frequency generating unit, configured to generate a first word frequency vector set of the keyword in the defect data to be analyzed, and a second word frequency vector set of the keyword in the defect knowledge base;

An exact matching unit is used to obtain a cosine similarity set between the first word frequency vector and the second word frequency vector using a cosine distance algorithm, so as to classify the keyword in the defect knowledge based on the cosine similarity set. exact match in the library.

Optionally, the automatic defect location analysis device further includes:

an exact matching judgment module, used for judging whether there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set;

If so, it is determined that the current exact match is successful;

If not, it is determined that the current exact match fails.

Optionally, the automatic defect location analysis device further includes:

A matching sample acquisition module, configured to acquire a matching defect sample corresponding to the target cosine similarity in the defect knowledge base when detecting that there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set data;

The result association display module is used for acquiring the matching problem cause and the matching repair scheme corresponding to the matching defect sample data, and displaying the matching problem cause and the matching repair scheme in association.

Optionally, the automatic defect location analysis device further includes:

The regular statistics module is used to regularly screen out the common defect data whose actual recurrence times exceed the preset times threshold from the defect knowledge base according to the preset time interval, and determine the common defect category, common defect category corresponding to the regular defect data, and common defect data. Defect causes and common fixes to generate a visual statistics table of common defects.

In addition, in order to achieve the above object, the present invention also provides an automatic defect location analysis device, the defect automatic location analysis device includes a processor, a memory, and a defect automatic location stored in the memory and executed by the processor. A location analysis program, wherein when the defect automatic location analysis program is executed by the processor, the steps of the above-mentioned defect automatic location analysis method are implemented.

In addition, in order to achieve the above object, the present invention also provides a computer-readable storage medium on which an automatic defect location analysis program is stored, wherein when the defect automatic location analysis program is executed by a processor, the The steps of the automatic defect location analysis method as described above.

The present invention provides an automatic defect location analysis method, device and computer-readable storage medium. The automatic defect location analysis method performs accurate matching based on a text similarity algorithm based on defect data to be analyzed first, so that in the face defect knowledge base When the existing common defect problems are found, the cause of the defect and the repair plan can be quickly matched; when the exact matching method is invalid for the current defect data to be analyzed, further fuzzy matching is performed, and the major categories that match the defect data in the knowledge base are first searched. That is, the defect type, and then correspondingly find the cause and repair plan corresponding to the defect type as the result of fuzzy matching, so that when no accurate result can be found, the closest result that can be found in the current knowledge base is given first; By updating the knowledge base and calibrating the defect data to be checked when the exact cause and repair scheme of the defect data to be verified are obtained, it is possible to improve the defect analysis results of the defect data only through fuzzy matching, and to adapt the knowledge base Optimization to further improve the efficiency of defect location analysis, thereby solving the technical problem of low efficiency of the existing software defect location analysis methods. In addition, compared with the traditional methods of manually locating defects such as query logs and packet capture analysis, the automatic defect locating and analysis method can realize self-calibration of the locating results through the adaptive optimization process, and can be applied to the software development and testing process very well. Good help developers and testers to quickly locate program defects, provide reliable and effective repair suggestions, and achieve rapid repair of defects. Especially when the scale of the program project is large, the business scenario is complex, and it involves multi-person collaborative development and testing, it can significantly shorten the time for defect investigation and positioning, improve the efficiency of development and testing, and ensure the high quality and reliability of software programs.

Description of drawings

Fig. 1 is the hardware structure schematic diagram of the defect automatic location analysis equipment involved in the embodiment scheme of the present invention;

FIG. 2 is a schematic flowchart of the first embodiment of the automatic defect location analysis method according to the present invention;

FIG. 3 is a schematic diagram of functional modules of the automatic defect location analysis device of the present invention.

The realization, functional characteristics and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments.

Detailed ways

It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

The automatic defect location analysis method involved in the embodiment of the present invention is mainly applied to an automatic defect location analysis device, which may be a device with display and processing functions, such as a PC, a portable computer, and a mobile terminal.

Referring to FIG. 1 , FIG. 1 is a schematic diagram of the hardware structure of the automatic defect location analysis device involved in the solution of the embodiment of the present invention. In this embodiment of the present invention, the automatic defect location analysis device may include a processor 1001 (eg, a CPU), a communication bus 1002 , a user interface 1003 , a network interface 1004 , and a memory 1005 . Wherein, the communication bus 1002 is used to realize the connection and communication between these components; the user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard); the network interface 1004 may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface); the memory 1005 can be a high-speed RAM memory, or a non-volatile memory, such as a disk memory, and the memory 1005 can optionally be a storage device independent of the aforementioned processor 1001 .

Those skilled in the art can understand that the hardware structure shown in FIG. 1 does not constitute a limitation on the automatic defect location analysis device, and may include more or less components than the one shown, or combine some components, or different components layout.

Continuing to refer to FIG. 1 , the memory 1005 as a computer-readable storage medium in FIG. 1 may include an operating system, a network communication module, and an automatic defect location analysis program.

In FIG. 1, the network communication module is mainly used to connect to the server and perform data communication with the server; and the processor 1001 can call the defect automatic location analysis program stored in the memory 1005, and execute the defect automatic location analysis method provided by the embodiment of the present invention. .

Based on the above hardware structure, various embodiments of the automatic defect location analysis method of the present invention are proposed.

Whether you are a software developer or a software tester, your day-to-day work involves software defects. Testers need to find program defects or functional problems as much as possible before the program project is officially launched, and fix them one by one by the developers, so as to ensure the high-quality and stable delivery of the program project according to the actual needs of users. This process often costs A lot of time, therefore, how to quickly and accurately investigate and locate the cause of the defect, and effectively repair it in a targeted manner, is a long-term problem that developers and testers need to face.

At present, the inspection and location of software defects in the industry are mainly carried out manually by querying logs, packet capture analysis and other methods. The efficiency of inspection and location depends on testing and the personal experience of developers. The location process is relatively blind and difficult, which often results in failure to timely detect defects. be repaired, thereby blocking the development and testing progress, thus causing the technical problem of inefficiency of the existing software defect location analysis methods.

In order to solve the above problem, the present invention provides an automatic defect location analysis method, that is, by firstly performing accurate matching based on the text similarity algorithm based on the defect data to be analyzed, so that when facing the existing common defect problems in the defect knowledge base, it is possible to quickly The cause of the defect and the repair plan are obtained by matching; when the exact matching method is invalid for the current defect data to be analyzed, the fuzzy matching is further carried out. The cause and repair scheme corresponding to the defect type are used as the result of fuzzy matching, so that when the exact result cannot be found, the closest result that can be found in the current knowledge base is given first; Update the knowledge base and calibrate the defect data to be verified based on the accurate cause and repair plan, so that the defect analysis results of the defect data only through fuzzy matching can be improved, and the knowledge base can be adaptively optimized to further improve the efficiency of defect location analysis. , so as to solve the technical problem of low efficiency of the existing software defect location analysis method. In addition, compared with the traditional methods of manually locating defects such as query logs and packet capture analysis, the automatic defect locating and analysis method can realize self-calibration of the locating results through the adaptive optimization process, and can be applied to the software development and testing process very well. Good help developers and testers to quickly locate program defects, provide reliable and effective repair suggestions, and achieve rapid repair of defects. Especially when the scale of the program project is large, the business scenario is complex, and it involves multi-person collaborative development and testing, it can significantly shorten the time for defect investigation and positioning, improve the efficiency of development and testing, and ensure the high quality and reliability of software programs.

Referring to FIG. 2 , FIG. 2 is a schematic flowchart of the first embodiment of the automatic defect location analysis method of the present invention.

The first embodiment of the present invention provides an automatic defect location analysis method, and the defect automatic location analysis method includes the following steps:

Step S10, acquiring defect data to be analyzed, and extracting keywords in the defect data to be analyzed, so as to accurately match the keywords in a preset defect knowledge base based on a preset text similarity algorithm, wherein the The defect knowledge base includes multi-category defect sample data, and defect causes and repair solutions corresponding to the multi-category defect sample data;

In this embodiment, the defect data to be analyzed is usually defect data encountered by software developers or software testers in their daily work, and can be obtained from a defect management platform, an automated testing platform, and a security vulnerability scanning tool. Keywords are words that can represent actual meanings in the defect data to be analyzed. Specifically, the defect data to be analyzed can be screened for invalid words without actual meaning, such as modal particles and pause words, so as to improve the extraction efficiency of keywords. The preset text similarity algorithm may specifically be Euclidean distance, cosine similarity, Jaccard distance, edit distance, Hamiltonian distance, etc., which is not specifically limited in this embodiment. The multi-category defect sample data is the defect sample data belonging to various defect categories, and the defect category can be UI display abnormality, interface error report, and data writing table error. security breaches, etc.

The method is applied to a defect automatic positioning analysis system (hereinafter referred to as the system) loaded with a defect automatic positioning analysis program. When the user submits new defect data to the defect management platform, the system will receive the defect data that needs to be analyzed currently reported by the defect management platform, and trigger an automatic comparison with the preset defect knowledge base in the system. The system will first perform keyword extraction according to the currently submitted phenomenon description of the defect to be analyzed, and the specific extraction method is usually natural language processing. After the system proposes the keyword of the phenomenon description of the defect to be analyzed, it uses the keyword to accurately match the defect sample data in the defect knowledge base. The specific exact matching method is usually to obtain the similarity between the keyword and each defect sample data, and compare the similarity with a preset similarity threshold. If the similarity exceeds the threshold, the system can determine that the exact match with the defect data to be analyzed is successful; if the similarity does not exceed the threshold, the system can determine that the exact match with the defect data to be analyzed fails. If the similarity between multiple keywords of the defect data to be analyzed exceeds the threshold in the knowledge base, the defect sample data with the highest similarity is selected as the exact matching object.

Step S20, when it is detected that the current exact match fails, identify the defect category to which the defect data to be analyzed belongs, and search the defect knowledge base for the target defect cause and target repair scheme corresponding to the defect category, to complete the process. Fuzzy matching to the defect data to be analyzed;

In this embodiment, when the accurate analysis of the defect data to be analyzed fails (specifically, there is no multi-category defect sample data whose similarity with the keyword exceeds a preset threshold in the knowledge base), the system continues to perform fuzzy matching on it. The system uses the defect data to be analyzed as the input of the preset classification algorithm, and obtains the defect category to which the defect data to be analyzed belongs. The system finds this defect category in the defect knowledge base, as well as the target defect cause and target repair plan corresponding to this defect category, as the result of this fuzzy matching. The system can directly display the target defect cause and target repair plan on the front end for users to obtain directly. It should be noted that, in practical application, the identification method of the defect category may be a method based on a classification algorithm, or a method of transforming the defect data to be identified into a structured query of the defect knowledge base. The classification algorithm may be a support vector machine (SVM, Support Vector Machine), a random forest, a decision tree, a nearest neighbor algorithm (K-NN, K-Nearest Neighbor), etc., which is not specifically limited in this embodiment.

Step S30, taking the fuzzy matching defect data to be analyzed as the defect data to be verified, and when acquiring the accurate defect cause and accurate repair scheme of the defect data to be verified, the accurate defect cause and the accurate repair scheme are obtained. The defect knowledge base is updated to calibrate the defect data to be calibrated.

In this embodiment, the defect data to be verified is defect data obtained by performing fuzzy matching after the exact matching fails.

The system marks the defect data to be analyzed after fuzzy matching with the identification to be verified as the defect data to be verified. At the time of defect acceptance, the system will associate the exact defect cause and correct repair plan of the defect data to be verified, associate it with the defect data to be verified and update it to the defect knowledge base, so as to verify the defect data to be verified. In the next fuzzy matching, the feature defects in the knowledge base will be compared first, so as to realize the self-adaptive localization result calibration. The system can also use the verified defect data and its corresponding defect causes, repair plans and defect categories as new training data sets to optimize the training of the classification algorithm to improve the recognition accuracy of the classification algorithm. At the same time, the system will diff the code, analyze the most recently submitted code with defects, and mark the changed code segments in red. Developers can refer to the analysis results and repair suggestions to quickly complete defect repairs.

As a specific embodiment, the automatic defect location analysis system first collects data, collects BUG data found in daily tests, and after repairing the cause of the mobile phone problem and repair plan, forms a closed quality loop; The mapping relationship between the three problem repair solutions, according to the problem appearance and classification, quickly locate the cause, and give repair suggestions; then build a knowledge base, maintain the system common problem set, conduct statistical analysis on a regular basis, identify quality hidden dangers, and share coding experience; finally Self-adaptive optimization, automatic positioning result calibration, continuously optimizes the positioning accuracy through the self-adaptive process.

In this embodiment, the present invention obtains the defect data to be analyzed, and extracts the keywords in the defect data to be analyzed, so that the keywords are stored in the preset defect knowledge base based on the preset text similarity algorithm. Exact matching, wherein the defect knowledge base includes multi-category defect sample data, as well as defect causes and repair solutions corresponding to the multi-category defect sample data; when detecting that the current accurate matching fails, identify the defect data to be analyzed The defect category to which it belongs, and find the target defect cause and target repair scheme corresponding to the defect category in the defect knowledge base, so as to complete the fuzzy matching of the defect data to be analyzed; The data is used as the defect data to be verified, and when the accurate defect cause and accurate repair scheme of the to-be-verified defect data are obtained, the accurate defect cause and accurate repair scheme are updated to the defect knowledge base, so as to correct the defect knowledge base. The defect data to be calibrated is calibrated. Through the above method, the present invention performs accurate matching based on the text similarity algorithm based on the defect data to be analyzed first, so that when facing the existing common defect problems in the defect knowledge base, the defect cause and repair scheme can be quickly matched to obtain the defect cause and repair plan; When the matching method is invalid for the current defect data to be analyzed, further fuzzy matching is performed. First, find the major category that matches the defect data in the knowledge base, that is, the defect type, and then find the cause and repair plan corresponding to the defect type. The result of fuzzy matching makes it possible to first give the closest result that can be found in the current knowledge base when the exact result cannot be found; by updating the knowledge base based on the exact cause and repair scheme of the defect data to be verified And calibrate the defect data to be verified, so that the defect analysis results of the defect data only through fuzzy matching can be improved, and the knowledge base can be adaptively optimized to further improve the efficiency of defect location analysis, thus solving the existing software defect location analysis. The technical problem of inefficiency of the way. In addition, compared with the traditional methods of manually locating defects such as query logs and packet capture analysis, the automatic defect locating and analysis method can realize self-calibration of the locating results through the adaptive optimization process, and can be applied to the software development and testing process very well. Good help developers and testers to quickly locate program defects, provide reliable and effective repair suggestions, and achieve rapid repair of defects. Especially when the scale of the program project is large, the business scenario is complex, and it involves multi-person collaborative development and testing, it can significantly shorten the time for defect investigation and positioning, improve the efficiency of development and testing, and ensure the high quality and reliability of software programs.

Further, based on the above-mentioned first embodiment shown in FIG. 2 , a second embodiment of the automatic defect location analysis method of the present invention is proposed. In this embodiment, step S20 includes:

When detecting that the current exact match fails, use a preset entity recognition model to identify the entity information of the defect data to be analyzed, and obtain the problem template based on the entity information;

Perform multi-level semantic analysis on the defect data to be analyzed to obtain the multi-level semantics of the defect data to be analyzed;

Using a preset probability graph model and combining the problem template and the multi-level semantics, predict that the defect data to be analyzed corresponds to the defect category in the defect knowledge base;

According to the defect category and the entity information, the defect data to be analyzed is converted into a structured query of the defect knowledge base, and the target defect cause and the target repair plan are obtained through the query, so as to complete the analysis of the defect to be analyzed. Fuzzy matching of data.

In this embodiment, when the system detects that the current exact match fails (for example, the highest value of the cosine similarity is lower than the preset similarity threshold), the system uses a Long Short-Term Memory (LSTM, Long Short-Term Memory)-conditional random field ( The entity recognition model of CRF, conditional random field) can identify the entity part from the natural language part of the defect data to be analyzed, and can map the entity in the natural sentence to the corresponding concept, and the natural language conceptualization can assist the model to learn More precise semantic information. The system can convert the defect knowledge base into a knowledge graph, and obtain more relational expressions by expanding the attributes (relationship types) in the knowledge graph. High-quality question templates are obtained by mapping entity concepts and extending attributes of knowledge base graphs.

The system first performs multi-level semantic analysis of the defect data to be identified, and then uses the preset probability graph model to comprehensively use the semantic analysis results of the problem, and through the obtained problem template, predicts the attribute type corresponding to the defect data to be identified in the knowledge graph. Finally, according to the obtained attribute type and the entity and other information in the defect data to be identified identified by the model, the defect data to be identified is converted into a structured query of the knowledge graph, and the current defect data closest to the defect data to be identified is obtained by querying the defect knowledge base. The target defect cause of the matcher and the target repair scheme are used to complete the fuzzy matching of the defect data to be analyzed.

Further, after step S30, it also includes:

Marking the calibrated defect data to be calibrated as characteristic defect data;

When it is detected that fuzzy matching is currently performed, the feature defect data is preferentially selected for matching.

In this embodiment, the system marks the defects of the calibrated defect data with a feature mark as the feature defect data, and will preferentially compare the feature defect data in the knowledge base in the next fuzzy matching, thereby realizing self-adaptive positioning result calibration. At the same time, the system will diff the code, analyze the most recently submitted code with defects, and mark the changed code segments in red. Development colleagues can refer to the analysis results and repair suggestions to quickly complete defect repairs.

It should be emphasized that, in order to further ensure the privacy and security of the above-mentioned characteristic defect data, the above-mentioned characteristic defect data may also be stored in a node of a blockchain.

Further, before step S10, it also includes:

Obtaining defective sample data, and performing pre-screening and format conversion on the defective sample data to obtain target sample data;

Classifying the target sample data based on a preset classification algorithm to obtain multi-category defect sample data corresponding to multiple defect categories;

Extracting and screening the multi-category defect sample data to obtain the defect cause and repair scheme corresponding to each defect category, and establishing a mapping relationship between the defect category, defect cause and repair scheme information;

When it is detected that the data amount of the multi-category defect sample data reaches a preset data amount threshold, the defect knowledge base is constructed to perform automatic defect location analysis based on the defect knowledge base.

In this embodiment, defect sample data may be collected from a defect management platform, an automated testing platform, and a security vulnerability scanning tool. The purpose of the format conversion operation is to convert defect sample data, which may belong to different types, into a uniform format. Specifically, the pre-screening method may be based on the specific identification in the sample defect card. The unified format after the conversion may specifically be a JSON format, a string format, or the like. The preset classification algorithm may specifically be a support vector machine (SVM, Support Vector Machine), a random forest, a decision tree, a nearest neighbor algorithm (K-NN, K-Nearest Neighbor), and the like.

Specifically, when the system builds a defect knowledge base in the early stage, it first collects defect sample data from defect management platforms, automated testing platforms, and security vulnerability scanning tools, and maintains basic defect information, including defect descriptions, causes of problems, and problem fixes. method, in which the cause of the problem and the repair method need to be manually maintained during the acceptance of the defective card at the initial stage; after the data collection is completed, it enters the processing layer, which is first preprocessed to screen out some invalid data such as false positives in the test or not actually defects. Then, the defect data collected by different platforms is converted into a unified format, and then processed by the KNN classification algorithm, and the defect screening is divided into some major categories, such as UI display exception, interface error, data writing table error, security loophole, etc.; data processing After completion, it will reach the analysis layer, where the collected defect data will be analyzed, and the mapping relationship between each defect appearance, problem cause, and solution will be established, and maintained in the knowledge base at the same time; when the defect samples in the knowledge base are When the data is maintained to a certain scale, the knowledge base at this time can be used as a defect knowledge base that can be used directly, and the automatic location analysis process can be performed.

Further, after step S30, it also includes:

According to preset time intervals, regularly screen out the common defect data whose actual recurrence times exceed the preset times threshold from the defect knowledge base, and determine the common defect category, common defect cause and common repair scheme corresponding to the common defect data , to generate a visual common defect statistics table.

In this embodiment, the preset time interval is a time period for regularly counting common defects, and the preset number of times threshold is a critical value for judging whether a defect is common, which can be flexibly set and changed according to actual needs, for example, a week, half a month and so on. The system is installed at a preset time interval, such as one month, and a visual report of common defects is generated every month. It can regularly analyze the classification of common defects, the cause of the problem, the number of recurrences, and the repair suggestions, so as to better identify the hidden quality risks of software projects. Targeted measures are taken to solve them, and some recurring defects of the same type are highlighted.

Further, by integrating the problem template and multi-level semantic analysis technology, the complexity of the model can be effectively reduced, and the defect data to be analyzed can be deeply understood, and the accuracy of the model can be improved; by marking the defect data for fuzzy matching with features identification, the feature defects in the knowledge base will be preferentially compared in the next fuzzy matching, so as to realize the self-adaptive positioning result calibration; The accumulation of defect repair experience forms closed-loop management, and continuously improves the efficiency of automatic defect location analysis based on the defect knowledge base; by regularly generating a statistical table of common defect problems, the representative results in the process of automatic defect location can be displayed on a regular basis , it is convenient to take targeted measures to solve and gradually avoid the occurrence of common problems.

Further, based on the first embodiment shown in FIG. 2 above, a third embodiment of the automatic defect location analysis method of the present invention is proposed. In this embodiment, the text similarity algorithm includes a cosine distance algorithm, and step S10 includes:

Using natural language processing technology to perform word segmentation processing on the defect data to be analyzed to extract keywords in the defect data to be analyzed;

In this embodiment, based on natural language processing technology, the given text (the text part in the defect data to be analyzed) is firstly segmented according to complete sentences, that is, for each sentence, word segmentation and part-of-speech tagging are performed, and filtering is performed. Remove stop words, keep only words with specified part of speech, such as nouns, verbs, etc., construct candidate keyword graph G=(V, E), where V is the node set, which is composed of candidate keywords; then according to TextRank (a key Word Extraction Algorithm) formula (TextRank is inspired by Google's PageRank, by dividing the text into several constituent units (words, sentences) and establishing a graph model, using the voting mechanism to sort the important components in the text, only using a single article The information of the document itself can realize keyword extraction), iteratively propagate the weight of each node until convergence. Finally, the node weights are sorted in reverse order to obtain the most important T words as candidate keywords.

generating a first word frequency vector set of the keyword in the defect data to be analyzed, and a second word frequency vector set of the keyword in the defect knowledge base;

A cosine similarity set between the first word frequency vector and the second word frequency vector is obtained by using a cosine distance algorithm, so as to accurately match the keyword in the defect knowledge base based on the cosine similarity set.

In this embodiment, the first word frequency vector set is the set of word frequency vectors of the currently extracted keywords in the defect data to be analyzed, and the second word frequency vector is the word frequency vectors of the currently extracted keywords in the defect knowledge base collection. Based on the cosine distance calculation principle, the system takes the currently extracted keywords as a keyword set, respectively calculates the word frequencies of the two parties for the keywords in this keyword set, and generates their respective word frequency vectors (the first word frequency vector and the second word frequency vector). word frequency vector); finally, calculate the cosine similarity of the two vector sets, and combine the cosine similarity values of each vector into a set. The greater the cosine similarity, the more similar. The system can take the problem cause and repair plan corresponding to the defect sample data with the highest similarity (higher than the preset similarity threshold) in the knowledge base as the current exact matching result.

Further, the cosine similarity set between the first word frequency vector and the second word frequency vector is obtained by using the cosine distance algorithm, so as to put the keyword in the defect knowledge base based on the cosine similarity set. After the exact matching steps in , it also includes:

Judging whether there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set;

If so, it is determined that the current exact match is successful;

If not, it is determined that the current exact match fails.

In this embodiment, the preset similarity threshold can be flexibly set according to actual requirements, which is not specifically limited in this embodiment. The system determines whether there is a target cosine similarity that exceeds the preset similarity threshold in the cosine similarity set, and if there is a target cosine similarity that exceeds the preset similarity threshold in the cosine similarity set, it is determined that the current exact match is successful; if the cosine similarity is similar If there is no target cosine similarity exceeding the preset similarity threshold in the degree set, that is, the cosine similarity of each word frequency vector is lower than the preset similarity threshold, it is determined that the current exact matching fails, and then the fuzzy matching can be turned.

Further, after step S10, it also includes:

When detecting that there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set, acquiring matching defect sample data corresponding to the target cosine similarity in the defect knowledge base;

Obtain the matching problem cause and matching repair scheme corresponding to the matching defect sample data, and display the matching problem cause and matching repair scheme in association.

In this embodiment, when detecting that there is a target cosine similarity in the cosine similarity set that exceeds the preset similarity threshold, the system directly obtains the matching defect sample data corresponding to the target cosine similarity in the defect knowledge base, and obtains the matching defect sample data. The cause of the matching problem and the matching repair scheme corresponding to the defect sample data in the knowledge base are directly displayed in the front end, so that users can view them directly.

The blockchain referred to in the present invention is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain, essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

Further, by using natural language processing technology and the principle of cosine distance to perform precise matching, the precise matching process is more accurate and efficient; by setting a similarity threshold to determine the success of precise matching, it makes the determination of precise matching results easier and more feasible. ; By correlating the matching results when the exact match is successful, users can view them directly, which improves the user experience.

In addition, as shown in FIG. 3 , in order to achieve the above purpose, the present invention also provides an automatic defect positioning and analysis device, and the automatic defect positioning and analysis device includes:

The exact matching module 10 is used for acquiring defect data to be analyzed, and extracting keywords in the defect data to be analyzed, so as to accurately match the keywords in a preset defect knowledge base based on a preset text similarity algorithm , wherein the defect knowledge base includes multi-category defect sample data, and defect causes and repair schemes corresponding to the multi-category defect sample data;

The fuzzy matching module 20 is used for identifying the defect category to which the defect data to be analyzed belongs when the current exact matching failure is detected, and searching the defect knowledge base for the target defect cause and target repair corresponding to the defect category scheme to complete the fuzzy matching of the defect data to be analyzed;

The defect calibration module 30 is configured to use the fuzzy matching defect data to be analyzed as the defect data to be verified, and when acquiring the accurate defect cause and accurate repair plan of the defect data to be verified, the accurate defect cause The defect knowledge base is updated with an accurate repair plan, so as to calibrate the defect data to be calibrated.

Optionally, the fuzzy matching module 10 includes:

a problem module acquiring unit, configured to use a preset entity recognition model to identify the entity information of the defect data to be analyzed, and obtain the problem template based on the entity information when it is detected that the current exact match fails;

a multi-level semantic parsing unit, configured to perform multi-level semantic parsing on the defect data to be analyzed to obtain the multi-level semantics of the defect data to be analyzed;

A defect category prediction unit, configured to use a preset probability graph model and combine the problem template and the multi-level semantics to predict the defect category in the defect knowledge base corresponding to the defect data to be analyzed;

The fuzzy matching completion unit is used to convert the defect data to be analyzed into a structured query of the defect knowledge base according to the defect category and the entity information, and obtain the target defect cause and target repair plan by querying Complete the fuzzy matching on the defect data to be analyzed.

Optionally, the automatic defect location analysis device further includes:

The feature marking module is used for feature marking the calibrated defect data to be calibrated as feature defect data;

The preferential matching module is configured to preferentially select the feature defect data for matching when it is detected that fuzzy matching is currently being performed.

Optionally, the automatic defect location analysis device further includes:

A screening and conversion module, used for acquiring defective sample data, and performing pre-screening and format conversion on the defective sample data to obtain target sample data;

a data classification module, configured to classify the target sample data based on a preset classification algorithm to obtain multi-category defect sample data corresponding to a plurality of defect categories;

The mapping establishment module is used for extracting and screening the multi-category defect sample data to obtain the defect cause and repair scheme corresponding to each defect category, and establish the mapping relationship between the defect category, defect cause and repair scheme information ;

The knowledge base building module is configured to build the defect knowledge base when it is detected that the data volume of the multi-category defect sample data reaches a preset data volume threshold, so as to perform automatic defect location analysis based on the defect knowledge base.

Optionally, the text similarity algorithm includes a cosine distance algorithm,

The exact matching module 20 includes:

A key extraction unit, used for performing word segmentation processing on the defect data to be analyzed by using natural language processing technology, so as to extract keywords in the defect data to be analyzed;

A word frequency generating unit, configured to generate a first word frequency vector set of the keyword in the defect data to be analyzed, and a second word frequency vector set of the keyword in the defect knowledge base;

An exact matching unit is used to obtain a cosine similarity set between the first word frequency vector and the second word frequency vector using a cosine distance algorithm, so as to classify the keyword in the defect knowledge based on the cosine similarity set. exact match in the library.

Optionally, the automatic defect location analysis device further includes:

an exact matching judgment module, used for judging whether there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set;

If so, it is determined that the current exact match is successful;

If not, it is determined that the current exact match fails.

Optionally, the defect automatic positioning analysis device further includes:

A matching sample acquisition module, configured to acquire a matching defect sample corresponding to the target cosine similarity in the defect knowledge base when detecting that there is a target cosine similarity exceeding a preset similarity threshold in the cosine similarity set data;

The result association display module is used for acquiring the matching problem cause and the matching repair scheme corresponding to the matching defect sample data, and displaying the matching problem cause and the matching repair scheme in association.

Optionally, the automatic defect location analysis device further includes:

The regular statistics module is used to regularly screen out the common defect data whose actual recurrence times exceed the preset times threshold from the defect knowledge base according to the preset time interval, and determine the common defect category, common defect category corresponding to the regular defect data, and common defect data. Defect causes and common fixes to generate a visual common defect statistics table.

The invention also provides a defect automatic positioning analysis device.

The automatic defect location analysis device includes a processor, a memory, and an automatic defect location analysis program stored on the memory and executable on the processor, wherein the automatic defect location analysis program is executed by the processor. , to realize the steps of the automatic defect location analysis method as described above.

Wherein, for the method implemented when the automatic defect location analysis program is executed, reference may be made to the various embodiments of the automatic defect location analysis method of the present invention, which will not be repeated here.

In addition, an embodiment of the present invention further provides a computer-readable storage medium.

The computer-readable storage medium of the present invention stores an automatic defect location analysis program, wherein when the defect automatic location analysis program is executed by the processor, the steps of the above-mentioned method for automatic defect location analysis are implemented.

The method implemented when the automatic defect location analysis program is executed may refer to the various embodiments of the automatic defect location analysis method of the present invention, which will not be repeated here.

It should be noted that, herein, the terms "comprising", "comprising" or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article or system comprising a series of elements includes not only those elements, It also includes other elements not expressly listed or inherent to such a process, method, article or system. Without further limitation, an element qualified by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article or system that includes the element.

The above-mentioned serial numbers of the embodiments of the present invention are only for description, and do not represent the advantages or disadvantages of the embodiments.

From the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform, and of course hardware can also be used, but in many cases the former is better implementation. Based on this understanding, the technical solutions of the present invention can be embodied in the form of software products in essence or the parts that make contributions to the prior art. The computer software products are stored in a storage medium (such as ROM/RAM) as described above. , magnetic disk, optical disk), including several instructions to make a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) execute the methods described in the various embodiments of the present invention.

The above are only preferred embodiments of the present invention, and are not intended to limit the scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields , are similarly included in the scope of patent protection of the present invention.

Claims (10)

1. An automatic defect positioning and analyzing method is characterized by comprising the following steps:

acquiring defect data to be analyzed, extracting keywords in the defect data to be analyzed, and accurately matching the keywords in a preset defect knowledge base based on a preset text similarity algorithm, wherein the defect knowledge base comprises multi-class defect sample data, and defect reasons and repair schemes corresponding to the multi-class defect sample data;

when the current accurate matching is detected to fail, identifying the defect type to which the defect data to be analyzed belongs, and searching a target defect reason and a target repair scheme corresponding to the defect type in the defect knowledge base to complete fuzzy matching of the defect data to be analyzed;

and taking the fuzzy-matched defect data to be analyzed as defect data to be verified, and updating the accurate defect reason and the accurate repair scheme into the defect knowledge base when the accurate defect reason and the accurate repair scheme of the defect data to be verified are obtained so as to calibrate the defect data to be calibrated.

2. The method according to claim 1, wherein the step of identifying the defect type to which the defect data to be analyzed belongs when detecting that the current exact match fails, and searching the defect knowledge base for a target defect cause and a target repair solution corresponding to the defect type to complete the fuzzy match of the defect data to be analyzed comprises:

when the current accurate matching is detected to fail, recognizing entity information of the defect data to be analyzed by using a preset entity recognition model, and obtaining the problem template based on the entity information;

performing multi-level semantic analysis on the defect data to be analyzed to obtain multi-level semantics of the defect data to be analyzed;

predicting the defect category of the defect data to be analyzed corresponding to the defect knowledge base by using a preset probability map model and combining the problem template and the multi-level semantics;

and converting the defect data to be analyzed into structured query of the defect knowledge base according to the defect category and the entity information, and querying to obtain the target defect reason and a target repair scheme so as to complete fuzzy matching of the defect data to be analyzed.

3. The method according to claim 1, wherein the step of using the fuzzy-matched defect data to be analyzed as defect data to be verified, and updating the accurate defect reason and the accurate repair plan into the defect knowledge base when acquiring the accurate defect reason and the accurate repair plan of the defect data to be verified, so as to calibrate the defect data to be calibrated further comprises:

performing characteristic marking on the calibrated defect data to be calibrated to serve as characteristic defect data;

and when the fuzzy matching is detected to be currently performed, preferentially selecting the characteristic defect data for matching.

4. The method for automatically locating and analyzing the defects according to claim 1, wherein before the step of obtaining the defect data to be analyzed and extracting the keywords from the defect data to be analyzed, the method further comprises:

acquiring defect sample data, and performing pre-screening and format conversion on the defect sample data to obtain target sample data;

classifying the target sample data based on a preset classification algorithm to obtain multi-class defect sample data corresponding to a plurality of defect classes;

extracting and screening the multi-class defect sample data to obtain defect reasons and repair schemes corresponding to the defect classes, and establishing a mapping relation among the defect classes, the defect reasons and the repair scheme information;

and when the data volume of the multi-class defect sample data is detected to reach a preset data volume threshold value, constructing the defect knowledge base so as to perform automatic defect positioning analysis based on the defect knowledge base.

5. The method of automatic defect localization analysis according to claim 1, wherein said text similarity algorithm comprises a cosine distance algorithm,

the step of obtaining the defect data to be analyzed, extracting the keywords in the defect data to be analyzed, and accurately matching the keywords in a preset defect knowledge base based on a preset text similarity algorithm comprises the following steps:

performing word segmentation processing on the to-be-analyzed defect data by using a natural language processing technology to extract keywords in the to-be-analyzed defect data;

generating a first word frequency vector set of the keywords in the defect data to be analyzed and a second word frequency vector set of the keywords in the defect knowledge base;

and a cosine distance algorithm is used for obtaining a cosine similarity set between the first word frequency vector and the second word frequency vector, so that the keywords are accurately matched in the defect knowledge base based on the cosine similarity set.

6. The method of claim 5, wherein after the step of obtaining the cosine similarity set between the first word frequency vector and the second word frequency vector by using a cosine distance algorithm to precisely match the keyword in the defect knowledge base based on the cosine similarity set, the method further comprises:

judging whether a target cosine similarity exceeding a preset similarity threshold exists in the cosine similarity set or not;

if so, judging that the current accurate matching is successful;

if not, judging that the current accurate matching fails.

7. The method for automatically locating and analyzing defects according to claim 6, wherein after the steps of obtaining the data of the defects to be analyzed and extracting the keywords from the data of the defects to be analyzed to precisely match the keywords in a preset defect knowledge base based on a preset text similarity algorithm, the method further comprises:

when the cosine similarity set is detected to have target cosine similarity exceeding a preset similarity threshold, acquiring matched defect sample data corresponding to the target cosine similarity in the defect knowledge base;

and acquiring a matching problem reason and a matching repair scheme corresponding to the matching defect sample data, and displaying the matching problem reason and the matching repair scheme in a correlation manner.

8. The method according to any one of claims 1 to 7, wherein the step of using the fuzzy-matched defect data to be analyzed as defect data to be verified, and updating the accurate defect reason and the accurate repair solution into the defect knowledge base when acquiring the accurate defect reason and the accurate repair solution of the defect data to be verified so as to calibrate the defect data to be calibrated further comprises:

and regularly screening common defect data with the actual reproduction times exceeding a preset time threshold from the defect knowledge base according to a preset time interval, and determining common defect types, common defect reasons and common repair schemes corresponding to the common defect data to generate a visual common defect statistical table.

9. A defect automatic localization analysis device, characterized in that the defect automatic localization analysis device comprises a processor, a memory, and a defect automatic localization analysis program stored on the memory and executable by the processor, wherein the defect automatic localization analysis program when executed by the processor implements the steps of the defect automatic localization analysis method according to any one of claims 1 to 8.

10. A computer-readable storage medium, wherein a defect automatic location analysis program is stored on the computer-readable storage medium, wherein the defect automatic location analysis program, when executed by a processor, implements the steps of the defect automatic location analysis method according to any one of claims 1 to 8.

Images