CN109376092B - Automatic analysis method for software defect reasons for defect patch codes - Google Patents
Automatic analysis method for software defect reasons for defect patch codes Download PDFInfo
- Publication number
- CN109376092B CN109376092B CN201811412560.2A CN201811412560A CN109376092B CN 109376092 B CN109376092 B CN 109376092B CN 201811412560 A CN201811412560 A CN 201811412560A CN 109376092 B CN109376092 B CN 109376092B
- Authority
- CN
- China
- Prior art keywords
- repair
- defect
- code
- tree
- code element
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 230000007547 defect Effects 0.000 title claims abstract description 147
- 238000004458 analytical method Methods 0.000 title claims abstract description 9
- 230000008439 repair process Effects 0.000 claims abstract description 125
- 239000013598 vector Substances 0.000 claims abstract description 43
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 12
- 238000013145 classification model Methods 0.000 claims abstract description 9
- 238000012360 testing method Methods 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000012986 modification Methods 0.000 claims description 3
- 230000004048 modification Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000004931 aggregating effect Effects 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 claims description 2
- 150000001875 compounds Chemical class 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/36—Prevention of errors by analysis, debugging or testing of software
- G06F11/3604—Analysis of software for verifying properties of programs
- G06F11/3608—Analysis of software for verifying properties of programs using formal methods, e.g. model checking, abstract interpretation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a software defect reason automatic analysis method facing to a defect patch code, which comprises the following steps: firstly, extracting a defect report and a repair code file of repaired defects from a defect library to obtain a defect data set; then, classifying and marking the defect data set according to the defect generation reason category; dividing a code block related to repair from the repair code file by using a code distinguishing tool; then extracting code element terms and repair modes from the code block; then constructing a repair tree of the defect repair code; then, converting the repair tree into a vector form; and finally, automatically learning and acquiring a defect reason classification model through a convolutional neural network model. The method of the invention represents the repair codes of the defects from a plurality of angles, makes full use of the concept knowledge contained in the repair codes, and links the representation with the defect reason classification to obtain the automatic classification model of the defect reasons, thereby being capable of more effectively assisting developers to position and repair the defects.
Description
Technical Field
The invention belongs to the field of software maintenance, and particularly relates to a method for automatically analyzing software defect reasons for defect patch codes.
Background
With the increase of the scale of software projects and the continuous evolution of software products, the number of software defects is also increased greatly. The traditional manual method for analyzing the defect reasons, positioning and repairing the defects cannot meet the requirement of quickly repairing the defects. A large amount of repair code of repaired defects is stored in a defect library of an open-source software defect tracking system, and the repair code implies various conceptual knowledge related to the defects, such as defect repair modes. Therefore, analyzing and characterizing the defect repair code may effectively assist developers in quickly understanding the cause of the defect, and locating and repairing the defect.
Current techniques for defect code analysis and description fall into two general categories: one class resembles the code to natural language and uses information retrieval based techniques to calculate text similarity of the code vocabulary and the defect report natural language vocabulary to aid in automatic defect localization. However, these methods have two problems, the first is the vocabulary mismatch problem: the terms used to describe defects in defect reports are different from the terms and code words used in the source file; the second is that a large amount of specific conceptual knowledge, such as code element type and code structure information, contained by the source code is not considered. The other type is focusing on the structure of the code, the change of the code on the tree structure is extracted by adopting a difference algorithm based on the tree structure, a repair example is obtained from historical defect repair, and automatic defect repair is assisted. Such approaches tend to favor the study of the code itself, ignoring the link between the defect and the repair code, such as the correlation of the cause of the defect to the mode of code repair. And the method is generally used for extracting a specific type of defect repair mode and has poor universality.
Disclosure of Invention
The invention solves the technical problem of describing software defect repair codes from multiple angles and developing a multivariate representation method of the software defect repair codes and an automatic defect classification method.
The technical solution for realizing the purpose of the invention is as follows: a software defect reason automatic analysis method facing defect patch codes comprises the following steps:
step 1, extracting a defect report and a repair code file of repaired defects from a defect library to obtain a defect data set;
step 2, self-defining the defect generation reason category, and classifying and marking the defect data set according to the defect generation reason category;
step 3, dividing a code block related to repair from the repair code file obtained in the step 1 by using a code distinguishing tool;
step 4, extracting code element terms and a repair mode from the code block obtained in the step 3;
step 5, constructing a repair tree of the defect repair code according to the code element terms and the repair mode obtained in the step 4;
step 6, converting the constructed repair tree of the defect repair codes into a vector form to obtain vector representation of the repair codes;
and 7, taking the vector representation of the repair code obtained in the step 6 and the classification label marked in the step 2 as the input of a convolutional neural network model, and automatically learning to obtain a defect reason classification model so as to finish automatic analysis of software defect reasons.
Compared with the prior art, the invention has the following remarkable advantages: 1) the method uses a code element term, a code element concept type and a code repair operation from multiple angles to comprehensively express the defect repair code, and the expression simultaneously comprises text vocabulary semantic features, concept type semantic features and structure semantic features; 2) the method utilizes the corresponding relation of the code repair mode and the defect reason classification on the AST hierarchical structure to closely connect the defect code repair mode and the defect reason classification; 3) the invention adopts a representation learning method based on a tree structure to represent the repair tree of the defect repair code, the representation method comprises the multi-angle repair code characteristics, the representation is used as the input of a deep learning model, the defect reason classifier is obtained by utilizing the advantage that the deep learning model can automatically learn the data characteristics, and the developer can be effectively assisted to search the defect generation reason.
The present invention is described in further detail below with reference to the attached drawing figures.
Drawings
FIG. 1 is a flowchart of an automatic analysis method for software defect reasons of defect patch codes according to the present invention.
FIG. 2 is a defect cause distribution diagram of the Mozilla project and Radare2 projects extracted according to an embodiment of the present invention.
FIG. 3 is a screenshot of a differentiated code before and after defect repair extracted in an embodiment of the present invention.
Fig. 4 is a schematic diagram of a repair tree of an AST structure according to an embodiment of the present invention.
Detailed Description
With reference to fig. 1, the present invention provides a method for automatically analyzing software defect reasons for defect patch codes, which comprises the following steps:
step 1, extracting a defect report and a repair code file of repaired defects from a defect library to obtain a defect data set.
And 2, self-defining the defect generation reason category, and classifying and marking the defects according to the defect generation reason category to obtain a code data set of the defect generation reason category. The generation reason types of the custom defects are shown in the following table 1:
TABLE 1 Defect cause Classification Table used in the present invention
And 3, dividing a code block related to repair from the repair code file obtained in the step 1 by using a code distinguishing tool. When the code blocks related to the modification are divided, the division granularity is a complete AST tree structure.
Step 4, extracting code element terms and repair modes from the code block obtained in step 3, specifically:
step 4-1, extracting code element terms from the code block acquired in the step 3 by using a predefined rule;
wherein, the code element terms, namely code text words, comprise identifiers, literal amounts and operators;
the predefined rules are: removing keywords of a programming language and common symbols except operators;
4-2, extracting a repair mode from the code block obtained in the step 3 by using an extraction technology based on an abstract syntax tree and combining a code distinguishing tool;
wherein the repair mode comprises a code element concept type and a repair operation; the conceptual type of the code element is the role of the code element in the code, and the repair operation is a modification operation on the code element.
Step 5, constructing a repair tree of the defect repair code according to the code element terms and the repair mode obtained in the step 4, which specifically comprises the following steps:
and constructing a repair tree based on the abstract syntax tree, wherein each node is a repair to one abstract syntax tree node and comprises the code element term, the code element concept type and the repair operation to the code element term and the code element concept type which are related to the node.
Step 6, converting the constructed repair tree of the defect repair codes into a vector form to obtain vector representation of the repair codes, wherein the method adopted for converting the repair tree into the vector form is a representation learning method based on a tree structure, and specifically comprises the following steps:
step 6-1, the leaf nodes of the repair tree are represented by direct vectors thereof, and the non-leaf nodes are represented by combined vectors;
the combination vector is a linear combination of a direct vector of a non-leaf node and a sub-node representation of the direct vector, wherein the sub-node is a leaf node or a non-leaf node;
let c1,…,cnFor children nodes other than the leaf node p, the combined vector of p is represented as:
p＝Wcomb1·vec(p)+Wcomb2·tanh(∑liWcoｄｅ,i·vec(ci)+bcode)
where vec (-) is a direct vector of nodes,
is node ciThe corresponding weight matrix is used to determine the weight,
for deviation,. liIs a weight coefficient;
are all combination parameters, are in [0,1 ]]Any value within the range;
a real number matrix;
the direct vector is a linear combination of related code element items, concept types of the elements and direct vectors of the repair actions, and the direct vector of a certain node f is expressed as:
f＝Wfe·Vec(fe)+Wft·Vec(ft)+Wfa·Vec(fa)
in the formula, Vec (f)e)、Vec(ft)、Vec(fa) Direct vectors of code element terms, code element concept types, repair operations, respectively;
are all combination parameters, are in [0,1 ]]Any value within the range, and Wfe+Wft+Wfa＝1；
As can be seen from the above, each node of the repair tree is represented as a distributed real-valued vector
Step 6-2, in the neural network based on the tree structure, applying a characteristic detector to slide on the repair tree to detect and extract the characteristics in the representation method in the step 6-1, and assuming that the repair tree has n nodes, the output of the characteristic detector is as follows:
in the formula (I), the compound is shown in the specification,
Ncis the number of feature detectors.
And 7, taking the vector representation of the repair code obtained in the step 6 and the classification label marked in the step 2 as the input of a convolutional neural network model, and automatically learning to obtain a defect reason classification model, which specifically comprises the following steps:
7-1, aggregating all the features extracted by the feature detector by using a standard dynamic pool layer of a unidirectional pool;
7-2, dividing the defect data set in the step 1 into a training set and a testing set;
and 7-3, taking the aggregated features as the input of an output layer of the convolutional neural network model, namely taking the vector representation of the repair code obtained in the step 6 and the classification label marked in the step 2 as the input of the convolutional neural network model, verifying the training set by using the convolutional neural network model, then performing cross verification on the test set to obtain an automatic defect classifier, and finishing automatic classification of the historical defects according to the generation reasons.
The present invention will be described in further detail with reference to examples.
Examples
With reference to fig. 1, the method for automatically analyzing software defect reasons for defect patch codes of the present invention includes the following steps:
step 1, in this embodiment, Mozilla and Radare2 defect libraries are used as data, and in consideration of the accuracy of defect data, a keyword matching search and crawler technology is adopted to screen out historical defects in fixed (repaired) and verified states, and front and rear code segments in submission information of the historical defects are extracted to serve as a data set of an experiment. FIG. 3 is a schematic diagram of the extracted distinguishing codes before and after repair of one of the defects.
And 2, defining the generation reason type of the defects as shown in the table 1. And (3) dividing the defect data set obtained in the step (1) into a training set, a verification set and a test set according to the proportion of the number of the defects to be 3:1: 1. And manually analyzing the defect report and the repair code in the data set, and carrying out reason classification and marking on each defect in the data set according to the defined defect generation reason. Wherein, the classification of the defect reason is labeled by using 19 sub-types of the classification of the defect reason. FIG. 2 is a diagram of the defect cause number distribution of the Mozilla item and Radare2 items extracted in this embodiment.
Step 3, extracting the code segments before and after repair in the submission information of the repaired defects from the repair code file obtained in the step 1 by using a code distinguishing tool, and dividing the repair code segments into code blocks with independent AST structures; taking the defect repair code shown in fig. 3 as an example, one of the repair-related if statements: if (0 ═ strlen (phones)) { ckd _ free (phones); return 0; is a code block with independent AST structure.
Step 4, extracting the code element terms of the code block obtained in the step 3 by using a predefined rule, and extracting the concept type containing the code elements and the repair mode of the code block obtained in the step 3 by using an extraction technology based on an Abstract Syntax Tree (AST);
and 4-1, extracting the code element terms of the code block obtained in the step 3 by using a predefined rule. Taking the defect repair code shown in fig. 3 and the if statement as an example, the common symbols and the redundant spaces are removed, and the code element words are extracted to obtain the code element terms of each defect. The code element terms in this embodiment include: '0', 'strlen', 'lights', and 'ckd _ free'.
And 4-2, extracting the concept type of the code elements and the repairing operation at the AST level by means of an AST extraction-based repairing mode extracting technology. The extracted code element concept type and repair operation in this embodiment include: "If Insert", "Condition Insert", "If-body Insert", "live Insert", "FunCall Insert", "Return Insert", "Name Insert", "Args Insert" and "Var Insert".
And 5, combining the code element terms, the code element concept types and the repair operation of each defect obtained in the step 4 to construct a repair tree. The defect repair code as in this embodiment indicates that if statement blocks containing jump statements are added, and the schematic diagram of the repair tree thereof is shown in fig. 4. For visual representation, the code element terms of each node that are actually extracted are omitted in fig. 4. As can be seen from FIG. 4, each node of the repair tree is a meta-repair pattern, e.g., f1＝(If Insert,*),f2＝(Condtn Insert,f1),f3＝(If-body Insert,f1),f4＝(Literal Insert,f2),f5＝(FunCall Insert,f2),f6＝(FunCall Insert,f3),and f7＝(Return Insert,f3) And so on. The above formalized representation of each node is fi＝(eti,pti) In which etiIs the repair of the current node, ptiIs the repair of its parent node. It can be seen that f1And f2And f3Respectively parent-child relationship, otherwise similar, that is, they are nested on the AST structure.
And 6, converting the repair tree constructed in the step 5 into a vector form by a representation learning method based on a tree structure to obtain vector representation of the repair code.
And 7, using the vector representation set of the repair codes obtained in the step 6 and the artificial classification labels marked in the step 2 as the input of the convolutional neural network model, and automatically learning to obtain a defect reason classification model. For example, the classification result obtained by using the classification model is as follows: 0302 the extreme conditions are ignored (see table 1 above).
The method of the invention represents the repair codes of the defects from a plurality of angles, makes full use of the concept knowledge contained in the repair codes, and links the representation with the defect reason classification to obtain the automatic classification model of the defect reasons, thereby being capable of more effectively assisting developers to position and repair the defects.
Claims (7)
1. A software defect reason automatic analysis method facing defect patch codes is characterized by comprising the following steps:
step 1, extracting a defect report and a repair code file of repaired defects from a defect library to obtain a defect data set;
step 2, self-defining the defect generation reason category, and classifying and marking the defect data set according to the defect generation reason category;
step 3, dividing a code block related to repair from the repair code file obtained in the step 1 by using a code distinguishing tool;
step 4, extracting code element terms and a repair mode from the code block obtained in the step 3;
step 5, constructing a repair tree of the defect repair code according to the code element terms and the repair mode obtained in the step 4;
step 6, converting the constructed repair tree of the defect repair codes into a vector form to obtain vector representation of the repair codes;
and 7, taking the vector representation of the repair code obtained in the step 6 and the classification label marked in the step 2 as the input of a convolutional neural network model, and automatically learning to obtain a defect reason classification model so as to finish automatic analysis of software defect reasons.
2. The method for automatically analyzing software defect reasons for defect patch codes as claimed in claim 1, wherein the categories of the causes of the custom defects in step 2 are shown in the following table 1:
TABLE 1 Defect cause Classification
3. The method for automatically analyzing software defect reasons facing defect patch codes as claimed in claim 1 or 2, wherein the step 3 divides the code blocks related to repair with a division granularity of a complete AST tree structure.
4. The method for automatically analyzing software defect causes for defect patch codes as claimed in claim 1, wherein the step 4 is to extract code element terms and repair modes from the code block obtained in the step 3, specifically:
step 4-1, extracting code element terms from the code block acquired in the step 3 by using a predefined rule;
wherein, the code element terms, namely code text words, comprise identifiers, literal amounts and operators;
the predefined rules are: removing keywords of a programming language and common symbols except operators;
4-2, extracting a repair mode from the code block obtained in the step 3 by using an extraction technology based on an abstract syntax tree and combining a code distinguishing tool;
wherein the repair mode comprises a code element concept type and a repair operation; the conceptual type of the code element is the role of the code element in the code, and the repair operation is a modification operation on the code element.
5. The method for automatically analyzing software defect causes for defect patch codes as claimed in claim 4, wherein the step 5 of constructing the repair tree of the defect repair code according to the code element terms and the repair mode obtained in the step 4 specifically comprises:
and constructing a repair tree based on the abstract syntax tree, wherein each node is a repair to one abstract syntax tree node and comprises the code element term, the code element concept type and the repair operation to the code element term and the code element concept type which are related to the node.
6. The method for automatically analyzing software defect causes of defect patch codes as claimed in claim 5, wherein the step 6 is to convert the constructed repair tree of the defect repair codes into a vector form to obtain the vector representation of the repair codes, and the method adopted is as follows: the representation learning method based on the tree structure specifically comprises the following steps:
step 6-1, the leaf nodes of the repair tree are represented by direct vectors thereof, and the non-leaf nodes are represented by combined vectors;
the combination vector is a linear combination of a direct vector of a non-leaf node and a sub-node representation of the direct vector, wherein the sub-node is a leaf node or a non-leaf node;
let c1，...，cnFor children nodes other than the leaf node p, the combined vector of p is represented as:
p＝Wcomb1·vec(p)+Wcomb2·tanh(∑liWcode，i·vec(ci)+bcode)
where vec (-) is a direct vector of nodes,
is node ciThe corresponding weight matrix is used to determine the weight,
for deviation,. liIs a weight coefficient;
are all combination parameters, are in [0,1 ]]Any value within the range;
a real number matrix;
the direct vector is a linear combination of code element terms, code element concept types and direct vectors of repair operations, and the direct vector of a certain node f is represented as:
in the formula, Vec (f)e)、Vec(ft)、Vec(fa) Direct vectors of code element terms, code element concept types, repair operations, respectively;
are all combination parameters, are in [0,1 ]]Any value within the range, and
as can be seen from the above, each node of the repair tree is represented as a distributed real-valued vector
Step 6-2, in the neural network based on the tree structure, applying a characteristic detector to slide on the repair tree to detect and extract the characteristics in the representation method in the step 6-1, and assuming that the repair tree has n nodes, the output of the characteristic detector is as follows:
in the formula (I), the compound is shown in the specification,
Ncis the number of feature detectors.
7. The method for automatically analyzing software defect causes for defect patch codes as claimed in claim 6, wherein the step 7 takes the vector representation of the repair code obtained in the step 6 and the classification label marked in the step 2 as the input of the convolutional neural network model, and automatically learns and obtains the cause classification model of the defect, specifically:
7-1, aggregating all the features extracted by the feature detector by using a standard dynamic pool layer of a unidirectional pool;
7-2, dividing the defect data set obtained in the step 1 into a training set and a testing set;
and 7-3, taking the aggregated features as the input of an output layer of the convolutional neural network model, namely taking the vector representation of the repair code obtained in the step 6 and the classification label marked in the step 2 as the input of the convolutional neural network model, verifying the training set by using the convolutional neural network model, then performing cross verification on the test set to obtain an automatic defect classifier, and finishing automatic classification of the historical defects according to the generation reasons.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811412560.2A CN109376092B (en) | 2018-11-26 | 2018-11-26 | Automatic analysis method for software defect reasons for defect patch codes |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811412560.2A CN109376092B (en) | 2018-11-26 | 2018-11-26 | Automatic analysis method for software defect reasons for defect patch codes |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109376092A CN109376092A (en) | 2019-02-22 |
| CN109376092B true CN109376092B (en) | 2021-07-13 |
Family
ID=65383010
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811412560.2A Active CN109376092B (en) | 2018-11-26 | 2018-11-26 | Automatic analysis method for software defect reasons for defect patch codes |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109376092B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110162245B (en) * | 2019-04-11 | 2020-12-08 | 北京达佳互联信息技术有限公司 | Analysis method and device of graphic operation, electronic equipment and storage medium |
| CN110221933B (en) * | 2019-05-05 | 2023-07-21 | 北京百度网讯科技有限公司 | Code defect auxiliary repairing method and system |
| CN110188047B (en) * | 2019-06-20 | 2023-04-18 | 重庆大学 | Double-channel convolutional neural network-based repeated defect report detection method |
| CN110349477B (en) * | 2019-07-16 | 2022-01-07 | 长沙酷得网络科技有限公司 | Programming error repairing method, system and server based on historical learning behaviors |
| CN111240978A (en) * | 2020-01-09 | 2020-06-05 | 上海丰蕾信息科技有限公司 | Data report generation and analysis method |
| CN111177016B (en) * | 2020-04-14 | 2020-08-21 | 四川新网银行股份有限公司 | Software test defect management method |
| CN112069069A (en) * | 2020-09-03 | 2020-12-11 | 平安信托有限责任公司 | Defect automatic positioning analysis method, device and readable storage medium |
| CN112181428B (en) * | 2020-09-28 | 2021-10-22 | 北京航空航天大学 | Abstract syntax tree-based open-source software defect data classification method and system |
| CN112860362B (en) * | 2021-02-05 | 2022-10-04 | 达而观数据(成都)有限公司 | Visual debugging method and system for robot automation process |
| CN113010420B (en) * | 2021-03-05 | 2023-05-30 | 南方科技大学 | Method and terminal equipment for promoting co-evolution of test codes and product codes |
| WO2023279254A1 (en) * | 2021-07-06 | 2023-01-12 | Huawei Technologies Co.,Ltd. | Systems and methods for detection of software vulnerability fix |
| CN115934147A (en) * | 2021-08-06 | 2023-04-07 | 中兴通讯股份有限公司 | Automatic software restoration method and system, electronic equipment and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105159822B (en) * | 2015-08-12 | 2017-08-29 | 南京航空航天大学 | A kind of software defect positioning method based on text part of speech and routine call relation |
| CN108829607B (en) * | 2018-07-09 | 2021-08-10 | 华南理工大学 | Software defect prediction method based on convolutional neural network |
-
2018
- 2018-11-26 CN CN201811412560.2A patent/CN109376092B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109376092A (en) | 2019-02-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109376092B (en) | Automatic analysis method for software defect reasons for defect patch codes | |
| CN110597735B (en) | Software defect prediction method for open-source software defect feature deep learning | |
| CN110134757B (en) | Event argument role extraction method based on multi-head attention mechanism | |
| CN112214610B (en) | Entity relationship joint extraction method based on span and knowledge enhancement | |
| CN110502361A (en) | Fine granularity defect positioning method towards bug report | |
| CN106407113B (en) | A kind of bug localization method based on the library Stack Overflow and commit | |
| CN109977205B (en) | Method for computer to independently learn source code | |
| CN113254507B (en) | Intelligent construction and inventory method for data asset directory | |
| CN111427775B (en) | Method level defect positioning method based on Bert model | |
| CN111309607B (en) | Software defect positioning method of code method level | |
| CN113138920B (en) | Software defect report allocation method and device based on knowledge graph and semantic role labeling | |
| Ke et al. | Interpretable test case recommendation based on knowledge graph | |
| CN109858550B (en) | Machine identification method for potential process failure mode | |
| Simanjuntak | Proposed framework for automatic grading system of ER diagram | |
| CN103559510B (en) | Method for recognizing social group behaviors through related topic model | |
| CN111723021B (en) | Defect report automatic allocation method based on knowledge base and representation learning | |
| CN112329440A (en) | Relation extraction method and device based on two-stage screening and classification | |
| CN115408506B (en) | NL2SQL method combining semantic analysis and semantic component matching | |
| CN114462387B (en) | Sentence pattern automatic discrimination method under no-label corpus | |
| CN110807096A (en) | Information pair matching method and system on small sample set | |
| Dau et al. | Bootstrapping code-text pretrained language model to detect inconsistency between code and comment | |
| Mz et al. | Development of Software Cost Estimation and Resource Allocation Using Natural Language Processing, Cosine Similarity and Function Point | |
| CN114372148A (en) | Data processing method based on knowledge graph technology and terminal equipment | |
| CN109299381B (en) | Software defect retrieval and analysis system and method based on semantic concept | |
| Zhang et al. | Research on Defect Location Method of C Language Code Based on Deep Learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |