Systems and methods for classification of software defect reports

Patent metadata

Field	Value
Publication number	US-10664696-B2
Application number	US-201815934855-A
Country	US
Kind code	B2
Filing date	Mar 23, 2018
Priority date	Apr 19, 2017
Publication date	May 26, 2020
Grant date	May 26, 2020

Abstract

Existing software defect text categorization approaches are based on use of supervised/semi-supervised machine learning techniques, which may require significant amount of labeled training data for each class in order to train the classifier model leading to significant amount of human effort, resulting in an expensive process. Embodiments of the present disclosure provide systems and methods for circumventing the problem of dependency on labeled training data and features derived from source code by performing concept based classification of software defect reports. In the present disclosure, semantic similarity between the defect category/type labels and the software defect report(s) is computed and represented in a concept space spanned by corpus of documents obtained from one or more knowledge bases, and distribution of similarity values are obtained. These similarity values are compared with a dynamically generated threshold, and based on the comparison, the software defect reports are classified into software defect categories.

First claim

What is claimed is:

1. A processor implemented method, comprising:
obtaining, by one or more hardware processors, input data comprising (a) one or more software defect reports; (b) one or more software defect categories, each software defect category from the one or more software defect categories comprising a class label and associated textual description thereof, and (c) a corpus of documents ( 302 );
segmenting, by the one or more hardware processors, input text of the one or more software defect reports into one or more segments based on the input data, each of the one or more segments comprises text content ( 304 );
performing analysis on the corpus of documents obtained from one or more knowledge bases to identify a subset of relevant documents from the corpus of documents ( 306 );
generating a concept space based on the identified subset of relevant documents ( 308 );
projecting into the concept-space, by the one or more hardware processors, text content of the one or more segments pertaining to at least one of the one or more software defect reports and the textual description of the one or more software defect categories to generate a concept-space representation for each of the one or more software defect reports and the one or more software defect categories ( 310 );
computing, by the one or more hardware processors, one or more similarities between the concept-space representation of each of the one or more software defect reports and each of the one or more software defect categories to obtain distribution of one or more similarity values specific to the one or more software defect reports to be classified ( 312 );
performing, by the one or more hardware processors, a comparison of distribution of the one or more similarity values with a dynamically generated threshold ( 314 ); and
classifying by the one or more hardware processors, the one or more software defect reports into the one or more software defect categories based on the comparison ( 316 ).

2. The processor implemented method as claimed in claim 1 , wherein the step of performing analysis comprises applying one or more document identification techniques on the corpus of documents obtained from the one or more knowledge bases to identify the subset of relevant documents.

3. The processor implemented method as claimed in claim 2 , wherein the one or more document identification techniques comprises at least one of one or more graph-theoretic analysis, one or more keyword identification techniques and one or more text clustering techniques.

4. The processor implemented method as claimed in claim 1 , wherein when the one or more similarity values are higher than the dynamically generated threshold, the one or more software defect reports are classified into the one or more software defect categories.

5. The processor implemented method as claimed in claim 1 , wherein the dynamically generated threshold is based on the distribution of the one or more similarity values.

6. A system ( 100 ) comprising:
a memory ( 102 ) storing instructions and one or more modules ( 108 );
one or more communication interfaces ( 106 ); and
one or more hardware processors ( 104 ) coupled to the memory ( 102 ) via the one or more communication interfaces ( 106 ), wherein the one or more hardware processors ( 104 ) are configured by the instructions to execute the one or more modules ( 108 ) comprising:
an input reader module ( 202 ) that is configured to: obtain input data comprising (a) one or more software defect reports; (b) each software defect category from the one or more software defect categories comprising a class label and associated textual description thereof, and (c) a corpus of documents;
a software defect report text segmentation module ( 204 ) that is configured segment input text of the one or more software defect reports into one or more segments based on the input data, each of the one or more segments comprises text content;
a concept-space creation module ( 206 ) that is configured to:
perform analysis on the corpus of documents obtained from one or more knowledge bases to identify a subset of relevant documents from the corpus of documents, and
generate a concept space based on the identified subset of relevant documents;
a projection module ( 208 ) that is configured to project into the concept-space, text content of the one or more segments pertaining to at least one of the one or more software defect reports and the textual description of the one or more software defect categories to generate a concept-space representation for each of the one or more software defect reports and the one or more software defect categories;
a concept-space similarity computation module ( 210 ) that is configured to compute one or more similarities between the concept-space representation of each of the one or more software defect reports and each of the one or more software defect categories to obtain distribution of one or more similarity values specific to the one or more software defect reports to be classified; and
a software defect classification module ( 212 ) that is configured to:
perform a comparison of distribution of the one or more similarity values with a dynamically generated threshold, and
classify the one or more software defect reports into the one or more software defect categories based on the comparison.

7. The system as claimed in claim 6 , wherein the concept-space creation module ( 206 ) performs the analysis by applying one or more document identification techniques on the corpus of documents obtained from the one or more knowledge bases to identify the subset of relevant documents.

8. The system as claimed in claim 7 , wherein the one or more document identification techniques comprises at least one of one or more graph-theoretic analysis, one or more keyword identification techniques and one or more text clustering techniques.

9. The system as claimed in claim 6 , wherein when the one or more similarity values are higher than the dynamically generated threshold, the one or more software defect reports are classified into the one or more software defect categories.

10. The system as claimed in claim 6 , wherein the dynamically generated threshold is based on the distribution of the one or more similarity values.

11. One or more non-transitory machine readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors causes:
obtaining, by the one or more hardware processors, input data comprising (a) one or more software defect reports; (b) one or more software defect categories, each software defect category from the one or more software defect categories comprising a class label and associated textual description thereof, and (c) a corpus of documents;
segmenting, by the one or more hardware processors, input text of the one or more software defect reports into one or more segments based on the input data, each of the one or more segments comprises text content;
performing analysis on the corpus of documents obtained from one or more knowledge bases to identify a subset of relevant documents from the corpus of documents;
generating a concept space based on the identified subset of relevant documents;
projecting into the concept-space, by the one or more hardware processors, text content of the one or more segments pertaining to at least one of the one or more software defect reports and the textual description of the one or more software defect categories to generate a concept-space representation for each of the one or more software defect reports and the one or more software defect categories;
computing, by the one or more hardware proce

Assignees

• Tata Consultancy Services Ltd

Inventors

• Patil Sangameshwar Suryakant

Classifications

• G06F16/355

  Creation or modification of classes or clusters · CPC title

• G06F16/93

  Document management systems · CPC title

• G06F40/289

  Phrasal analysis, e.g. finite state techniques or chunking · CPC title

• G06K9/6224

  Physics · mapped topic

• G06F40/30 Primary

  Semantic analysis · CPC title