Compositional Vector Space Models for Improved Bug Localization

Wang, Shaowei; Lo, David; Lawall, Julia

doi:10.1109/icsme.2014.39

Cited by 59 publications

(28 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hit@10 only considers the performance of a fault localization tool when a small number of program elements (in our case: program blocks) are inspected. The measure was previously used by information retrieval (IR) based bug localization studies that find buggy program files given a textual bug report [31,35,38,40,43,51]. Admittedly, there is no large scale study that shows positive correlation (or its absence) between improving Hit@10 (or other rank-based metrics) and time and effort saved when developers use a fault localization technique to debug various kinds of bugs.…”

Section: Discussionmentioning

confidence: 99%

Fusion fault localizers

Lucia

Xia

2014

Proceedings of the 29th ACM/IEEE International Conference on Automated Software Engineering

Self Cite

View full text Add to dashboard Cite

Many spectrum-based fault localization techniques have been proposed to measure how likely each program element is the root cause of a program failure. For various bugs, the best technique to localize the bugs may differ due to the characteristics of the buggy programs and their program spectra. In this paper, we leverage the diversity of existing spectrum-based fault localization techniques to better localize bugs using data fusion methods. Our proposed approach consists of three steps: score normalization, technique selection, and data fusion. We investigate two score normalization methods, two technique selection methods, and five data fusion methods resulting in twenty variants of Fusion Localizer. Our approach is bug specific in which the set of techniques to be fused are adaptively selected for each buggy program based on its spectra. Also, it requires no training data, i.e., execution traces of the past buggy programs.We evaluate our approach on a common benchmark dataset and a dataset consisting of real bugs from three medium to large programs. Our evaluation demonstrates that our approach can significantly improve the effectiveness of existing state-of-the-art fault localization techniques. Compared to these state-of-the-art techniques, the best variants of Fusion Localizer can statistically significantly reduce the amount of code to be inspected to find all bugs. Our best variants can increase the proportion of bugs localized when developers only inspect the top 10% most suspicious program elements by more than 10% and increase the number of bugs that can be successfully localized when developers only inspect up to 10 program blocks by more than 20%.

show abstract

Section: Discussionmentioning

confidence: 99%

Fusion fault localizers

Lucia

Xia

2014

Proceedings of the 29th ACM/IEEE International Conference on Automated Software Engineering

Self Cite

View full text Add to dashboard Cite

show abstract

“…Besides language models, many other text analysis techniques, e.g., information retrieval, topic modeling, etc., have been used by researchers to analyze software and its related textual artifacts [3], [7], [22], [24], [25]. We highlight a few software engineering studies that make use of text analysis below.…”

Section: Text Analysis and Software Engineeringmentioning

confidence: 99%

NIRMAL: Automatic identification of software relevant tweets leveraging language model

Sharma

Tian

2015

2015 IEEE 22nd International Conference on Software Analysis, Evolution, and Reengineering (SANER)

Self Cite

View full text Add to dashboard Cite

Twitter is one of the most widely used social media platforms today. It enables users to share and view short 140-character messages called "tweets". About 284 million active users generate close to 500 million tweets per day. Such rapid generation of user generated content in large magnitudes results in the problem of information overload. Users who are interested in information related to a particular domain have limited means to filter out irrelevant tweets and tend to get lost in the huge amount of data they encounter. A recent study by Singer et al. found that software developers use Twitter to stay aware of industry trends, to learn from others, and to network with other developers. However, Singer et al. also reported that developers often find Twitter streams to contain too much noise which is a barrier to the adoption of Twitter. In this paper, to help developers cope with noise, we propose a novel approach named NIRMAL, which automatically identifies software relevant tweets from a collection or stream of tweets. Our approach is based on language modeling which learns a statistical model based on a training corpus (i.e., set of documents). We make use of a subset of posts from StackOverflow, a programming question and answer site, as a training corpus to learn a language model. A corpus of tweets was then used to test the effectiveness of the trained language model. The tweets were sorted based on the rank the model assigned to each of the individual tweets. The top 200 tweets were then manually analyzed to verify whether they are software related or not, and then an accuracy score was calculated. The results show that decent accuracy scores can be achieved by various variants of NIRMAL, which indicates that NIRMAL can effectively identify software related tweets from a huge corpus of tweets.

show abstract

“…IR-Based Bug Localization: Various IR-based bug localization approaches that employ information retrieval techniques to calculate the similarity between a bug report and a program element (e.g., a method or a source code file) have been proposed [39,30,22,44,57,41,47,48,53].…”

Section: Related Workmentioning

confidence: 99%

Information retrieval and spectrum based bug localization: better together

Oentaryo

2015

Proceedings of the 2015 10th Joint Meeting on Foundations of Software Engineering

Self Cite

112

View full text Add to dashboard Cite

Debugging often takes much effort and resources. To help developers debug, numerous information retrieval (IR)-based and spectrum-based bug localization techniques have been proposed. IR-based techniques process textual information in bug reports, while spectrum-based techniques process program spectra (i.e., a record of which program elements are executed for each test case). Both eventually generate a ranked list of program elements that are likely to contain the bug. However, these techniques only consider one source of information, either bug reports or program spectra, which is not optimal. To deal with the limitation of existing techniques, in this work, we propose a new multi-modal technique that considers both bug reports and program spectra to localize bugs. Our approach adaptively creates a bug-specific model to map a particular bug to its possible location, and introduces a novel idea of suspicious words that are highly associated to a bug. We evaluate our approach on 157 real bugs from four software systems, and compare it with a state-of-the-art IR-based bug localization method, a state-of-the-art spectrum-based bug localization method, and three state-of-the-art multi-modal feature location methods that are adapted for bug localization. Experiments show that our approach can outperform the baselines by at least 47.62%, 31.48%, 27.78%, and 28.80% in terms of number of bugs successfully localized when a developer inspects 1, 5, and 10 program elements (i.e., Top 1, Top 5, and Top 10), and Mean Average Precision (MAP) respectively.

show abstract

Compositional Vector Space Models for Improved Bug Localization

Cited by 59 publications

References 39 publications

Fusion fault localizers

Fusion fault localizers

NIRMAL: Automatic identification of software relevant tweets leveraging language model

Information retrieval and spectrum based bug localization: better together

Contact Info

Product

Resources

About