Detecting duplicate bug reports helps reduce triaging efforts and save time for developers in fixing the same issues. Among several automated detection approaches, text-based information retrieval (IR) approaches have been shown to outperform others term of both accuracy efficiency. However, those IR-based do not detect well on technical issues written different descriptive terms.

Recent research has successfully applied the statistical n-gram language model to show that source code exhibits a good level of repetition. The is shown have predictability in supporting suggestion and completion. However, state-of-the-art approach capture regularities/patterns based only on lexical information local context units. To improve predictability, we introduce SLAMC, novel semantic for code. It incorporates into tokens models such annotations, called sememes, rather than their...

Locating buggy code is a time-consuming task in software development. Given new bug report, developers must search through large number of files project to locate code. We propose BugScout, an automated approach help reduce such efforts by narrowing the space when they are assigned address report. BugScout assumes that textual contents report and its corresponding source share some technical aspects system which can be used for locating given develop specialized topic model represents those...

Bug localization refers to the automated process of locating potential buggy files for a given bug report. To help developers focus their attention those is crucial. Several existing approaches from report face key challenge, called lexical mismatch, in which terms used reports describe are different and code tokens source files. This paper presents novel approach that uses deep neural network (DNN) combination with rVSM, an information retrieval (IR) technique. rVSM collects feature on...

Reusing existing library components is essential for reducing the cost of software development and maintenance. When evolve to accommodate new feature requests, fix bugs, or meet standards, clients libraries often need make corresponding changes correctly use updated libraries. Existing API usage adaptation techniques support simple such as replacing target calls a deprecated API, however, cannot handle complex adaptations creating object be passed different method, adding an exception...

Word2Vec is a class of neural network models that as being trainedfrom large corpus texts, they can produce for each unique word acorresponding vector in continuous space which linguisticcontexts words be observed. In this work, we study thecharacteristics vectors, called API2VEC or API embeddings, the elements within sequences source code. Ourempirical shows close proximity vectorsfor reflects similar usage contexts containing thesurrounding APIs those elements. Moreover, captureseveral...

Learning and remembering how to use APIs is difficult. While code-completion tools can recommend API methods, browsing a long list of method names their documentation tedious. Moreover, users easily be overwhelmed with too much information. We present novel recommendation approach that taps into the predictive power repetitive code changes provide relevant recommendations for developers. Our tool, APIREC, based on statistical learning from fine-grained context in which those were made....

n-gram statistical language model has been successfully applied to capture programming patterns support code completion and suggestion. However, the approaches using face challenges in capturing at higher levels of abstraction due mismatch between sequence nature n-grams structure syntax semantics source code. This paper presents GraLan, a graph-based its application GraLan can learn from corpus compute appearance probabilities any graphs given observed (sub)graphs. We use develop an API...

In this paper, we present a large-scale study of repetitiveness code changes in software evolution. We collected large data set 2,841 Java projects, with 1.7 billion source lines (SLOC) at the latest revisions, 1.8 million change revisions (0.4 fixes), 6.2 changed files, and 2.5 SLOCs. A is considered repeated within or cross-project if it matches another having occurred history project project, respectively. report following important findings. First, could be as high 70-100% small sizes...

Prior research has shown that source code also exhibits naturalness, i.e. it is written by humans and likely to be repetitive. The researchers showed the n-gram language model useful in predicting next token a file given large corpus of existing code. In this paper, we investigate how well statistical machine translation (SMT) models for natural languages could help migrating from one programming another. We treat as sequence lexical tokens apply phrase-based SMT on lexemes those tokens. Our...

n-gram statistical language model has been successfully applied to capture programming patterns support code completion and suggestion. However, the approaches using face challenges in capturing at higher levels of abstraction due mismatch between sequence nature n-grams structure syntax semantics source code. This paper presents GraLan, a graph-based its application GraLan can learn from corpus compute appearance probabilities any graphs given observed (sub)graphs. We use develop an API...

Code completion helps improve developers' programming productivity. However, the current support for code is limited to context-free templates or a single method call of variable on focus. Using software libraries development, developers often repeat API usages certain tasks. Thus, tool could make use usage patterns. In this paper, we introduce GraPacc, graph-based, pattern-oriented, context-sensitive approach that based database such GraPacc represents and manages patterns multiple...

The same software product nowadays could appear in multiple platforms and devices. To address business needs, companies develop a programming language then migrate it to another one. support that process, semi-automatic migration tools have been proposed. However, they require users manually define the mappings between respective APIs of libraries used two languages. reduce such manual effort, we introduce StaMiner, novel data-driven approach statistically learns from corpus corresponding...

Code completion helps improve developers' programming productivity. However, the current support for code is limited to context-free templates or a single method call of variable on focus. Using software libraries development, developers often repeat API usages certain tasks. Thus, tool could make use usage patterns. In this paper, we introduce GraPacc, graph-based, pattern-oriented, context-sensitive approach that based database such GraPacc represents and manages patterns multiple...

The links between the bug reports in an issue-tracking system and corresponding fixing changes a version repository are not often recorded by developers. Such linking information is crucial for research mining software repositories measuring defects maintenance efforts. However, state-of-the-art bug-to-fix link recovery approaches still rely much on textual matching commit/change logs cannot handle well cases where their contents textually similar.

Prior research shows that directly applying phrase-based SMT on lexical tokens to migrate Java C# produces much semantically incorrect code. A key limitation is the use of sequences in model and translate source code with well-formed structures. We propose mppSMT, a divide-and-conquer technique address novel training migration algorithms using three phases. First, mppSMT treats program as sequence syntactic units maps/translates such two languages one another. Second, syntax-directed...

An extended unsteady vortex-lattice method is developed to study the aerodynamics of insect flapping wings while hovering and during forward flight. Leading-edge suction analogy vortex-core growth models are used as an extension, which incorporated into a conventional in effort overcome challenges that arise when simulating such wing–wake interaction leading-edge effects. A convergence analysis was carried out derive optimal aerodynamic mesh time-step size for flapping-wing models. parallel...

In this work, we develop T2API, a statistical machine translation-based tool that takes given English description of programming task as query, and synthesizes the API usage template for by learning from training data. T2API works in two steps. First, it derives elements relevant to described input statistically StackOverflow corpus text descriptions corresponding code. To infer those elements, also considers context words textual often go together corpus. The inferred with their relevance...

In the era of mobile computing, developers often need to migrate code written for one platform in a programming language another different platform, e.g., from Java Android C# Windows Phone. The migration process is performed manually or semi-automatically, which are required define translation rules and API mappings. This paper presents semSMT, an automatic tool C#. semSMT utilizes statistical machine automatically infer existing migrated code, thus, requires no manual defining rules. video...

Problem. Code migration between languages is challenging partly because different require developers to use software libraries and frameworks. For example, in Java, Java Development Kit library (JDK) a popular toolkit while .NET the main framework used C# development. requires not only mappings language constructs (e.g., statements, expressions) but also among APIs of libraries/frameworks two languages. write file, one can FileWriter.write FileWriter, C#, achieve same function with...

Software developers often make use of the online forums such as StackOverflow (SO) to learn how software libraries and their APIs. However, code snippets in a forum contain undeclared, ambiguous, or largely unqualified external references. Such declaration ambiguity reference present challenges for learning correctly In this paper, we propose StatType, statistical approach resolve fully qualified names (FQNs) API elements snippets. Unlike existing approaches that are based on heuristics,...

Statistical language models (LMs) have been applied in several software engineering applications. However, they issues dealing with ambiguities the names of program and API elements (classes method calls). In this paper, inspired by success Deep Neural Network (DNN) natural processing, we present Dnn4C, a DNN model that complements local context lexical code both syntactic type contexts. We designed context-incorporating to use annotations for source order learn distinguish tokens different...

Cellular unjamming is the collective fluidization of cell motion and has been linked to many biological processes, including development, wound repair, tumor growth. In growth, uncontrolled proliferation cancer cells in a confined space generates mechanical compressive stress. However, because multiple cellular molecular mechanisms may be operating simultaneously, role stress transitions during progression remains unknown. Here, we investigate which mechanism dominates dense, mechanically...

Abstract Air pollution poses a significant threat to the health of environment and human well-being. The air quality index (AQI) is an important measure that describes degree its impact on health. Therefore, accurate reliable prediction AQI critical but challenging due non-linearity stochastic nature particles. This research aims propose hybrid deep learning model based Attention Convolutional Neural Networks (ACNN), Autoregressive Integrated Moving Average (ARIMA), Quantum Particle Swarm...