Xiaoyu Zhou

Link prediction in social networks: the state-of-the-art

In social networks, link prediction predicts missing links in current networks and new or dissolution links in future networks, is important for mining and analyzing the evolution of social networks. In the past decade, many works have been done about the link prediction in social networks. The goal of this paper is to comprehensively review, analyze and discuss the state-of-the-art of the link prediction in social networks. A systematical category for link prediction techniques and problems is presented. Then link prediction techniques and problems are analyzed and discussed. Typical applications of link prediction are also addressed. Achievements and roadmaps of some active research groups are introduced. Finally, some future challenges of the link prediction in social networks are discussed.创新点对社交网络中的链接预测研究现状进行系统回顾、分析和讨论, 并指出未来研究挑战. 在动态社交网络中, 链接预测是挖掘和分析网络演化的一项重要任务, 其目的是预测当前未知的链接以及未来链接的变化. 过去十余年中, 在社交网络链接预测问题上已有大量研究工作. 本文旨在对该问题的研究现状和趋势进行全面回顾、分析和讨论. 提出一种分类法组织链接预测技术和问题. 详细分析和讨论了链接预测的技术、问题和应用. 介绍了该问题的活跃研究组. 分析和讨论了社交网络链接预测研究的未来挑战.

Source code size estimation approaches for object-oriented systems from UML class diagrams: A comparative study

Background: Source code size in terms of SLOC (source lines of code) is the input of many parametric software effort estimation models. However, it is unavailable at the early phase of software development. Objective: We investigate the accuracy of early SLOC estimation approaches for an object-oriented system using the information collected from its UML class diagram available at the early software development phase. Method: We use different modeling techniques to build the prediction models for investigating the accuracy of six types of metrics to estimate SLOC. The used techniques include linear models, non-linear models, rule/tree-based models, and instance-based models. The investigated metrics are class diagram metrics, predictive object points, object-oriented project size metric, fast&&serious class points, objective class points, and object-oriented function points. Results: Based on 100 open-source Java systems, we find that the prediction model built using object-oriented project size metric and ordinary least square regression with a logarithmic transformation achieves the highest accuracy (mean MMRE=0.19 and mean Pred(25)=0.74). Conclusion: We should use object-oriented project size metric and ordinary least square regression with a logarithmic transformation to build a simple, accurate, and comprehensible SLOC estimation model.

A Constraint Based Bug Checking Approach for Python

Python is a powerful dynamically typed programming language. Dynamic typing brings great flexibility for programming. However, lack of static type checking, it is hard to detect some bugs before run time. We present a constraint framework based on Python’s structural equivalence type system. The framework does not introduce any new language features, thus without losing benefits of Python’s dynamic typing. Constraints are extracted from source code via static analysis and are used to check bugs, such as passing wrong parameters to function. A case study shows how to use the framework to check validity of function calls.

Inferring weak references for fixing Java memory leaks

Lots of efforts have yet been devoted to the detection of memory leaks. However, very few work concerns on the problem of memory leak fixing. In this paper, we propose a dynamic approach for the weak reference based memory leak fixing. The approach can automatically infer the weakenable references in a program, rank them and report them to the user. The user can then weaken references according to the reports to fix the memory leaks.

Are Anti-patterns Coupled? An Empirical Study

The interactions between anti-patterns are claimed to affect maintenance. However, little work has been conducted to examine how anti-patterns interact. In this paper, we aim to investigate which pairs of anti-patterns tend to be coupled, i.e., interact with each other. We employ Fishers exact test and Wilcoxon rank-sum test to identify coupled anti-patterns in the same class and coupled classes. Analyzing the relationships amongst 10 kinds of anti-patterns in five open-source projects, our results show that 1) several kinds of anti-patterns tend to be coupled, but some are conflicting, 2) the effect of anti-patterns on their dependent or co-changed ones are significant but small, 3) in ArgoUML, Xalan and Xerces-J, the classes infected with dependent anti-patterns are mostly (69.9% ~ 100%) modified in maintenance activities. Our findings offer empirical evidences for the existence of anti-pattern interactions, which provides valuable implications for practitioners and researchers.

Slicing Concurrent Programs Based on Program Reachability Graphs

Program slicing is an effective and established technique to analyze sequential programs. However, owing to intransitive dependences between statements, it is hard to get precise slices of concurrent programs with traditional slicing algorithms for sequential programs. This paper proposes a novel representation, thread interaction reachability graph (TIRG), for concurrent programs with shared memory, interleaving semantics and mutual exclusion. Based on TIRG, dependences can be analyzed globally and we construct a new dependence graph (MSDG), where each node is a 2-tuple composed of program state and statement. The research shows that dependences in MSDG are transitive and more precise slice can be obtained by traversing MSDG compared with previous approaches.

Dependence Analysis for C Programs with Combinability of Dataflow Facts under Consideration

To avoid the precision loss caused by combining dataflow facts impossible to occur in the same execution path in dependence analysis for C programs, this paper first proposes a flow-sensitive and context-insensitive points-to analysis algorithm and then presents a new dependence analysis approach based on it. The approach makes more sufficient consideration on the executable path problem and can avoid invalid combination between points-to relations and between points-to relations and reaching definitions. The results of which are therefore more precise than those of the ordinary dependence analysis approaches.

Understanding metric-based detectable smells in Python software: A comparative study

Abstract Context Code smells are supposed to cause potential comprehension and maintenance problems in software development. Although code smells are studied in many languages, e.g. Java and C#, there is a lack of technique or tool support addressing code smells in Python. Objective Due to the great differences between Python and static languages, the goal of this study is to define and detect code smells in Python programs and to explore the effects of Python smells on software maintainability. Method In this paper, we introduced ten code smells and established a metric-based detection method with three different filtering strategies to specify metric thresholds (Experience-Based Strategy, Statistics-Based Strategy, and Tuning Machine Strategy). Then, we performed a comparative study to investigate how three detection strategies perform in detecting Python smells and how these smells affect software maintainability with different detection strategies. This study utilized a corpus of 106 Python projects with most stars on GitHub. Results The results showed that: (1) the metric-based detection approach performs well in detecting Python smells and Tuning Machine Strategy achieves the best accuracy; (2) the three detection strategies discover some different smell occurrences, and Long Parameter List and Long Method are more prevalent than other smells; (3) several kinds of code smells are more significantly related to changes or faults in Python modules. Conclusion These findings reveal the key features of Python smells and also provide a guideline for the choice of detection strategy in detecting and analyzing Python smells.

A Specification-Based Approach to the Testing of Java Memory Bloat

Inefficient use of memory may cause memory bloat, and the bloat may lead to performance slowdowns or even crashes. To address the problem, lots of efforts have been devoted to the diagnosing of bloat, especially the bloat caused by memory leaks. However, testing and identifying the executions potentially containing bloat, which act as key steps before diagnosing, are still challenging. In this paper, we introduce a memory specification tool named MemSpec to help programmers test memory inefficiencies in Java programs. The specification works as a test oracle to automatically determine whether a program suffers from memory bloat. Users can write their own memory specifications using our specification language. MemSpec weaves the specification checking mechanisms into the tested programs. During test runs, memory bloat will be reported if the some specifications are not satisfied. With this specification-based approach, a tester does not need to manually watch the program executions to identify bloat. The automatically reporting of memory bloat can reduce the manual efforts in memory inefficiency testing. We test the proposed approach on ArgoUML. The initial results show that it is effective.

Side-effect analysis with fast escape filter

Side-effect analysis is a fundamental static analysis used to determine the memory locations modified or used by each program entity. For the programs with pointers, the analysis can be very imprecise. To improve the precision of side-effect analysis, many approaches design more elaborate background pointer analyses in order to obtain smaller side-effect sets, but very few approaches consider to achieve better precisions by refining the side-effect analysis algorithms themselves. To address the problem, this paper presents a new side-effect analysis approach that uses Gay and Steensgaards fast escape analysis to filter superfluous side-effects. The approach does not need to modify the background pointer analysis and can filter side-effects in the intraprocedural level and the interprocedural level. The experimental results show that it can effectively improve the analysis precision within a short extra time.