毕业论文报告 孙悦明 2012.04.

Presentation on theme: "毕业论文报告 孙悦明 2012.04."— Presentation transcript:

1 毕业论文报告 孙悦明

2 论文题目 使用本体结构辅助系统化调研

3 提纲 论文工作介绍 系统化调研简介 结构化摘要简介 工作整体介绍 SLRONT的构建 扩展SLRONT为COSONT 实例化COSONT

4 系统化调研简介 系统化调研 SLR Protocol 定义研究问题 各个步骤的执行准则 Identification of Research
Study Selection Study Quality Assessment Data Extraction Data Synthesis

5 结构化摘要简介 结构化摘要 完备性与清晰性 Background Object Method Result Conclusion

6 工作整体介绍 工作流程图

7 SLRONT的构建 Review Protocol部分

8 SLRONT的构建 Primary Study部分

9 扩展为COSONT 主要扩展Structured abstract部分

10 COSONT 估算方法部分的扩展

11 实例化COSONT—非结构化摘要 Background Method Conclusion part

12 实例化COSONT—基于规则的分析 MAINSEN CONCLU 定位结果 R1: S = PP, NP VP R2: S = NP VP
In this paper, we propose an approach that converts cost estimation into a classification problem and that classifies new software projects in one of the effort classes, each of which corresponds to an effort interval. MAINSEN R1: S = PP, NP VP R2: S = NP VP CONCLU R3: S = (PP,)+ NP VP 定位结果 Results of the study show a significant correlation between the software development effort and all three models. Sentences Total Found Precision MAINSEN Sentence 347 267 76.95% CONCLU Sentence 244 70.3%

13 实例化COSONT—抽取本体信息 抽取的名词词组对比 抽取结论 - amod(method-9, effective-8)
The results show that KNN is an effective method. - amod(method-9, effective-8) - nsubj(method-9, KNN-5) - cop(method-9, is-6) - det(method-9, an-7) Paper Manual Auto Aver Recall per paper Student 1 85 652 508 80.102% Student 2 406 342 84.281% Student 3 472 336 72.428% Student 4 503 390 77.613% This paper describes a controlled experiment of student programmers performing maintenance tasks on a C++ program.

14 实例化COSONT

15 支持SLR中的关键步骤 辅助系统化调研第二步 寻找带有regression and neural network的文章 专家判断：
自动化方法 Student ID Num of Papers Paper Identified Time (Person*Hour) 1 87 6 8.5 2 7.5 3 9 4 86 10 Total 347 11 35

16 Q&A

17 感谢各位老师和同学！