TSE_2022_Cross Project Code Context Prediction for Software Development Tasks

The dataset of 3,469 code context models (Code Context Models.zip) and 14,510 code context patterns (Topological Patterns.zip) from Platform, PDE, and ECF. <strong>Code Context Models</strong> There are 10 files in each directory (i.e., working period), including: 1. interaction events.txt: the interaction events during this working period 2. event timestamps.txt: the timestamps of this working period, including: start datetime of the first event start datetime of the last event end datetime of the last event duration of interaction events (in seconds) 3. code elements_extracted.txt: the code elements the developer accesses during this working period, extracted from interaction events (e.g., org.eclipse.ecf.provider.irc.ui/src&amp;lt;org.eclipse.ecf.internal.irc.ui.wizards{IRCConnectWizardPage.java[IRCConnectWizardPage~getConnectID) 4. code elements_resolved.txt: the resolved code elements (e.g., 'org.eclipse.ecf.internal.irc.ui.wizards.IRCConnectWizardPage[getConnectID') 5. repository urls.txt: the urls of the git repositories identified from the code elements 6. code elements_total.txt: all the code elements extracted from the code snapshots 7. code context model.txt: the code context model with the code elements in [4]. This file is like: vertices: A@#c0 B@#f0 edges: c0@#f0@#declare A and B are the code elements, while c0, f0 are their ids (c: class, f: function). The relation between them is 'declare'. 8. code context model_total.txt: the code context model with the code elements in [6] 9. stereotype roles_total.txt: the stereotype roles of all the code elements in [6] 10. code context model_abstract.txt: the abstract code context model of which the code elements are abstracted as stereotype roles <strong>Topological Patterns</strong> There are 9, 8, and 7 pattern groups mined from Platform, PDE, and ECF respectively. Each pattern group is named 'project_X_MinSupp', where 'X' denotes the X-th cluster of code context models and 'MinSupp' = 0.02. The information about the X-th cluster can be found in the paper's appendix. There are various patterns in a pattern group. For example: t # 25<br> v 0 DATA_PROVIDER <br> v 1 COLLABORATOR-CONSTRUCTOR <br> e 0 1 declare<br> 12 The id of this pattern is 25, and there are two vertices 0 and 1, corresponding to DATA_PROVIDER and COLLABORATOR-CONSTRUCTOR respectively. There is a 'declare' edge from DATA_PROVIDER to COLLABORATOR-CONSTRUCTOR. This pattern occurs 12 times in all the abstract code context models

本数据集涵盖来自Platform、PDE及ECF的3469份代码上下文模型（Code Context Models，对应压缩包为Code Context Models.zip）与14510份代码上下文模式（Topological Patterns，对应压缩包为Topological Patterns.zip）。

### 代码上下文模型（Code Context Models）
每个目录（即对应一个工作周期）下包含10个文件，具体如下：
1. 交互事件记录文件（interaction events.txt）：记录当前工作周期内产生的交互事件
2. 事件时间戳文件（event timestamps.txt）：包含当前工作周期的以下时间信息：首个事件的开始时间、最后一个事件的开始时间、最后一个事件的结束时间、交互事件总时长（单位：秒）
3. 提取代码元素记录文件（code elements_extracted.txt）：从交互事件中提取得到的开发者在当前工作周期内访问的代码元素（示例格式：org.eclipse.ecf.provider.irc.ui/src<org.eclipse.ecf.internal.irc.ui.wizards{IRCConnectWizardPage.java[IRCConnectWizardPage~getConnectID）
4. 已解析代码元素记录文件（code elements_resolved.txt）：已完成解析的代码元素（示例：'org.eclipse.ecf.internal.irc.ui.wizards.IRCConnectWizardPage[getConnectID'）
5. 代码仓库URL记录文件（repository urls.txt）：从代码元素中识别出的Git仓库URL
6. 全量代码元素记录文件（code elements_total.txt）：从代码快照中提取的全部代码元素
7. 单批次代码上下文模型文件（code context model.txt）：包含第4项中代码元素的代码上下文模型，格式示例如下：
vertices: A@#c0 B@#f0
edges: c0@#f0@#declare
其中A、B为代码元素，c0、f0为其标识（c代表类，f代表函数），二者间的关系为“声明”。
8. 全量代码上下文模型文件（code context model_total.txt）：包含第6项中全部代码元素的代码上下文模型
9. 全量代码元素原型角色记录文件（stereotype roles_total.txt）：第6项中全部代码元素对应的原型角色
10. 抽象代码上下文模型文件（code context model_abstract.txt）：将代码元素抽象为原型角色后得到的抽象代码上下文模型

### 拓扑模式（Topological Patterns）
从Platform、PDE及ECF中分别挖掘得到9组、8组及7组模式集群。每组模式集群以`project_X_MinSupp`命名，其中X代表第X个代码上下文模型集群，MinSupp取值为0.02。第X个集群的相关信息可参见论文附录。
每个模式集群包含多种模式，示例如下：
t # 25<br> v 0 DATA_PROVIDER <br> v 1 COLLABORATOR-CONSTRUCTOR <br> e 0 1 declare<br> 12
该模式的标识为25，包含两个顶点0和1，分别对应DATA_PROVIDER与COLLABORATOR-CONSTRUCTOR，存在一条从DATA_PROVIDER指向COLLABORATOR-CONSTRUCTOR的“声明”关系边。该模式在所有抽象代码上下文模型中共出现12次。