The dataset, named ACcoding, was built by a university teaching group. The latest release of ACcoding dataset is available at https://zenodo.org/record/6522395. By the time of constructing the dataset (May 6, 2022), the dataset contains 4,046,652 task-solving records submitted by 27,444 students on 4,559 programming tasks over a span of 6 years. The large size of the dataset, combined with its rich functional features, empowers educators to trace students’programming progress and choose appropriate programming tasks for specific training purposes.
You can download the ACcoding dataset from https://zenodo.org/record/6522395, which includes 6 SQL files. Assuming that you have already created a MySQL database, you can load the data files by executing the command
source xxx.sqlin MySQL. Alternatively, you can use tools like Navicat or DataGrip to load the SQL files.
ACcoding dataset is constructed and stored in a relational database in the form of tables. Note that in the table below, names in lower cases are entities, and those in upper cases are relations between the entities.
| Name | Description | Attributes | Quantity |
|---|---|---|---|
| Users | undergraduate users | id, last_login | 27444 |
| Problems | programming tasks | difficulty, time_limit, memory_limit | 4559 |
| Contests | regular ACM contests | start_time, end_time | 606 |
| Problem-tags | knowledge point (KP) tags of tasks | tag_id, problem_id, weight | 100 |
| Submissions | source code submitted by users | code, language, time_cost, memory_cost | 4046652 |
| FINISH | links between users and submissions | - | 4046652 |
| SUBMIT | links between submissions and tasks | time, result | 4046652 |
| CONTAIN | links between tasks and KPs | - | 4397 |
| BELONG TO | links between contests and tasks | order | 4756 |
Users:
| type | description | |
|---|---|---|
| id | int | use ID |
| last_login | datetime | user latests login time |
Problems:
| type | description | |
|---|---|---|
| id | int | problem ID |
| access_level | enum | 3 levels: private, protect, public |
| difficulty | decimal(10, 5) | difficulty of problem, ranges from 0 to 10 |
| updated_at | datetime | last revision time of the problem |
| creator_id | int | author ID of the problem |
Contests:
| type | description | |
|---|---|---|
| id | int | contest ID |
| start_time | datetime | start time of the contest |
| end_time | datetime | end time of the contest |
| updated_at | datetime | last revision time of the problem |
| access_level | enum | 3 levels: private, protect, public |
Problem-tags:
| type | description | |
|---|---|---|
| tag_id | int | tag ID associated with the problem |
| problem_id | int | problem ID |
Submissions:
| type | description | |
|---|---|---|
| id | int | contest ID |
| lang | enum | 'c++','c','python','java','python2','python3' |
| result | enum | 'WT','JG','AC','WA','CE','REG','MLE','REP','PE','TLE','IFNR','OFNR','EFNR','OE' |
| score | float | score of the submission, ranges from 0~100 |
| time_cost | int | runtime consumed by the code |
| memory_cost | int | memory consumed by running the code |
| code_length | int | length of the code |
| detail | text | compiler result of the code |
| creator_id | int | author ID of the submission |
| problem_id | int | problem ID |
| contest_id | int | contest ID |
Figure (a), (b) and (c) are the distribution of knowledge points, feedback result types and submission languages, respectively.
The following example is about visualizing embedding results for programming task difficulty, AC ratios, and heats (popularity) through ACcoding dataset.
cd data_process/experiment/visualize/
python visual_transx.py