### Prototype Application A prototype application can be found under following link: [Link to web application][1] ### Code See ReadMe.md in the code zip file on how to install the dependencies. To create a LinSets.zip diagram, click the link to upload in the navigation bar. Then, upload a dataset (examples can be found in the subfolder datasets) and select a pipeline configuration. ### Computational Experiments Evaluation (folder `computational experiments`) The input data is in the folder `dataset_dblp`, which was generated with `generate_dblp_data.py`. The results of the experiments are in the folder `results`, which were produced by the program `runExperiment.py`. Both folders `dataset_dblp` and `results` have been compressed to reduce space. The experimental results are presented in the jupyter notebook `evaluation.ipynb`, which is structured into three parts 1, 2, and 3, corresponding to the questions A, B, and C posed in the paper. The notebook creates the plots used in the paper, but also contains further results. The output of each code block contains a description of the resulting plot or presented data. The plots are also given in the folder `plots` as pdf files. Further, a description of all plots can be found in `DESCRIPTION.md`. To run experiments and create plots Python3 is required and the packages listed in `requirements.txt` must be installed. Further, to create the plots latex is required, and `results.zip` has to be unpacked in the same folder as `evaluation.ipynb`. The authors have used Ubuntu and texlive. To run the experiments the following setup is required. - The packages in `requirements.txt` must be installed. - The folder structure has to be as follows ``` ├── code ├── experiments │ ├── dataset_dblp │ └── runExperiment.py ``` - A running version of Gurobi has to be installed, and the requirements for `code` have to be installed. The output of a single experiment in the `results` folder is a json containing: - `file`: the file name - `heurisic`: `true` if this was a run of the heuristic pipeline, false otherwise - `compressopt`: The compression variant. 0 for Gamma_1, 1 for Gamma_2, 2 for Gamma_3. - `bound`: The bound for the maximum number of sets per row. - `optimizegapstime`: time required for optimizing gaps. - `compressldtime`: Time required for the compression of the linear diagram. - `rowordertime`: time required for ordering the rows. - `num_cols`: The number of elements of the instance. - `num_rows`: The number of sets of the instance. - `num_blocks`: The number of blocks produced by the column ordering algorithm. - `num_row_blocks`: The number of column-wise blocks produced by the row ordering algorithm. - `informationDensity`: an array containing the values of information density of the computed linear diagram. The first value is the space used of the uncompressed linear diagram divided by the number of rows times number of columns. The second value is this ratio for the compressed linear diagram. The third value is the ratio between the first two values in the array. - `num_unique_columns`: The number of unique elements in the instance. - `num_unique_rows`: The number of unique sets in the instance. - `optimizegapsmemory_exceeded`: true if the memory limit was exceeded during column ordering, false otherwise. - `compressldmemory_exceeded`: true if the memory limit was exceeded during compression, false otherwise. - `rowordermemory_exceeded`: true if the memory limit was exceeded during row ordering, false otherwise. - `coloringmemory_exceeded`: always false, artifact of old code. - `compressed`: A dictionary assigning each set a row. - `rows`: The order of rows. - `columns`: The order of columns (elements). ### User Study (folder 'user study') ``` ├── datasets ├── stimuli ├── study └── evaluation ├── results-survey.csv ├── clean_raw_study.py ├── cleaned_survey.csv ├── Comments.txt ├── evaluation_study.ipynb └── plots ``` All files related to the user study can be found in the user study folder. The structure is indicated above. The datasets folder contains all raw datasets that were used to generate the stimuli. The file name indicates the task-style pair (style 1 = \Gamma 0) The stimuli folder contains the previously mentioned generated images and the post-processing we applied. The study folder contains a complete screen capture of our study (note that order is reversed). The evaluation folder contains all results regarding the study and is described below. The evaluation can be found in the subfolder evaluation. The raw data can be found in the `results-survey.csv`. The script `clean_raw_study.py` cleans the raw data and creates two files which are already included. First, `Comments.txt` contains the participant's responses of the free-form text field. Second, `cleaned_survey.csv` is further used in the statistical analysis. The statistical analysis itself can be found in `evaluation_study.ipynb`. It contains and creates the plots used in the paper und more statistical information. Lastly, the sub-folder plots contains the plots that were generated from the study data. There is a description of the plots inside the folder. [1]: https://metrosets.ac.tuwien.ac.at/linsets-zip/