README.md

intent_viz

Maruta and Kato. Intent-aware Visualization Recommendation for Tabular Data. WISE 2021.

Setup

Clone our repository

git clone https://github.com/kasys-lab/intent-viz.git
cd intent-viz

Download large files

Our model requires word embedding model and trained model

Download pre-trained word embedding vector

To get word embedding model, please download glove.6B.zip.

And put glove.6B.100d.txt file in unzipped glove.6B.zip file to data/.

wget http://downloads.cs.stanford.edu/nlp/data/glove.6B.zip
unzip glove.6B.zip
mv glove.6B.100d.txt ./data/
rm glove*

Download trained model

To get trained model, please access https://drive.google.com/drive/folders/1G2WTiNGCOzu2W4E3IFB6WT62RJXJkFeZ?usp=sharing and download predict_vis_columns.pt.

And put predict_vis_columns.pt to saved_models/

Installation of Poetry (skip this step if Poetry has already been installed)

$ curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py | python

Installation of required Python packages

$ source ~/.bash_profile
$ poetry install

Prediction

Predict a visualization type and visualized column percents

$ poetry run python demo.py <data path>
$ poetry run python demo.py "./data/input_data.json"

Input data

Json format pair data of a visualization intent and tabular data.

{   "visualization_intent": "Population trends in Italy",
    "data": 
    [
        {
            "Year": 2015,
            "population": 100,
            "GDP": 2.3
        },{
            "Year": 2016,
            "population": 110,
            "GDP": 2.4
        },{
            "Year": 2017,
            "population": 130,
            "GDP": 2.9
        },{
            "Year": 2018,
            "population": 170,
            "GDP": 3.0
        },{
            "Year": 2019,
            "population": 200,
            "GDP": 3.1
        },{
            "Year": 2020,
            "population": 220,
            "GDP": 3.6
        }
    ]
}

Output

Output visualuzation type and visualized column percent

predict visualization type : Line chart
predict visualized columns percent
header  :  percent
Year  :  1.0
population  :  0.9150764707515349
GDP  :  0.0