Skip to content

Latest commit

 

History

History
88 lines (65 loc) · 3.6 KB

README.md

File metadata and controls

88 lines (65 loc) · 3.6 KB
Raw

VisCARS: Graph-Based Context-Aware Visualization Recommendation System

version

Installation

Create a virtual environment using virtualenv or anaconda3:

conda create -n myenv python=3.9
conda activate myenv

Install the latest version from PyPI in your environment:

pip install --upgrade viscars

Basic usage

Load the dataset

from rdflib import Graph

graph_ = Graph()
graph_.parse('../data/protego/protego_ddashboard.ttl')
graph_.parse('../data/protego/protego_zplus.ttl')
graph_.parse('../data/protego/visualizations.ttl')

Initialize the two-stage recommendation pipeline

from viscars.dao import ContentRecommenderDAO, VisualizationRecommenderDAO
from viscars.recommenders.cacf import ContextAwareCollaborativeFiltering

# Initialize Content Recommender (stage 1)
content_dao = ContentRecommenderDAO(graph_)
content_recommender = ContextAwareCollaborativeFiltering(content_dao, cbcf_w=0.5, ubcf_w=0.5, verbose=False)

# Initialize Visualization Recommender (stage 2)
vis_dao = VisualizationRecommenderDAO(graph_)
visualization_recommender = ContextAwareCollaborativeFiltering(vis_dao, ubcf_w=1, verbose=False)

Run the pipeline for a user and context

# user = 'https://dynamicdashboard.ilabt.imec.be/users/4'  # Operator
user = 'https://dynamicdashboard.ilabt.imec.be/users/5'  # Nurse

context = 'http://example.com/tx/patients/zplus_6'  # Diabetes

content_recommendations = content_recommender.predict(user, context, k=5)

# Find cutoff for Multiple-View recommendation
# We recommend the top x items, where x is the average number of items rated by users in the context
ratings = content_dao.ratings[(content_dao.ratings['c_id'] == context)]
c = int(ratings.value_counts('u_id').mean())

visualization_recommendations = []
for recommendation in content_recommendations[:c]:
    # Recommend visualizations
    recommendations = visualization_recommender.predict(user, recommendation['itemId'], k=5)
    visualization_recommendations.append({'propertyId': recommendation['itemId'], 'visualizationId': recommendations[0]['itemId']})

Example output

propertyId visualizationId
.../things/zplus_6.lifestyle/properties/enriched-call .../things/visualizations/enriched-call
.../things/zplus_6.60%3A77%3A71%3A7D%3A93%3AD7%2Fservice0009/properties/org.dyamand.types.health.GlucoseLevel .../things/visualizations/time-series-line-chart-with-time-range-selector
.../things/zplus_6.AQURA_10_10_145_9/properties/org.dyamand.aqura.AquraLocationState_Protego%20User .../things/visualizations/scrolling-table

Citation

VisCARS: Knowledge Graph-based Context-Aware Recommender System for Time-Series Data Visualization and Monitoring Dashboards

@article{moens2024viscars,
  title={VisCARS: Knowledge Graph-based Context-Aware Recommender System for Time-Series Data Visualization and Monitoring Dashboards},
  author={Moens, Pieter and Volckaert, Bruno and Van Hoecke, Sofie},
  journal={IEEE Transactions on Visualization and Computer Graphics},
  year={2024},
  publisher={IEEE}
}