Recomandation Tool : library recommendation
We make available the source code of LisbScanDroid for study and experimentation in this directory GitHub: : https://github.com/alexingfds/LibScanDroid. This directory contains four (4) main directories. The The Results directory contains the results that were generated. The experimentation directory contains the results of the experiments that have been done with LibScanDroid and another CrossRec library recommendation tool.
The Resources directory contains the database we used. The viewing directory contains json format files that allow you to view with the LibScanDroid site. The LibScanDroid website is available in the following directory: https://github.com/stilab-ets/LibScanDroid. This directory contains a file index.html which allows you to launch the site for viewing libraries. The data directory contains the json format for visualization.
Install python 3.9 .
sudo apt install python3.9In this directory there is the following files that can be run separately:
- VisualisatonLibScanDroidRUN.py : For genarting recommandations and visualistion file. We can set the value of parameters (maxEpsiolon, Epsion, minPoint ..)
- tenFoldsTrainingRUN : perfoming ten folds cross validation (PUC: pattern usage cohesion)
- RecallRateDBSCANRUN : Evaluate Recall rate and MRR. (Evaluation of recommandation)
If you encounter any difficulties in working with the tool or the datasets, please do not hesitate to contact us at githe following email: richardson.alexandre.1@ens.etsmtl.ca . We will try our best to answer you as soon as possible.