PaperRecommender is a general framework for relevant scientific paper recommendation, developed for experimental purpose in our research project. The source code was written in Java using Ant-based NetBeans project format, which can be opened in NetBeans IDE. For the latest source code, please see the Test Site branch.
We also built several datasets for scientific paper recommendation experiments based on a novel method to build ground truth data. These datasets contain metadata of over 4 million scientific papers in computer science, crawled from Microsoft Academic Search (MAS), finished in November 2012.
Dataset 1: https://drive.google.com/file/d/0B8gXe63FdGk5WTljdHdsSUw3UEk (.zip, 376 MB)
Dataset 2: https://drive.google.com/file/d/0B8gXe63FdGk5QlNsQmhVekx1SlU (.zip, 379 MB)
Dataset 3: https://drive.google.com/file/d/0B8gXe63FdGk5ZjlTWS1hZ0w0Tnc (.zip, 376 MB)
Dataset 4: https://drive.google.com/file/d/0B8gXe63FdGk5Q0pfNE1oNUVJZVU (.zip, 528 MB)
PaperRecommender is a free software under MIT License.
The dataset is provided under open ODC-BY License 1.0.
Corresponding paper:
If you find the codes or data useful, please cite the following paper.
Hung Nghiep Tran, Tin Huynh, Kiem Hoang. A Potential Approach to Overcome Data Limitation in Scientific Publication Recommendation. KSE 2015.
For more information, please visit the website: https://sites.google.com/site/tranhungnghiep