1 / 13

Enhancing Scholarly Paper Recommendations through User Profiles and Citation Analysis

This paper presents methodologies for improving scholarly paper recommendations, focusing on user profile construction and paper characterization. We explore how to identify potential papers that should cite a target paper, utilizing a Paper-Citation Matrix for collaborative filtering. The effectiveness of our proposed approach is evaluated through experimental results, highlighting metrics such as Normalized Discounted Cumulative Gain (NDCG) and Mean Reciprocal Rank (MRR). Our findings suggest that full content analysis of potential papers significantly enhances recommendation accuracy compared to fragment-based approaches.

silver
Télécharger la présentation

Enhancing Scholarly Paper Recommendations through User Profiles and Citation Analysis

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Scholarly Paper Recommendation Exploiting Potential Papers 25th November, 2011 Kazunari Sugiyama

  2. Outline of My Former Research • How to construct user profile for scholarly paper recommendation

  3. Publication list new old (‘05) (‘11) (‘02) (‘03) Individual paper Citation papers (‘06) (‘07) (‘09) References (‘05) (‘01) (‘04) (‘03) Reference papers

  4. Outline of My Current Research • How to characterize candidate papers to recommend

  5. Proposed Approach (‘07) (‘09) Potential paper that should cite the target paper (‘06) (‘05) References (‘01) (‘03) (‘04)

  6. How to Find Potential Papers Pi (i=1, … ,N): All papers in the dataset Pcj (j=1, … ,N): Papers as citation papers in the dataset 0.368 0.536 0.211 0.472

  7. Characterize the Target Paper using Potential Papers (‘06) (‘07) (‘09) (‘05) Potential paper that should cite the target paper 7

  8. Paper-Citation Paper Matrix for CF

  9. Experimental Data

  10. Evaluation Measure • Normalized Discounted Cumulrative Gain (NDCG) • nDCG@5, nDCG@10 • Mean Reciprocal Rank (MRR)

  11. Experimental Results • Define optimal values using training set • Neighbors of target paper in CF, Number of potential papers

  12. Experimental Results • Apply optimal values obtained from training set to test set

  13. Conclusion • In order to provide better recommendation, we proposed how to characterize candidate papers to recommend. • Full contents of potential paper gives better recommendation accuracy compared with “fragments only” or “potential paper + fragments”

More Related