Recommendations in the Scientific World

1
Recommendations in the Scientific World

• John Delacruz
• Oct 8, 2007

2
Outline

• Problems in science
• Finding papers
• Filtering and Merit
• Journalfire
• What is it?
• Data set
• Other data sets

3
Current Problem (Biology)

• Finding papers
• 739,303 articles in 2006
• 20 journals per discipline
• Filtering and merit
• Journal title
• 83 rejection rate

4
Solution

• Recommendation system
• Democratic rating system
• New measures for merit
• Stop relying on journal title

5
Journalfire
6
Recommendation systems

• Content-based recommendations
• The user will be recommended items similar to the
  ones the user preferred in the past
• Collaborative recommendations
• The user will be recommended items that people
  with similar tastes and preferences liked in the
  past
• Hybrid approaches
• These methods combine collaborative
• and content-based methods

Adomavicius and Tuzhilin, 2005
7
Journalfire data

• Pubmed
• Title
• Authors
• Journal
• Abstract
• MESH terms
• Date

• Journalfire
• Rating
• Tags
• Journal clubs
• User Favorites
• Date

8
Content based recommendations
author
author
keywords
article
article
journal
date
tags
9
Collaborative recommendations
Adomavicius and Tuzhilin, 2005
article
article
user lists
co-occurrence of articles
10
New collaborators
keywords
article
article
author
author
journal
journal
date
11
Other data sets
12
Faculty of 1000
6,500 ratings
13
Neurotree
14
Tagging