Quick Evaluation of Research Impacts at Conferences Using SNS

We are investigating ways of evaluating research impact as soon as possible after publication. Traditionally, the research impact or importance of academic journals has been evaluated using citation relations, such as the impact factor and the citation half-life. However, these citation-based method...

Full description

Saved in:
Bibliographic Details
Main Authors: Fukuda, Satoshi, Nakahashi, Hikaru, Nanba, Hidetsugu, Takezawa, Toshiyuki
Format: Conference Proceeding
Language:English
Subjects:
Alignment
Bibliometrics
Citation analysis
component
Conferences
Estimation
Expert systems
Indexes
Mathematical analysis
Measurement
research impact
research paper
Scientific papers
Twitter
Workshops
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We are investigating ways of evaluating research impact as soon as possible after publication. Traditionally, the research impact or importance of academic journals has been evaluated using citation relations, such as the impact factor and the citation half-life. However, these citation-based methods require long periods to evaluate research impact and therefore are not suitable for evaluating the current impact of research papers at conferences. To solve this problem, we are studying the automatic evaluation of research impact using Twitter. Researchers participating in academic conferences often post their opinions or comments on Twitter. Here, research papers (presentations) that have many comments are considered to be outstanding and to have strong impact during the conference. In this paper, we propose a method for automatically aligning tweets with research papers. The procedure consists of the following three steps: (1) detecting valuable tweets, (2) aligning each valuable tweet with a research paper, and (3) calculating the research impact of each research paper by the number of aligned tweets. We conducted some experiments to confirm the effectiveness of our method. From the results, we obtained an MRR score of 0.223, which outperformed a baseline method.
ISSN:1529-4188
2378-3915
DOI:10.1109/DEXA.2015.64