Some key points about tpday's paper

[Kazunari SUGIYAMA]

Dear all

Unfortunately, the members who could join today's meeting was
very small. So I would like to share why today's paper

"A Location-Based News Article Recommendation
with Explicit Localized Semantic Analysis'' (SIGIR'13)

has been accepted by SIGIR:

(1) Compare proposed approach with some existing approaches
(It's important to show comparison between *some* baselines
not only one baseline.)

(2) Proposed approach outperforms existing approaches

(3) Well-structured.


The following improvements make this paper much better:

(4) Clarify optimization of damping factor in PageRank

(5) Evaluation using MRR not Rt10 (Rating of Top10)


This paper is worth reading for those who try to submit to SIGIR
to learn how to write a SIGIR acceptable paper.

In addition, recently, graph-based social media analysis is emerging.
If you work on this field and try to employ PageRank to analyze
graph structures, please keep (4) in mind.

I hope that the information above is helpful for your work.


Best regards,
Kazunari Sugiyama