Ronnie's page 
subscribe
I'm a Phd Student in Informatics.My research interests are Databases & Data Mining.More about me at my personal homepage.
An overall picture of PubMed
Topic Hubs
|
Watchlist activity
This user hasn't added anything to this watchlist.
Learn more.
will let you be notified of changes to this item on your watchlist page (for example: you�ll get notified about new comments, new visualizations, etc).
Ronnie's Contributions
funciones para deseño del Mario Graph
Business Intelligence, KDD and Data Mining People
Co-related topics to HIV research from PubMed 2005
Most referenced topics in protein research by the Top-10 most frequent authors in PubMed 2005
Treemap of authors and its associated keywords from PubMed 2005
funciones para deseño del Mario Graph
Comments [see all 11 comments]
| Subject | Comment |
|---|---|
Giovani's profile
| Jovem, vai lá dar um "positive rating" nas nossas visualizações! Tb se qu... |
|
Ronnie's profile
| Hi Frank, many thanks for this clearly and complete response! ;0) As yo... |
Co-related topics to HIV research from PubMed 2005
| Zooming HIV-Infections. |
|
Association between authors and keywords in PubMed 2005
| Everyone has connection to sequence-analysis. |
|
Co-related Keywords in PubMed 2005
| Keywords associated to FEAR! |
|
Frank van Ham's profile
| Hi Frank! I have few questions about network diagrams: the nodes's size... |
Messages
(4)
|
Fernanda B. Viegas
says:
Oi, Ronnie, tudo bem? Bem vindo ao Many Eyes! Que legal voce ter criado o PubMed topic hub; eu ja' me adicionei ao grupo.
O que voce esta' estudando em Portugal? |
|
Ronnie
says:
Oi Fernanda! Tudo tranquilo! E vc? Muito legal este serviço q vcs disponibilizaram! Veio mesmo em boa hora! Estamos realizando alguma "garimpagem" sobre os dados do PubMed, e andavamos a procurar de compartilhar datasets e visualizacoes. Enfim, ta bem Legal!
Eu trabalho c/ multidimensional data mining...uma combinacao de varias tecnicas e algoritmos de mining e databases. Tem um link da minha pagina com os trabalhos q realizamos por estes lados. Quem sabe no futuro, possamos explorar estas tecnicas de visualizacao junto com os metodos q temos desenvolvido por aki...principalmente na area de "Gradients", Top-K queries, Rankings. Legal q vc se adicionou! Na medida do possivel vou acrescentando mais coisas tb. Tenho q terminar a minha tese tb por estes tempos! :0) |
|
Frank van Ham
says:
Hi Ronnie, in response to your questions:
The networks are treated as undirected currently, so the size of the nodes is the sum of the in and out degrees. The same holds for strong and weakly connected components, since we are not taking directions into account, the is no difference between strong and weakly connected. Two nodes are either connected or not. Strongly connected components are identified before layout and are split off. The layout algorithm tries to satisfy two requirements: connected nodes need to be as close as possible and unconnected nodes need to be as far apart as possible. The final positions are being determined by an algorithm from multidimensional scaling called "stress majorization". Alternatives would be incremental force directed graph layouts, but these are worse at finding the 'optimal' configuration. For references see http://www.research.att.com/~yehuda/pubs/majorization.pdf or http://en.wikipedia.org/wiki/Force-based_algorithms. I'm still working on visualization of edge weights, you can always manually filter the graph by weight if you want. Hope this helps, keep up the good work! |
|
Ronnie
says:
Hi Frank,
many thanks for this clearly and complete response! ;0) As you may know PubMed DB is a very big database (50gb at least!), ...so we came up with that idea of using an top-k approach to generalize queries over the most important objects in the DB. We hope to get other datasets by using top-k with graph components...so we will be able to filter out that components manually. This also help us to keep them sizing 5Mb. However, sometimes we can see that "small word figure" even by using such data reduction. For sure, we will keep our work here!!! Thanks also for that references! I will read carefully! See ya!! |