The Data Mining Forum 
IMPORTANT: This is the old Data Mining forum.
I keep it online so that you can read the old messages.
Please post your new messages in the new forum: https://forum2.philippe-fournier-viger.com/index.php
Mining based on the graphs
Posted by:
Nima
Date: October 16, 2018 01:11AM
Hi,
Is there any data mining method based on the graph data structure?
Is it efficient to mine frequent patterns based on graph data structures?
Waiting for your helpful information.
Nima
Is there any data mining method based on the graph data structure?
Is it efficient to mine frequent patterns based on graph data structures?
Waiting for your helpful information.
Nima
Re: Mining based on the graphs
Posted by:
webmasterphilfv
Date: October 16, 2018 01:32AM
Hello,
Yes, there are many algorithms to discover patterns in graph.
For example, there are several algorithms to discover frequent subgraphs in a graph or multiple graphs. Such algorithms are GSpan, FSG, etc. For an overview of frequent subgraph mining, here is an introduction to frequent subgraph mining. Besides that people will also discover other types of patterns such as cliques in graphs.
Besides that, there are lot of other tasks that can be done with graphs such as community detection in social network, relational classification (inferring the labels of unlabelled nodes in a graph), studying the evolution of a graph, etc.
How is the performance? It depends on the algorithms and the type of graph. Some algorithms are certainly more scalable than others.
What kind of graph data do you want to analyze?
Edited 1 time(s). Last edit at 10/16/2018 01:33AM by webmasterphilfv.
Yes, there are many algorithms to discover patterns in graph.
For example, there are several algorithms to discover frequent subgraphs in a graph or multiple graphs. Such algorithms are GSpan, FSG, etc. For an overview of frequent subgraph mining, here is an introduction to frequent subgraph mining. Besides that people will also discover other types of patterns such as cliques in graphs.
Besides that, there are lot of other tasks that can be done with graphs such as community detection in social network, relational classification (inferring the labels of unlabelled nodes in a graph), studying the evolution of a graph, etc.
How is the performance? It depends on the algorithms and the type of graph. Some algorithms are certainly more scalable than others.
What kind of graph data do you want to analyze?
Edited 1 time(s). Last edit at 10/16/2018 01:33AM by webmasterphilfv.
Re: Mining based on the graphs
Posted by:
Nima
Date: October 16, 2018 02:07AM
Thanks for your prompt response.
I actually know that algorithm for mining graph datasets.
Maybe my question was not clear.
I want to know is there any algorithm for mining frequent patterns from binary datasets( Retail, Chess, etc) based on graph data structure such as a directed graph.
Best
I actually know that algorithm for mining graph datasets.
Maybe my question was not clear.
I want to know is there any algorithm for mining frequent patterns from binary datasets( Retail, Chess, etc) based on graph data structure such as a directed graph.
Best
Re: Mining based on the graphs
Posted by:
webmasterphilfv
Date: October 16, 2018 07:54AM
Hi Nima,
I see.
I remember that a few years ago I saw a paper about some algorithms called FPGraph or something like that. Some person contacted with me by e-mail to ask me to implement this algorithm. But it was not published in a good journal if I remember well, and maybe even that it was an incorrect algorithm because the paper lacked details from what I remember.
After that I did not see such paper. But maybe there are. Actually, it is an interesting idea ;-) maybe you can do something with this.
Best regards,
I see.
I remember that a few years ago I saw a paper about some algorithms called FPGraph or something like that. Some person contacted with me by e-mail to ask me to implement this algorithm. But it was not published in a good journal if I remember well, and maybe even that it was an incorrect algorithm because the paper lacked details from what I remember.
After that I did not see such paper. But maybe there are. Actually, it is an interesting idea ;-) maybe you can do something with this.
Best regards,