you can also do a map of sets, with the type as key
On Sat, Oct 4, 2014 at 12:57 AM, Mohana Krishna, IIT Bombay, India <
mohana...@gmail.com> wrote:
> In my algorithm, if there are say 17 types, then I need to enumerate all
> possible cliques for each of size-3 combinations of total 17 types.
> S
In my algorithm, if there are say 17 types, then I need to enumerate all
possible cliques for each of size-3 combinations of total 17 types.
Similarly for size-4, all possible cliques for each of size-4 combinations
of total 17 types and so on. So, is the idea of using sets in java good
(since
Not sure, depends on your algorithm. I thought one set per clique-size is
good enough.
On Sat, Oct 4, 2014 at 12:43 AM, Mohana Krishna, IIT Bombay, India <
mohana...@gmail.com> wrote:
> That way, if the maximum clique size I can have is 17, then at size-3
> level I need to check 17C3 different cl
That way, if the maximum clique size I can have is 17, then at size-3 level
I need to check 17C3 different clique combinations. Do I need to store
results of all these in 17 different sets? Is my understanding correct?
Thanks for the reply, Michael.
On Sunday, 21 September 2014 04:27:05 UTC+5:
As there are no temporary tables in cypher, I'd probably write some java
code in a server extension that holds your intermediate results in sets for
the 3,4,5 size cliques and iterate over these sets to find the next largest
cliques.
Michael
On Fri, Oct 3, 2014 at 10:46 PM, Mohana Krishna, IIT B
Michael,
How can I go forward? I wish to store temporary results as my application
does level wise result building. Any way to get/emulate the functionality
of temporary tables/views?? Please give some input.
On Sunday, 21 September 2014 04:27:05 UTC+5:30, Mohana Krishna, IIT Bombay,
India w
I will enumerate all size-3 cliques (all of nodes have relationship with
one another) from existing graph, then using some way (a test on
relationships) I get size-4 cliques (all 4 nodes have edges with one
another again). For size-5 I will use results from size-4 and I don't need
size-3 anymor
What would your algorithm look like if you'd describe it in terms of nodes
(instead of tables) and relationships instead of joins?
Michael
On Sun, Sep 21, 2014 at 3:27 PM, Mohana Krishna, IIT Bombay, India <
mohana...@gmail.com> wrote:
> Hi Michael,
>
> I am developing an apriori sort of algorit
Hi Michael,
I am developing an apriori sort of algorithm where I use previously
constructed tables to generate next level tables. Each level depends on
exactly on the immediate previous level and I do not need level-k tables
once I create level-(k+1) tables.
Being more specific, I am generatin
Hi Michael,
I am developing an apriori sort of algorithm where I use previously
constructed tables to generate next level tables. Each level depends on
exactly on the immediate previous level and I do not need level-k tables
once I create level-(k+1) tables.
Being more specific, I am generatin
10 matches
Mail list logo
