Adding edges to existing vertices

[Adhitya Rajagopalan]

Hi,

I have a Titan 1.0 graph around 120K vertices with 30 properties each (running with Cassandra storage backend and ElasticSearch for indexing). Now, I need to establish the edges between the vertices based on the properties of the edges.

Eg) The vertices have a properties called level which has values like 1.1,1.2,1.3..... and 1.1.1, 1.1.2, 1.2.1, 1.2.2.......

I have to create edges between {1.1 and 1.1.1}, {1.1 and 1.1.2}, {1.2 and 1.2.1}, {1.2 and 1.2.2}......

How do I achieve this?

Thanks in advance

Yours,

Adhitya 

[Daniel Kuppitz]

Given this sample graph:

g.addV(id, "1").
  addV(id, "2").
  addV(id, "2.1").
  addV(id, "1.1").
  addV(id, "1.2").
  addV(id, "1.1.1")

...the following query worked pretty well:

gremlin> g.V().as("a").V().as("b").filter(
             select("a","b").by(id).
               where("b",gt("a")).
               filter {
                 ids = it.get()
                 ids["a"].length() < ids["b"].length() &&
                 ids["b"].substring(0, ids["a"].length()).equals(ids["a"]) &&
                 ids["b"].substring(ids["a"].length()) ==~ /\.[^.]*/
               }).addE("link").from("a").to("b")
==>e[0][1-link->1.1]
==>e[1][1-link->1.2]
==>e[2][2-link->2.1]
==>e[3][1.1-link->1.1.1]

Of course you can just use the level property instead of the id. And btw., I don't think it can be done without lambdas.

Cheers,

Daniel

--
You received this message because you are subscribed to the Google Groups "Aurelius" group.
To unsubscribe from this group and stop receiving emails from it, send an email to aureliusgraph...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/aureliusgraphs/cf28c523-fe1c-4609-be37-5e1d3a4a1f51%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Adhitya Rajagopalan]

This is the code that I have right now:

conf = new BaseConfiguration()

conf.setProperty("gremlin.graph","com.thinkaurelius.titan.core.TitanFactory")

conf.setProperty("storage.backend", "cassandra")

conf.setProperty("storage.hostname", "127.0.0.1")

conf.setProperty("storage.batch-loading", true)

conf.setProperty("storage.cassandra.keyspace", "graph")

graph = GraphFactory.open(conf)

g = graph.traversal()

m = graph.openManagement()

PropertyKey pk1= m.makePropertyKey("pk1").dataType(String.class).make();

PropertyKey pk2= m.makePropertyKey("pk2").dataType(String.class).make();

PropertyKey pk3= m.makePropertyKey("pk3").dataType(String.class).make();

PropertyKey pk4= m.makePropertyKey("pk4").dataType(String.class).make();

PropertyKey pk5= m.makePropertyKey("pk5").dataType(String.class).make();

PropertyKey CFGACH_LEVEL_CODE= m.makePropertyKey("CFGACH_LEVEL_CODE").dataType(String.class).make();

PropertyKey pk7= m.makePropertyKey("pk7").dataType(String.class).make();

PropertyKey pk8= m.makePropertyKey("pk8").dataType(String.class).make();

PropertyKey pk9= m.makePropertyKey("pk9").dataType(String.class).make();

PropertyKey pk10= m.makePropertyKey("pk10").dataType(String.class).make();

PropertyKey pk11= m.makePropertyKey("pk11").dataType(String.class).make();

m.commit()

batchSize = 10000

counter = new java.util.concurrent.atomic.AtomicLong()

cache = [:]

mutate = { ->

  if (0 == counter.incrementAndGet() % batchSize) {

    graph.tx().commit()

  }

}

new File("C:/Users/PJT755/data.csv").eachLine { def line ->

  def (id1, id2, id3, id4, id5, id6, id7, id8, id9, id10, id11) = line.split(",", 11)

v = graph.addVertex("CFGACH_LEVEL_CODE", id6,"pk1", id1, "pk2", id2, "pk3", id3,"pk4",id4, "pk5",id5, "pk7", id7, "pk8", id8, "pk9", id9, "pk10", id10, "pk11", id11)

   mutate()  

};

graph.tx().commit();

g.V().as("a").V().as("b").filter(

             select("a","b").by(CFGACH_LEVEL_CODE).

               where("b",gt("a")).

               filter {

                 ids = it.get()

                 ids["a"].length() < ids["b"].length() &&

                 ids["b"].substring(0, ids["a"].length()).equals(ids["a"]) &&

                 ids["b"].substring(ids["a"].length()) ==~ /\.[^.]*/

               }).addE("link").from("a").to("b")

I am using Titan 1.0.0 with storage backend as Cassandra.

[Adhitya Rajagopalan]

Thanks a lot for your reply. When I run the code that you had posted, I get the following error:

No signature of method: org.apache.tinkerpop.gremlin.process.traversal.dsl.graph

.DefaultGraphTraversal.V() is applicable for argument types: () values: []

Possible solutions: is(java.lang.Object), by(groovy.lang.Closure), any(), min(),max(), sum()

Display stack trace? [yN]

How do I overcome this problem?

On Friday, July 22, 2016 at 12:07:58 PM UTC+5:30, Adhitya Rajagopalan wrote:

[Daniel Kuppitz]

Ah, I always run into this trap.  Titan doesn't support mid-traversal V()'s (yet). You can try this instead 

g.V().as("a").aggregate("x").select("x").unfold().as("b").filter(

  ...

Cheers,

Daniel

--

You received this message because you are subscribed to the Google Groups "Aurelius" group.
To unsubscribe from this group and stop receiving emails from it, send an email to aureliusgraph...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/aureliusgraphs/33f76edc-0945-440e-ae79-521de7e035d1%40googlegroups.com.

[Adhitya Rajagopalan]

Even after changing the code to:

g.V().as("a").aggregate("x").select("x").unfold().as("b").filter(

             select("a","b").by("CFGACH_LEVEL_CODE").

               where("b",gt("a")).

               filter {

                 ids = it.get()

                 ids["a"].length() < ids["b"].length() &&

                 ids["b"].substring(0, ids["a"].length()).equals(ids["a"]) &&

                 ids["b"].substring(ids["a"].length()) ==~ /\.[^.]*/

               }).addE("link").from("a").to("b")

I get the same error. Should the property key be in quotes?

[Adhitya Rajagopalan]

Sorry. This is the new error I got

No signature of method: org.apache.tinkerpop.gremlin.process.traversal.dsl.graph

.DefaultGraphTraversal.addE() is applicable for argument types: (java.lang.Strin

g) values: [link] 

On Monday, July 25, 2016 at 2:45:07 PM UTC+5:30, Daniel Kuppitz wrote:

[Daniel Kuppitz]

This should work in 3.0.x:

.addOutE("a", "link", "b")

...instead of  .addE("link").from("a").to("b") (just in case it wasn't obvious).

See: http://tinkerpop.apache.org/docs/3.0.0-incubating/#addedge-step

Cheers,

Daniel

To view this discussion on the web visit https://groups.google.com/d/msgid/aureliusgraphs/1e48cc65-f30e-41e1-b294-e1921a0959f6%40googlegroups.com.

[f201...@dubai.bits-pilani.ac.in]

hello Daniel

how would the code change if the edges have to be created based on multiple vertex properties?

 Also pls explain what this code does (right of the ==):

ids["b"].substring(ids["a"].length()) ==~ /\.[^.]*/

Yours

Shaswat

[Daniel Kuppitz]

how would the code change if the edges have to be created based on multiple vertex properties?

You mean for the by() part? You can use whatever you want (e.g. values("prop1","prop2") or valueMap("prop1","prop2"), whatever works best for you.

Also pls explain what this code does (right of the ==):

It cuts off the the first property value from the second and then ensures that the remainder starts with a period, but then contains no more periods.

For example:

First value:             AAA.BBB

Second value:            AAA.BBB.CCC

Remainder after the cut:        .CCC

Matches '\.[^.]*':       Yes

First value:              AAA.BBB

Second value:             AAA.BBB.CCC.DDD

Remainder after the cut:         .CCC.DDD

Matches '\.[^.]*':        No

Cheers,

Daniel

To view this discussion on the web visit https://groups.google.com/d/msgid/aureliusgraphs/60d89d2c-e0b2-4cdc-a932-14af0042de05%40googlegroups.com.

[SHASWAT ANANDHAN]

HI Daniel,

               Thank you for your explanation.

 By multiple parameters I mean the following:

  In addition to the level_code, if I have another vertex property key, say prop_key, edges should be established only if 'a.prop_key=b.prop_key' and also the above level_code hierarchy.

eg.   Suppose we have,

        Vertex1={(prop_key=2),......,(level_code=1.1),..}

        Vertex2={(prop_key=3),......,(level_code=1.1),..}

        Vertex3={(prop_key=2),......,(level_code=1.1.1),..}

        Vertex4={(prop_key=2),......,(level_code=1.1.2),..}

        Vertex5={(prop_key=2),......,(level_code=1.1.3),..}

        Vertex6={(prop_key=3),......,(level_code=1.1),..}

On Friday, July 22, 2016 at 12:07:58 PM UTC+5:30, Adhitya Rajagopalan wrote:

[SHASWAT ANANDHAN]

In continuation to the previous reply, for the vertex set in the example,

            edges must be created between 

            {Vertex1,Vertex3} and {Vertex1,Vertex4} and {Vertex1,Vertex5}..... 

There should be NO vertex between {Vertex2,Vertex3} etc.

On Friday, July 22, 2016 at 12:07:58 PM UTC+5:30, Adhitya Rajagopalan wrote:

[Daniel Kuppitz]

You can just use another filter step:

g.V().as("a").aggregate("x").select("x").unfold().as("b").

  filter(select("a","b").by("prop_key").where("a", eq("b"))).

  filter(select("a","b").by("level_code").

           where("b",gt("a")).

           filter {

             ids = it.get()

             ids["a"].length() < ids["b"].length() &&

             ids["b"].substring(0, ids["a"].length()).equals(ids["a"]) &&

             ids["b"].substring(ids["a"].length()) ==~ /\.[^.]*/

           }).

  addOutE("a", "link", "b")

Cheers,

Daniel

--

You received this message because you are subscribed to the Google Groups "Aurelius" group.
To unsubscribe from this group and stop receiving emails from it, send an email to aureliusgraph...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/aureliusgraphs/7f69e613-e665-4f2b-a55f-81e951f8ea04%40googlegroups.com.

[SHASWAT ANANDHAN]

When I use this method, exactly 101 edges get created. I am sure that more than 10K edges should be created. 

[Daniel Kuppitz]

Can you show the properties of 2 vertices that were not properly linked?

Cheers,

Daniel

To view this discussion on the web visit https://groups.google.com/d/msgid/aureliusgraphs/e544836d-95d3-499e-afde-c94114017741%40googlegroups.com.

[SHASWAT ANANDHAN]

Dear Daniel,

                  These are the properties of the 2 vertices between which a edge should have been created!. But I find that no edge has been created between these 2 vertices. 

gremlin> g.V(41087024).valueMap()

==>[CDI:[NULL], CPC:[AGB], CAS:[A], CNI:[ENG-208], CPO:[8], CR:[0], CDP:[NULL], CCI:[ENG-218], PART_NO:[N515A], CPS:[A], CSN:[N515A-3019], CPPN:[340-046-60-20], CN:[340-046-50-40], CNN:[0], CAN:[7200], CCC:[%], CFD:[2012-02-07 00:00:00.000], CZ:[210], CAO:[8], CPT:[O], COE:[8], CFGACH_LEVEL_CODE:[1.46.3.1], CLN:[3], CARF:[N], CRMF:[Y]]

gremlin> g.V(83017848).valueMap()

==>[CDI:[NULL], CPC:[AGBA], CAS:[A],CNI:[ENG-206], CPO:[8], CR:[0], CDP:[NULL], CCI:[ENG-208], PART_NO:[N515A], CPS:[A], CSN:[N515A-2020], CPPN:[340-046-50-40], CN:[CFM56-5B], CNN:[0], CAN:[7200], CCC:[%], CFD:[2012-02-07 00:00:00.000], CZ:[210], CAO:[8], CPT:[O], COE:[8],CFGACH_LEVEL_CODE:[1.46.3], CLN:[2], CARF:[N], CRMF:[N]]

  
Cheers

Shaswat

[Daniel Kuppitz]

Hmm, that's odd, the clearly match the pattern. And you're sure that the transaction didn't fail? Maybe just do a count and see how many matches you get. If the number is higher than the number of edges that were created, we need to think about another strategy. Maybe write the id pairs into a file and then process them afterwards and frequently commit after X new edges.

The count-query:

g.V().as("a").aggregate("x").select("x").unfold().as("b").

  filter(select("a","b").by("PART_NO").where("a", eq("b"))).

  filter(select("a","b").by("CFGACH_LEVEL_CODE").

           where("b", gt("a")).

           filter {

             ids = it.get()

             ids["a"].length() < ids["b"].length() &&

             ids["b"].substring(0, ids["a"].length()).equals(ids["a"]) &&

             ids["b"].substring(ids["a"].length()) ==~ /\.[^.]*/

           }).count()

Cheers,

Daniel

To view this discussion on the web visit https://groups.google.com/d/msgid/aureliusgraphs/86351fca-5398-4212-a7e7-23c47b654e5a%40googlegroups.com.