I just created a "super" label and a constraint on the id of its nodes.
Just to make it more clear, at first i had 16 labels (which could be
represented by the roles of each person inside a company) and only one
relationship between 2 different nodes (which is of only one kind, DIRECTS
for example). The "super" label i created now is something like
"MemberOfCompany" and added a constraint on it
Should i add ON CREATE before SET and CALL the addLabels method or is it
going to be redundant/disruptive?
When i need to MERGE relationships this is the code i use
String query =
"UNWIND $relationships AS relationship " +
"MATCH (x {id: relationship.source_id}), (y {id:
relationship.target_id}) " +
"CALL apoc.merge.relationship(x, relationship.type,
{source_id: relationship.source_id, target_id: relationship.target_id,
type: relationship.type}, relationship, y) YIELD rel RETURN rel";
session.run( query, params );
I'm suspecting this could be highly inefficient aswell, especially since i
dont have a unique id for each relationship and i use a combo of source,
target and type to determine its uniqueness. Is there anything here i could
do differently to improve the efficiency and performances of this part?
Thanks again
Il giorno giovedì 7 febbraio 2019 22:47:16 UTC+1, eve.freema ha scritto:
>
> Sure thing. Let us know how it works out. :)
>
> On Thu, Feb 7, 2019 at 4:43 PM Andrea Balzoni <[email protected]
> <javascript:>> wrote:
>
>> Thanks a lot for your input, i'll surely try what you suggested and add a
>> constraint to my schema :)
>>
>> Thankd again for taking your time to answer
>>
>> Il giorno gio 7 feb 2019, 22:04 Eve Freeman <[email protected]
>> <javascript:>> ha scritto:
>>
>>> Something like (these are totally untested, sorry for bugs)--also, this
>>> label AllCategories should probably be more what your entities actually are:
>>>
>>> create constraint on (allCategories:AllCategories) assert
>>> allCategories.id is unique;
>>>
>>> UNWIND $nodes AS n
>>> MERGE (node:AllCategories {id: n.id})
>>> SET node += n
>>> RETURN node
>>>
>>> If you want to retain category as a label, you could add it in a second
>>> part with something like:
>>> UNWIND $nodes AS n
>>> MERGE (node:AllCategories {id: n.id})
>>> SET node += n
>>> call apoc.create.addLabels([ id(node) ], [ n.category ]) yield node
>>> RETURN node
>>>
>>> On Thu, Feb 7, 2019 at 3:52 PM Andrea Balzoni <[email protected]
>>> <javascript:>> wrote:
>>>
>>>> Ids are supposed to be globally unique, regardless of their label
>>>> (again, i'm not sure this is the best approach, but i thought it was the
>>>> one that made more sense), i'm not sure what kind of info i could provide
>>>> so feel free to ask...
>>>>
>>>> Il giorno gio 7 feb 2019, 21:24 Eve Freeman <[email protected]
>>>> <javascript:>> ha scritto:
>>>>
>>>>> Whoops, "if so" should be "if not".
>>>>>
>>>>> On Thu, Feb 7, 2019 at 3:22 PM Eve Freeman <[email protected]
>>>>> <javascript:>> wrote:
>>>>>
>>>>>> Are the ids unique to each label? If so, maybe you can merge on one
>>>>>> "inclusive" label and then add an additional category label, and then
>>>>>> you
>>>>>> can have just one constraint. No idea what you're doing here, so giving
>>>>>> my
>>>>>> best guess.
>>>>>>
>>>>>> On Thu, Feb 7, 2019 at 2:40 PM Andrea Balzoni <
>>>>>> [email protected] <javascript:>> wrote:
>>>>>>
>>>>>>> That sounds like a lot of constraints or am i wrong?
>>>>>>>
>>>>>>> Il giorno gio 7 feb 2019, 20:19 Eve Freeman <[email protected]
>>>>>>> <javascript:>> ha scritto:
>>>>>>>
>>>>>>>> Yes. You would need to add a constraint for each label + id
>>>>>>>> combination.
>>>>>>>>
>>>>>>>> On Thu, Feb 7, 2019 at 1:57 PM Andrea Balzoni <
>>>>>>>> [email protected] <javascript:>> wrote:
>>>>>>>>
>>>>>>>>> Sadly the ids i'm trying to add are not going to be necessarily
>>>>>>>>> unique, hence why i'm using merge. I think i grasped what a constrant
>>>>>>>>> is,
>>>>>>>>> but i cant say the same about indexes. So you are telling me that
>>>>>>>>> adding a
>>>>>>>>> constraint on ids would make the query faster? Sorry about questions
>>>>>>>>> that
>>>>>>>>> might sound stupid, but i'm fairly new to neo4j...
>>>>>>>>>
>>>>>>>>> Il giorno gio 7 feb 2019, 18:59 Eve Freeman <[email protected]
>>>>>>>>> <javascript:>> ha scritto:
>>>>>>>>>
>>>>>>>>>> One other note--if you know these are unique ids, you can avoid
>>>>>>>>>> using merge, and just use create, which will be quicker still.
>>>>>>>>>>
>>>>>>>>>> On Thu, Feb 7, 2019 at 12:54 PM Eve Freeman <[email protected]
>>>>>>>>>> <javascript:>> wrote:
>>>>>>>>>>
>>>>>>>>>>> Well, what a merge will do is check whether something exists
>>>>>>>>>>> before it creates it, "match or create". The match step will get
>>>>>>>>>>> linearly
>>>>>>>>>>> slower as it has to scan larger and larger lists of nodes on a
>>>>>>>>>>> label. If
>>>>>>>>>>> you add an index (at minimum), or unique constraint/node key
>>>>>>>>>>> (preferred,
>>>>>>>>>>> especially if you require uniqueness and will eventually have
>>>>>>>>>>> concurrent
>>>>>>>>>>> requests of this nature), it will be able to do the match step much
>>>>>>>>>>> quicker.
>>>>>>>>>>>
>>>>>>>>>>> On Thu, Feb 7, 2019 at 12:50 PM Andrea Balzoni <
>>>>>>>>>>> [email protected] <javascript:>> wrote:
>>>>>>>>>>>
>>>>>>>>>>>> Honestly i heard of those, but i'm not using any. I'm quite new
>>>>>>>>>>>> to it, so i wasnt sure what their purpose is.
>>>>>>>>>>>>
>>>>>>>>>>>> Il giorno gio 7 feb 2019, 18:43 Eve Freeman <[email protected]
>>>>>>>>>>>> <javascript:>> ha scritto:
>>>>>>>>>>>>
>>>>>>>>>>>>> What's your schema? Do you have a unique constraint or index
>>>>>>>>>>>>> on the labels + id you're merging on?
>>>>>>>>>>>>>
>>>>>>>>>>>>> Eve
>>>>>>>>>>>>>
>>>>>>>>>>>>> On Thu, Feb 7, 2019 at 12:23 PM <[email protected]
>>>>>>>>>>>>> <javascript:>> wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>>> Hello everyone, i'm trying to MERGE thousands of nodes using
>>>>>>>>>>>>>> UNWIND and apoc procedures (for a dynamic label creation) using
>>>>>>>>>>>>>> Java Driver
>>>>>>>>>>>>>> as follows:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> String query =
>>>>>>>>>>>>>> "UNWIND $nodes AS n " +
>>>>>>>>>>>>>> "CALL apoc.merge.node([n.category], {id: n.id},
>>>>>>>>>>>>>> n) YIELD node RETURN node";
>>>>>>>>>>>>>> session.run( query, params );
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> My database contains something like 150k nodes divided by 16
>>>>>>>>>>>>>> labels and 130k relationships of the same type. It worked fine
>>>>>>>>>>>>>> so far, it
>>>>>>>>>>>>>> never failed to deliver, but now that i'm trying to create
>>>>>>>>>>>>>> something like
>>>>>>>>>>>>>> 4k new nodes and 5k new relationships at once it took more than
>>>>>>>>>>>>>> 30 minutes.
>>>>>>>>>>>>>> I feel like it should be way faster than this.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Am i doing anything wrong? Is it really supposed to be this
>>>>>>>>>>>>>> slow? Is there anything i can do to make this process smoother
>>>>>>>>>>>>>> and to
>>>>>>>>>>>>>> improve performances?
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Thanks in advance
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> --
>>>>>>>>>>>>>> You received this message because you are subscribed to the
>>>>>>>>>>>>>> Google Groups "Neo4j" group.
>>>>>>>>>>>>>> To unsubscribe from this group and stop receiving emails from
>>>>>>>>>>>>>> it, send an email to [email protected]
>>>>>>>>>>>>>> <javascript:>.
>>>>>>>>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>>>>>>>>
>>>>>>>>>>>>> --
>>>>>>>>>>>>> You received this message because you are subscribed to the
>>>>>>>>>>>>> Google Groups "Neo4j" group.
>>>>>>>>>>>>> To unsubscribe from this group and stop receiving emails from
>>>>>>>>>>>>> it, send an email to [email protected]
>>>>>>>>>>>>> <javascript:>.
>>>>>>>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>>>>>>>
>>>>>>>>>>>> --
>>>>>>>>>>>> You received this message because you are subscribed to the
>>>>>>>>>>>> Google Groups "Neo4j" group.
>>>>>>>>>>>> To unsubscribe from this group and stop receiving emails from
>>>>>>>>>>>> it, send an email to [email protected] <javascript:>
>>>>>>>>>>>> .
>>>>>>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>>>>>>
>>>>>>>>>>> --
>>>>>>>>>> You received this message because you are subscribed to the
>>>>>>>>>> Google Groups "Neo4j" group.
>>>>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>>>>> send an email to [email protected] <javascript:>.
>>>>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> You received this message because you are subscribed to the Google
>>>>>>>>> Groups "Neo4j" group.
>>>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>>>> send an email to [email protected] <javascript:>.
>>>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>>>
>>>>>>>> --
>>>>>>>> You received this message because you are subscribed to the Google
>>>>>>>> Groups "Neo4j" group.
>>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>>> send an email to [email protected] <javascript:>.
>>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>>
>>>>>>> --
>>>>>>> You received this message because you are subscribed to the Google
>>>>>>> Groups "Neo4j" group.
>>>>>>> To unsubscribe from this group and stop receiving emails from it,
>>>>>>> send an email to [email protected] <javascript:>.
>>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>>
>>>>>> --
>>>>> You received this message because you are subscribed to the Google
>>>>> Groups "Neo4j" group.
>>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>>> an email to [email protected] <javascript:>.
>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google
>>>> Groups "Neo4j" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send
>>>> an email to [email protected] <javascript:>.
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "Neo4j" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to [email protected] <javascript:>.
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "Neo4j" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [email protected] <javascript:>.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
--
You received this message because you are subscribed to the Google Groups
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
