Hi Michael,
Thanks for yo kind reply. I'm testing your code. As I replied earlier, I already swithed to Spark Graphx and its connected components algorithm perfectly solved my problem. Besides, Spark can read files on HDFS and run on clusters. It only took few seconds to accomplish my task. So I just temporarily put neo4j aside and thank you again for your answer! Best Regards, Guitao On Thursday, July 10, 2014 9:05:58 PM UTC+8, Michael Hunger wrote: > > > > match (n:user_id) > match (n)-[:mapping*]-(d:user_id) > return n.value as user_id, > case when min(d.value) > n.value then n.value else min(d.value) end as > people_id > > > > > could you also try to run: > > match (n:user_id) > match (n)-[:mapping*]-(d:user_id) > RETURN count(*) > > Would you be able to share your Neo4j database with me? > > Thanks, > > Michael > > On Thu, Jul 10, 2014 at 3:05 AM, Guitao Ding <[email protected] > <javascript:>> wrote: > >> >> Hi Michael, >> >> The data is from one website. The user_id is the ID of each registered >> user. And every user_id is linked to one or more cookie_ids (the value of >> one cookie). >> Also one cookie_id is linked to one or more user_ids. So user_id and >> cookie_id are many to many mapping. My case is to find all user_ids linked >> with each other (via cookie_ids, and no limit on the path length) and >> assign them one unique ID (here i used then smallest user_id in the path). >> >> for example: >> All user_ids (user_id1, user_id2, user_id3) in the following path should >> be assigned one same ID (e.g. user_id1) >> user_id1----cookie_id1----user_id2-----cookie-id2-----user_id3 >> >> I imported the data to neo4j and use different labels for user_id and >> cookie_id. There is only one type of relationship (cookie_id -- user_id) >> and the direction doesn't matter. >> >> On Thursday, July 10, 2014 6:01:14 AM UTC+8, Michael Hunger wrote: >> >>> This is a graph global query, with unlimited paths, so it might generate >>> many billions or trillions of paths to look at. >>> Esp. if you don't provide a direction. >>> >>> if your nodes are all users, then you do the equivalent of finding all >>> paths between the cross product of 16M^2 >>> >>> Perhaps you can describe your actual use-case that you try to solve? >>> >>> Michael >>> >>> >>> >>> On Wed, Jul 9, 2014 at 4:23 PM, Guitao Ding <[email protected]> >>> wrote: >>> >>>> Hi all, >>>> >>>> I'm leaning to user neo4j for relation analysis recently. Today I found >>>> it took too long for my cypher query took to finish. >>>> >>>> I used the batch importer >>>> <https://github.com/jexp/batch-import/tree/20> to import all data into >>>> neo4j. And I wanted to find the smallest user_id connected (directly or >>>> undirectly, path length no limit) to each user_id. Below is the details: >>>> >>>> neo4j version: 2.1.2 >>>> nodes num: 16M (three labels) >>>> relation num: 10M >>>> cypher query: >>>> >>>> match (n:user_id)-[:mapping*]-(d:user_id) >>>> with n.value as user_id, >>>> case when min(d.value) > n.value then n.value else min(d.value) end as >>>> people_id >>>> return user_id, people_id >>>> >>>> >>>> What should I do to improve my query performance? Any suggestions would >>>> be appreciated! >>>> >>>> Thanks in advance. >>>> >>>> Guitao >>>> >>>> -- >>>> 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. >>>> >>> >>> -- >> 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.
