which operator to use while loading data from Rapid Miner to neo4j
what operators can be used?
https://community.rapidminer.com/discussion/36711/connecting-neo4j-graph-database-to-rapidminer
You can tell us if it works or not.
Related
I'm working with a large data set that really warrants a graph db. My goal is to visualize identify trends in the data set to make decisions.
I'm currently using neo4j and i really like the tool, however the nodes returned are capped at 300. This number is only a fraction of my data, and doesnt really allow me to gain the insight i've been looking for, even with queries to filter out portions. Additionally, I'd really like to add node weights and color per conditions, which isn't possible using just neo4j.
Has anybody found a solution to this problem. I'd imagine there may be some client side libraries designed for these sorts of problems. Alternatively, I wouldn't be opposed to switching to some other graph db better suited to solve these problems.
I would suggest using Neo4j Bloom. This will provide you better visualization of your Neo4j data.
I am learning neo4j. I am accessing neo4j via REST api(s) supported by the server mode. CRUD operations are implemented using neo4jOperations. For experimentation , I have benchmarked its read operations but I have found that methods : 'query' and 'queryForObjects' are taking huge execution time, although I am querying via a field which is indexed. Traversals are not complex.
I have : around 500K+ nodes, 900K+ relationships.
neo4j version : 3.0.8.
Is there any solution to improve the performance of query on neo4j in server mode?
Without looking at your actual queries and model it is hard to say why the performance would not be up to your expectations. Try to run the queries through the Neo4j browser and either EXPLAIN or PROFILE them, that may give you a hint of where the issue is.
Having said that, you really should move to version 3.2.1 and access the server over the bolt:/ protocol. That by itself should already significantly improve things.
Regards,
Tom
I am wondering how to use Neo4j to find the MST? Most examplesI found was using Hadoop to find it.
I don't think that this is possible in Cypher, given how current algorithms determine an MST (if I'm wrong on this, I'd love to know).
Instead, I'd recommend implementing one of the algorithms used for determining an MST, e.g. Prim's Algorithm. It's quite straight forward and, with the help of heaps and adjacency lists, is relatively performant.
A quick search for the algorithm will turn up many links.
I'm sure leveraging Neo4j's Core API or Traversal API might even help things integrate even more closely, possibly without needing to represent the entire graph as an adjacency list first. And of course you can do that with Neo4j in Embedded Mode or turn it into a Server Plugin in case you're running Neo4j in Server Mode.
Let us know what you come up with!
I would be really grateful if someone show me the way to calculate minimum path with a dynamic programming algorithm like Floyd and Warshall.The algorithm has to calculate the path at every interaction,it has to making decision about which nodes choose taking in consideration the nodes already traversed.
I have made a little explaination:
https://drive.google.com/file/d/0B3i9KFQXzB89YXl0VkEzaDZDMHc/edit?usp=sharing
My graph is stored in a neo4j environment and it can increase in a serious way his dimension. I use rest with everyman php neo4j libraries.
What is the best way to do this?
Traversal, cypher, gremilins, code a custom algorithm starting from http://components.neo4j.org/neo4j-graph-algo/1.4/xref/org/neo4j/graphalgo/impl/shortestpath/FloydWarshall.html ?
Tnx in advance
At the end my solution was to use traversal framework and build unmanaged extension.
M.
I'm testing Neo4J as a potentially more efficient alternative to (non-distributed) Mahout for item-based collaborative filtering (i.e. 'recommend items for User based on his and others' preferences), and have seen excellent examples using Gremlin, e.g. http://markorodriguez.com/2011/09/22/a-graph-based-movie-recommender-engine/ - but none for Cypher.
Is this practical/feasible with Cypher, or should I just bite the bullet and start using Gremlin (and the REST API)?
We've done these kinds of recommendations using Neo4j and Cypher in particular, are are really pleased with the results.
Of course it could span into a couple of Cypher queries depending on the complexity of logic you have but it's entirely do-able.
I realize this is a extremely simplified approach but it might help you compare a gremlin/cypher kind of approach:
Gremlin: http://blog.everymansoftware.com/2012/02/similarity-based-recommendation-engines.html
Cypher: http://thought-bytes.blogspot.in/2012/02/similarity-based-recommendations-with.html
Disclaimer: I am the author of the Thought Bytes post
have you tried my open source project? It's name is reco4j, it is a graph based recommender engine based on neo4j as graph database.
It is in an early stage but it works for your use case.
Cheers,
Alessandro