Graph Algorithms or Graph data science Library

Hello, I 've install Graph data science Library ,but while previously link prediction algorithms were working now they dont exist in the database,Should i do something to fix it or to return to Graph algorithms Library is there a difference between them?.My supervisor ask me to do my researh with the last Library for Data Science for my research.I saw that graph data science is experimental and also the link prediction algorithms were working in the Graph algorithms.Which one is the best way to predict edges in a social network and is somenone know why these algorithms dissapeared from my database?

The graph algorithms library has been deprecated, and you should use the graph data science library. The link prediction algorithms from graph algos were all ported to GDS:

The link prediction algorithms in GDS are the same as the ones in graph algorithms, just updated to use the new, standardized syntax. They're labeled alpha/experimental because they have not (yet) been reimplemented by our product engineering team -- for more information on the tiers, see the tier description from our documentation.

Hello thanks for your reply but while the link prediction algorithms exist in my database in Gds Libriary now there isnt and i dont know why this seems to me i have uninstall the library and the problem still there.Although i would prefer to give me your opinion as expert of neo4j about my syrvey because i have been doing all my dissertation in Graph algorithms library.Especially i occupied with all centrality measures,community detection algorithms and link prediction .If there a difference between these Libraries in certain topics .Also i want to ask you why shortest paths algorithms take a lot of time?Particurarly i have 4,612 nodes and 1.300.000 edges and for one node takes 5 minutes to finde the shortest paths.

Στις Δευ, 27 Απρ 2020 στις 4:17 π.μ., ο/η Alicia Frame via Neo4j Online Community έγραψε:

We no longer support, and are no longer updating, the graph algorithms library. The infrastructure to execute algorithms, all product tier algorithms, and the entire API have been rewritten for GDS, and this library includes numerous features not available in the graph algos library (eg. graph mutability and export; K1 coloring and modularity calculations; Neo4j 4.0 compatibility). In general, the GDS is more optimized (20 - 30% faster) and has numerous bug fixes. If you need to port code over, check out our migration guide.

How long shortest path takes depends on many factors: how you load your graph into memory, whether you're using named or anonymous graphs, the structure of the underlying graph, and the path finding algorithm you select (eg. 1:1 versus all pairs shortest path). Check out the "common usage" chapter of our docs for recommended best practices.

It is a simple graph,with ids in nodes and timestamps between edges,which are three types.I think that doesnt have too much information to take so long.Although with new library all the queries became slower.I dont know why ,maybe because i have 3.5.12 version in y neo4j desktop,if i update the version in 4.0 the databases that already existed will be lost ?

Στις Δευ, 27 Απρ 2020 στις 5:01 μ.μ., ο/η Alicia Frame via Neo4j Online Community έγραψε: