graphinius
graphinius copied to clipboard
Graphinius
Generic graph (analysis) library in Typescript
- [ ] work out and polish neo4j based examples - [ ] write Medium article - [ ] post in every Graph, ML, AI, Data Science & Software group...
See title: - [ ] do this with nodes (make consistent with TypedNode Logic, or even abstract out to BaseNode) - [ ] same with graphs Also: - [x] design...
- [ ] go through issues - [ ] go through documentation (JSDoc) - [ ] decide on format of design document - [ ] add all design decisions &...
Todos - [ ] neighborhood definition - [ ] counting rules - [ ] RW parameters - [ ] find w2v JS / WASM / TF.js implementation to feed RW...
- otherwise any change in the order we instantiate nodes will break our test cases, which is really unacceptable !!
with SharedArrayBuffer & Atomics - [ ] (de)-serialize our PR Array datastructure via SharedArrayBuffer - [ ] implement "fast distributed PR" from paper - [ ] measure performance on different...
- [ ] Theory incl. documentation / diagrams - [ ] WebWorker based (just for validation on super-small graphs) - [ ] WASM (C++ / Rust) - [ ] TF.js...
We have a slight difference in results between our PageRank and NetworkX pagerank, since we are "pulling" ranks iteratively, while NetworkX seems to use the canonical google matrix... - Try...
- [ ] only built on typed arrays - [ ] holding an adjlist representation of the BaseGraph - [ ] updated in sync with the BaseGraph
The node strength is just the sum of the weights of all the adjacent edges (or DEFAULT if unweighted, which in case of DEFAULT==1 would just be the all_degree). Reason:...