Plans

Modes

Currently, there’s two modes to display the graph: Python’s WxWidgets interface and a web interface.

There exists a potential to get more information out of the dataset by expanding available modes.

Community detection

I should try several community detection 1 2 methods.

Adjacency graph

A mode for an adjacency graph will require a bit more work, for example, exporting only a top N tags and limit tag lengths so everything can be displayed.

Word2Vec

I am not sure if it can be used as-is, but there were some works that remind me it can be useful to try later. 3 4

Experiments

Edge weighting

Edge weights in pair_mgr are currently divided by an edge_count parameter. I am not sure it is an ideal option that allows to see the maximum amount of details.

Weighting by relation

  • Will add edges between tags like Abstract style and Abstract add more context?

  • How to weight those edges properly?

Argparse

  • Save contents for tag_manager and pair_manager


Footnotes

1

Understanding Community Detection Algorithms with Python NetworkX

2

Louvain

3

Node2vec: Scalable Feature Learning for Networks; How node2vec works — and what it can do that word2vec can’t

4

Paper2vec: Citation-Context Based Document Distributed Representation for Scholar Recommendation by Han Tian and Hankz Hankui Zhuo