Erick

changed the way that SocketNER retrieves data from the socket server

1

`SocketNER.tag_text` would now retrieve data in chunks of size 4092, and bail after a two-second timeout. May address issue #14 and issue #11. It will probably desirable to make the...

See [this page](https://hpsrepository.asu.edu/rest)

The JSTOR DfR reader currently reads in N-gram data, as specified by the user. The reader should also be able to read in full-text content, and transform that content into...

Should consider ways to interoperate with [graph-tool](https://graph-tool.skewed.de/static/doc/quickstart.html). - `tethne.networks` methods can return graph-tool `Graph`s - `gtGraphCollection` - NetworkX -> graph-tool converter

Should explore options for interoperating with iGraph, e.g. - NetworkX -> iGraph converter - `iGraphCollection` - Methods in `tethne.networks` can return igraph `Graph` objects instead of NetworkX `Graph`s.

E.g. only journals for which there is at least one year with a non-zero value.

http://gexf.net/format/

Right now we just rely on the NLTK stoplist and the NLTK Porter stemmer. Users may wish to use different stoplists or stemmers, and we should make that easy.

Writing even a moderately sized graph (~300k nodes, ~900k edges) overloads Neo4j, since it holds the entire transaction on the Java Heap until the very end (stops responding, or throws...

If a node or edge property has value `None`, the JSONEncoder will serialize that value as `null` without complaining. Neo4j chokes on the `null` value, and returns a 500 server...