Encoding Schemes [6/6]

[X] iob
[X] bio (iob2)
[X] iobes
[X] bilou
[X] bmewo
[X] Enum of the types

to spans code [2/4]

[X] Code it
[ ] extensive testing (with Hypothesis?)
[X] Use a NamedTuple `Span(type: str, start: int, end: int, tokens: List[int])`
[ ] Programmatic access to errors

transitions code [1/1]

[X] Actually code it

Conversions [10/10]

[X] iob ↔ bio
[X] iob ↔ iobes
[X] iob ↔ bilou
[X] iob ↔ bmeow
[X] bio ↔ iobes
[X] bio ↔ bilou
[X] bio ↔ bmeow
[X] bilou ↔ iobes
[X] bilou ↔ bmeow
[X] iobes ↔ bmeow

Testing [0/3]

[ ] Conversions
[ ] parsing
[ ] transitions

Documentation [0/2]

[-] Docstrings [2/6]
- [ ] __init__
- [ ] parse
- [X] convert
- [X] utils
- [ ] transition
- [ ] write
[ ] Read the docs

(Short) Paper at NLPOSS-2020 [0/2] <2020-07-01 Wed>

[ ] Use it to find bugs in released tagging datasets
[ ] Write the paper lol
The main Idea is that this is a vetted and well tested library for processing and evaluating span based task
Highlight places where I used it to find bugs in published work
Examples of using it to process spans at Work, Use the robust parser to create spans and operate on those instead of actually looking at the token level tags