So, the community bonding period has finally come to an end and the coding phases will start, more fun follows now!
I already started a bit of coding, I ported some of the plugins.
Things done in this week:
1. Added the Dockerfile.
@yukiisbored created the dockerfile for corobo. He is our devops guy at coala, he does the awesome work of deploying coala website, webservices, blog, gitmate etc. He is awesome!
2. The test suite for the repo has been set.
Tests are written in the
tests/ directory. Errbot comes with a pytest
testbot that can be used in testing. The fixture is working
3. Start contributing
Since the repository structure is already laid and the test suite has already been made, outside contributions on various plugins are highly encouraged, for a quick guide headout to our wiki.
I did some documentation fixes and fixed one wrong assertion upstream in the errbot repo. The gitter channel is quite active, it is helpful to have support upstream :D
5. Regarding topic modeling:
After recognizing the topic for the question, the next thing is to get the answer from the document. This part is what I call “Passage retrieval”. This was a bit difficult one. There are many possible alternatives.
- We can search the documentation with the topic obtained from the question using a search engine.
- Use tfidf on the documentation with topic and use the one with the highest score as the appropriate doc.
6. Next step
After retrieving the documentation, other thing would be linking to the documentation and ideally summarizing the same. For summarizing we would be using the summarizer available in the gensim already.(Most probably, subjected to change.)