Wikipedia DB

[freedomtrain]

Thanks so much for your help!!!!
I have set up your fork of Lucida and it is using the Wikipedia DB.
The answers and hit and miss.....Is there a way to improve them ?

Thanks

[JeremyARussell]

The short answer is yes, but it isn't as simple as one might like yet.

The Lucida project already has the feature of adding in knowledge one fact at a time via the Learn page. Through there you can add text only knowledge (which I have tested and confirms works) and image knowledge (which I have not tested).

The long answer.

The QA (question and answering) system uses OpenEphyra, which doesn't seem to have an easy way to add information to a database but it's supposed to be set up so you can analyze documents and websites to generate knowledge databases. If someone was to generate one of those databases it should be a fairly simple matter of adding it to the list of databases used to answer questions, the trick is achieving the prior.

Of course since within the flask web app / command center there is code to add knowledge one fact at a time it should also be fairly trivial to modify the code to implement a knowledge importing feature, but then we would still need to process the data using some sort of crawling/parsing system to provide questions and answers to whatever feature get's implemented to import the qa data anyways.

[EmpireofKings]

I think it would be best to scratch OpenEphyra. The clinc team talked about using yodaQA.
(claritylab#89)
They also forked a yodaQA
(https://github.com/claritylab/yodaqa)

Let me know if you have any questions.

[freedomtrain]

@EmpireofKings @EmpireofKings
Was yodaQA added to lucida ?
Wondering how difficult it would be to add yodaQA to lucida.

[EmpireofKings]

It does appear that @yunshengb did add YodaQA to lucida at one point.
(https://github.com/yunshengb/KuberLucida) *Scroll down and you'll see a .yaml for YodaQA.
If i was a bettin man, i'd say that the repository provided below, is also his attempt at implementing it into lucida. I havent tried it, so im not sure if it works, but it appears to have all the necessary thrift wrappers (Follow the second link to see what im talking about).
(https://github.com/yunshengb/YodaThrift) #1.
(EmpireofKings/YodaThrift@13c65f4) #2.

[EmpireofKings]

It shouldn't be that difficult, don't forget to modify start_all_tmux.sh in the /tools directory, as well as the config.py in the commandcenter.

I'm not sure if yodaQA will be a noticeable improvement standalone. It would probably be wise to use "Ensemble" so you can use YodaQA and OpenEphyra at the same time. Here's the link.
(https://github.com/claritylab/ensemble)