disease-whybot
try to elicit intuitive theories about disease from adaptive natural language prompts
Questions
- D is a disease.
- Next level “causes”:
- Bob has D and this causes him to S.
- Bob has D because he C.
- Bob has D, so he should A.
- Next level “details”:
- Bob should A, but he doesn’t A, because he R.
- 1000 people have D. How many will S?
- 1000 people C. How many will get D?
- Bob has D and he A. How much does it help that he A?
- Bob has D. He should A. How difficult is it for him to do that?
- Next level?
- Bob has D and this causes him to S. It also causes him to S2.
- Bob has D because he C, but also because he C2.
Analsysis
- Do people mention diseases that matter? That are frequent?
- Do people know what the symptoms are? How common they are?
- Do people know what the treatments are? How effective they are?
- Do people know what the causes are? How predictive they are?
Incremental data logging
The file log_data.php contains the initial logic to log data.
Navigating to http://localhost:8000/log_data.php?input={%22some%22:%20%22data%22}&userid=erindb will log {"some": "data"} to the file data/erindb.txt. If the file exists, it will append. If the file does not exist, it will be created.
There's an ajax call inside the experiment now to incrementally log data always all the time.
To Do List
- conjunctions, disjunctions, negation
- e.g. "eats too much and doesn't listen to her doctor or maybe has some precondition"