Dynamic rule generator

A computer-implemented method comprising:

• receiving an input question with a context about a text;
• determining, using the context as input for a rule generating model, a list of rules that proves the input question, wherein the rule generating model generates one or more rules that leads to the input question;
• determining, based at least in part on the list of rules, a proof graph;
• determining, based at least in part on the proof graph, a complete proof is found;
• and
• returning the complete proof as an answer to the input question.

A system comprising:

• one or more processors;
• and
• one or more computer-readable media storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
• receiving an input query with a context, wherein the context includes a list of facts;
• determining, using the context as input for a rule generating model, a list of rules that proves the input query, wherein the rule generating model generates one or more rules that leads to the input query;
• determining, based at least in part on confidence scores associated with the list of rules, a ranking for the list of rules;
• determining, based at least in part on the ranking for the list of rules, a ranked list of explanations;
• and
• returning the ranked list of explanations for the input query.