Difference between revisions of "Vulcan/SystemPrototype"
From Knowitall
(→Status) |
(→Status) |
||
Line 10: | Line 10: | ||
== Status == | == Status == | ||
− | + | Sample MLN programs and output from Tuffy can be found [[Vulcan/SystemPrototype/SampleIO| here]]. | |
+ | * <b>Input: Evidence relating to </b> | ||
+ | * <b>Knowledge:</b> Worked out the facts and rules required. | ||
+ | * <b>Output: </b> The system outputs scores for each query predicate. If query is not in output then score is zero. | ||
− | * <b> | + | * <b>How do you know that it works?</b> |
− | |||
− | |||
<blockquote> | <blockquote> | ||
* In all three examples the correct answer is assigned higher score compared to the incorrect ones. | * In all three examples the correct answer is assigned higher score compared to the incorrect ones. | ||
Line 33: | Line 34: | ||
<blockquote> | <blockquote> | ||
* Use predicates with small arity. For example, avoid writing rules entire nested tuples as predicates. | * Use predicates with small arity. For example, avoid writing rules entire nested tuples as predicates. | ||
− | * For now we can compute this from the score of its components: Score(nested_tuple) = Score(top tuple) * Score (nested). | + | * The only reason we'd need a nested tuple is for the purpose of computing the score. For now we can compute this from the score of its components: Score(nested_tuple) = Score(top tuple) * Score (nested). |
+ | |||
</blockquote> | </blockquote> |
Revision as of 19:42, 27 August 2013
Overview
The prototype is designed to work on three questions. We want the system to output the following:
- Score for the input proposition.
- New facts inferred.
- Facts and rules used in scoring.
Status
Sample MLN programs and output from Tuffy can be found here.
- Input: Evidence relating to
- Knowledge: Worked out the facts and rules required.
- Output: The system outputs scores for each query predicate. If query is not in output then score is zero.
- How do you know that it works?
- In all three examples the correct answer is assigned higher score compared to the incorrect ones.
- Facts inferred by larger number of steps have a lower score compared to those inferred by a smaller number of steps.
- What other diagnostics do we have?
- Inferred facts along with their probabilities.
- Rules that are reachable from the query fact. i.e., Clauses in the MLN that are relevant to the inference of the query fact.
- What diagnostics do we NOT have?
- Connections between the clauses in the MLN.
- A reconstruction/visualization of the MLN network.
- What does this exercise suggest?
- Use predicates with small arity. For example, avoid writing rules entire nested tuples as predicates.
- The only reason we'd need a nested tuple is for the purpose of computing the score. For now we can compute this from the score of its components: Score(nested_tuple) = Score(top tuple) * Score (nested).