<p dir="ltr">Probabilistic Logic Programming and Imprecise Probabilities: Current Work.<span>Theresa Swift</span></p><p dir="ltr">Most formulations of Probabilistic Logic Programming (PLP) rely on<span>point probabilities, usually drawn from small categorical</span><span>distributions. However, imprecise probabilities arise from ignorance,</span><span>from computational approximations such as bounded rationality, or</span><span>other reasons. There have been numerous formulations for imprecise</span><span>probabilities, with PLP over the well-founded semantics a newer and</span><span>sometimes controversial approach.</span></p><p dir="ltr">This presentation considers various topics related to imprecise<span>probabilities and PLP. As a first topic, the use of restraint and the</span><span>well-founded semantics can provide a tractable approximation to</span><span>intractable PLP queries. As a second, the use of Dempster-Shafer</span><span>belief and plausibility functions within PLP can model upper and lower</span><span>bounds in cases of partial ignorance of probability masses. As a</span><span>third, the use of T-norms can provide wide but tractable bounds on</span><span>probabilities. And finally, the relation of PLP under the</span><span>well-founded semantics to the credal stable semantics will be</span><span>discussed.</span></p><p dir="ltr">It should be noted that this work is on-going so that the results<span>presented are provisional.</span></p>
