A task-oriented agent-based mechanism for mathematical assistant systems
Web Intelligence and Agent Systems
| Hi-index | 0.00 |
We present an agent-based mechanism that acts as a mediator modulebetween theorem proving systems and mathematical knowledge basescontaining information that is necessary for the constructions ofproofs. Unlike the more popular user-oriented mediators who work asinformation agents to provide the so-called value-added services tothe collected data before presenting it to users or userapplications, our (multi-)agents are more task-oriented. That is,our agents work in tandem with the user or user application on thetasks the user is trying to solve. This approach is particularlysuitable to mathematical knowledge retrieval in theorem proving as(i) checking for applicable axioms/definitions/theorems from theknowledge base can be done independently from the proof searchprocess concurrently carried out by the prover, and (ii) the proverand the mediator operate on two different search spaces and thesearch outcome brought about by the mediator can be of greatbenefit to the prover, e.g. to avoid the prover from exploring manyunnecessary or irrelevant proof steps, to keep the prover's searchspace more manageable and the constructed proof morecomprehensible.