A na?ve solver is one approach that can be used to identify prospective solutions based on data on(or projected to be on) a Blackboard Architecture’s blackboard. The na?ve solver approach doesn t implement heuristics...A na?ve solver is one approach that can be used to identify prospective solutions based on data on(or projected to be on) a Blackboard Architecture’s blackboard. The na?ve solver approach doesn t implement heuristics or other techniques to determine what solution paths to attempt first. Instead, it runs the blackboard forward(simulating what would occur if data were gradually added to the blackboard at a faster-than-real time rate). The approach doesn t guarantee that an optimal solution will be found and will need to be run repetitively to create multiple solutions for comparison. This paper assesses the effect of pre-pruning the blackboard’s facts and rules to remove those that are not relevant(e.g., facts that cannot be asserted, rules that cannot be triggered) or which produce irrelevant facts and pruning actions that produce irrelevant facts(and/or trigger other similarly useless actions). It describes the Blackboard implementation and its utility, explains the pruning process used and presents quantitative and qualitative assessment of the utility of pruning to a na?ve solver’s operations. This value is extrapolated to facilitate consideration of a more robust pruning process which also removes low-value facts, actions and rules in addition to those being removed due to their uselessness.展开更多
基金supported by a Grant-In-Aid of Research from Sigma Xithe Scientific Research Society,North Dakota EPSCoR(NSF#EPS-814442)+2 种基金a Summer Doctoral Fellowship from the University of North Dakota School of Graduate StudiesFacilities and equipment utilized in this work have been provided by the University of North Dakota Department of Computer ScienceNorth Dakota EPSCo R(NSF#EPS-814442)
文摘A na?ve solver is one approach that can be used to identify prospective solutions based on data on(or projected to be on) a Blackboard Architecture’s blackboard. The na?ve solver approach doesn t implement heuristics or other techniques to determine what solution paths to attempt first. Instead, it runs the blackboard forward(simulating what would occur if data were gradually added to the blackboard at a faster-than-real time rate). The approach doesn t guarantee that an optimal solution will be found and will need to be run repetitively to create multiple solutions for comparison. This paper assesses the effect of pre-pruning the blackboard’s facts and rules to remove those that are not relevant(e.g., facts that cannot be asserted, rules that cannot be triggered) or which produce irrelevant facts and pruning actions that produce irrelevant facts(and/or trigger other similarly useless actions). It describes the Blackboard implementation and its utility, explains the pruning process used and presents quantitative and qualitative assessment of the utility of pruning to a na?ve solver’s operations. This value is extrapolated to facilitate consideration of a more robust pruning process which also removes low-value facts, actions and rules in addition to those being removed due to their uselessness.
