Organic Synthesis is a computationally challenging practical problem concerned with constructing a target molecule from a set of initially available molecules via chemical reactions. This paper demonstrates how organic synthesis can be formulated as a planning problem in Artificial Intelligence, and how it can be explored using the state-of-the-art domain independent planners. To this end, we develop a methodology to represent chemical molecules and generic reactions in PDDL 2.2, a version of the standardized Planning Domain Definition Language popular in AI. In our model, derived predicates define common functional groups and chemical classes in chemistry, and actions correspond to generic chemical reactions. We develop a set of benchmark problems. Since PDDL is supported as an input language by many modern planners, our benchmark can be subsequently useful for empirical assessment of the performance of various state-of-the-art planners.
This thesis formulates organic chemistry synthesis problems as Artificial Intelligence planning problems and uses a combination of techniques developed in the field of planning to solve organic synthesis problems. To this end, a methodology for axiomatizing organic chemistry is developed, which includes axiomatizing molecules and functional groups, as well as two approaches for representing chemical reactions in a logical language amenable to reasoning. A novel algorithm for planning specific to organic chemistry is further developed, based on which a planner capable of identifying 75 functional groups and chemical classes is implemented with a knowledge base of 55 generic chemical reactions. The performance of the planner is empirically evaluated on two sets of benchmark problems and analytically compared with a number of competing algorithms.
v
This thesis formulates organic chemistry synthesis problems as Artificial Intelligence planning problems and uses a combination of techniques developed in the field of planning to solve organic synthesis problems. To this end, a methodology for axiomatizing organic chemistry is developed, which includes axiomatizing molecules and functional groups, as well as two approaches for representing chemical reactions in a logical language amenable to reasoning. A novel algorithm for planning specific to organic chemistry is further developed, based on which a planner capable of identifying 75 functional groups and chemical classes is implemented with a knowledge base of 55 generic chemical reactions. The performance of the planner is empirically evaluated on two sets of benchmark problems and analytically compared with a number of competing algorithms.
v
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.