The problem is simulated using a discrete event based simulation program and solved on a more practical scale using the heuristic search techniques discussed in the previous sections.
Parsimony analysis was conducted with the program PAUP* 4.0b10, using the heuristic search option [63].
The MP tree [ 68] was obtained using the heuristic search method.
The last two trees were obtained using the heuristic search algorithm.
People use the heuristic way for different reasons, among other things because they are not able to devote enough attention to the information – a common event in noisy daily life.
This indicates the heuristic gives good results, and in the following, we use the heuristic solution for the three instances not solvable by ILP.
MP bootstrap percentages were obtained after 1,000 replications using the same heuristic search strategy using Paup*.
When using the population heuristic, the disambiguator always chooses the most populated place.
We detect two events from commits: "commit" (step 5), i.e., committing an issue-resolving patch to a central repository, and "backout" (step 6), using the same heuristics we used in previous work (Souza et al. 2014, 2015).
Since the metadata heuristic, by itself, cannot be used to disambiguate all the toponyms detected in an article, we used the population heuristic to complement this task.
We used the two heuristics, Population and Distance heuristics, separately to compare their respective precisions.
