Master Degree in Computer Science, University of Bari. Case study for Artificial Intelligence course. I work on this project with my classmate (and dear friend) Giovanni Mastroscianni.
Prolog version used: YAP 6.2.2
-
family.datis the dataset on which we made our experiments. The dataset represents a family tree. The file contains a predicateexamples(P, N, G, R). The predicate's arguments represent rispectively all positive examples, some negative ones, the information about the gender of every person, some facts about relationships (A married B, C is the parent of D, ...). -
/kb-construction: this folder contains the files that work con KB construction. The goal is start from a dataset.dat likefamily.datand build up adata.kbfile that store the information about the single dataset in a multi-head clause format, i.e.: [head_1, head_2, ..., head_n] :- body_1, body2, ..., body_m. This folder contains also a module that deal with dataset corrupution. If you are working only with positive examples you can corrupt the examples for generate new ambiguous facts and make more intresting the experiments. -
/knowledge-base: it's where the output of thekb-construcition's process is stored. -
/graph-construction: This folder contains files that deal with the graph construction process starting from a knowledge base. This phase is based on a set of ad-hoc constraint (stored in the/graph-construction/constraintsfolder), that explain what kinds of fact can coexist in the knowledge base. Based on these constraints, a set of labeled edge (attack edge or support edge) is generated, and the graph is built. Outputs of this phase (graphs) will be placed in the project folder/graph-construction/output/ -
/graph-fixing: this folder contains files and tools to extract extensions from graph using ASPARTIX tool. A set of file.tun will be produced. -
/output: in this folder will be placed the final output of this study (a set of file.tun). You can use INTHELEX system to learn a theory from the starting dataset.