public void GetNTriplesRecursively()
{
NTripleCollection nTriples = reader.GetTriplesRecursvely(MockFolderPath + "skos_categories_en.nt", "http://dbpedia.org/resource/Category:Wars");
Assert.AreEqual(4, nTriples.Triples.Count);
Assert.IsTrue(nTriples.Triples.Any(x => x.Triple.Item1 == "http://dbpedia.org/resource/Category:Vietnam_War"));
Assert.IsTrue(nTriples.Triples.Any(x => x.Triple.Item1 == "http://dbpedia.org/resource/Category:World_War_II"));
}
public void ReadNTripleFile()
{
NTripleCollection nTriples = reader.ReadFile(MockFolderPath + "skos_categories_en.nt");
Assert.AreEqual(16, nTriples.Triples.Count);
Assert.AreEqual("http://dbpedia.org/resource/Category:Futurama", nTriples.Triples[4].Triple.Item1);
Assert.AreEqual("http://www.w3.org/2004/02/skos/core#broader", nTriples.Triples[4].Triple.Item2);
Assert.AreEqual("http://dbpedia.org/resource/Category:Works_by_Matt_Groening", nTriples.Triples[4].Triple.Item3);
}
public void AddWikipediaLinks()
{
List <Individual> individuals = MockIndividuals();
NTripleCollection wikipediLinks = reader.ReadFile(MockFolderPath + "wikipedia_links_en.nt");
individuals = generator.AddWikipediaLinks(individuals, wikipediLinks);
Assert.AreEqual("http://en.wikipedia.org/wiki/Anchor", individuals.Single(x => x.Name == "Anchor").WikipediaLink);
Assert.AreEqual("http://en.wikipedia.org/wiki/Keel", individuals.Single(x => x.Name == "Keel").WikipediaLink);
}
public void GenerateIndividualFromNTriple()
{
NTripleCollection nTriples = reader.ReadFile(MockFolderPath + "article_categories_en.nt");
List <Individual> individuals = new List <Individual>();
nTriples.Triples.ForEach(x => individuals.Add(new Individual(x)));
Assert.AreEqual(4, individuals.Count);
Assert.IsTrue(individuals.Single(x => x.Name == "Operation_Brushwood").Category == "World_War_II");
}
public void GenerateEntityFromNTriple()
{
NTripleCollection nTriples = reader.ReadFile(MockFolderPath + "skos_categories_en.nt");
List <Entity> entities = new List <Entity>();
nTriples.Triples.ForEach(x => entities.Add(new Entity(x)));
Assert.AreEqual(16, entities.Count);
Assert.IsTrue(entities.Single(x => x.Name == "Wars_involving_the_United_States").Parent == "Wars_by_country");
}
public void AddShortAbstracts()
{
List <Individual> individuals = MockIndividuals();
NTripleCollection shortAbstracts = reader.ReadFile(MockFolderPath + "short_abstracts_en.nt");
individuals = generator.AddShortAbstracts(individuals, shortAbstracts);
Assert.IsTrue(individuals.Single(x => x.Name == "Anchor").ShortAbstract.StartsWith("An anchor is a device, normally made of metal, that is used"));
Assert.IsTrue(individuals.Single(x => x.Name == "Keel").ShortAbstract.StartsWith("In boats and ships, keel can refer to either"));
}
public List <Individual> AddShortAbstracts(List <Individual> individuals, NTripleCollection shortAbstracts)
{
foreach (Individual i in individuals)
{
NTriple triple = shortAbstracts.Triples.Where(x => x.Triple.Item1 == "http://dbpedia.org/resource/" + i.Name && x.Triple.Item2 == "http://www.w3.org/2000/01/rdf-schema#comment").FirstOrDefault();
if (triple != null)
{
i.ShortAbstract = triple.Triple.Item3;
}
}
return(individuals);
}
public List <Individual> AddWikipediaLinks(List <Individual> individuals, NTripleCollection wikipediaLinks)
{
foreach (Individual i in individuals)
{
NTriple triple = wikipediaLinks.Triples.Where(x => x.Triple.Item3 == "http://dbpedia.org/resource/" + i.Name && x.Triple.Item2 == "http://xmlns.com/foaf/0.1/primaryTopic").FirstOrDefault();
if (triple != null)
{
i.WikipediaLink = triple.Triple.Item1;
}
}
return(individuals);
}
static void Main(string[] args)
{
try
{
string path = args[0] + @"\";
Console.WriteLine("Starting...");
const string RootCategory = "http://dbpedia.org/resource/Category:Nautical_terms";
NTripleReader reader = new NTripleReader();
Console.WriteLine("Reading N-triples...");
NTripleCollection skolCategories = reader.GetTriplesRecursvely(path + "skos_categories_en.nt", RootCategory);
List <string> categories = skolCategories.Subjects;
categories.Add(RootCategory);
NTripleCollection articleCategories = reader.GetTriples(path + "article_categories_en.nt", categories, NTripleReader.NtripleSearchType.Object);
NTripleCollection wikipediLinks = reader.GetTriples(path + "wikipedia_links_en.nt", articleCategories.Subjects, NTripleReader.NtripleSearchType.Object);
NTripleCollection shortAbstracts = reader.GetTriples(path + "short_abstracts_en.nt", articleCategories.Subjects, NTripleReader.NtripleSearchType.Subject);
Console.WriteLine("Reading N-triples... - Done");
List <Entity> entities = skolCategories.Triples.Select(x => new Entity(x)).ToList();
List <Individual> individuals = articleCategories.Triples.Select(x => new Individual(x)).ToList();
OntologyGenerator ontologyGenerator = new OntologyGenerator();
List <Entity> individualEntities = ontologyGenerator.GenerateEntityForEachIndividual(individuals);
entities.AddRange(individualEntities);
Console.WriteLine("Adding Wikipedia links and short abstracts...");
individuals = ontologyGenerator.AddWikipediaLinks(individuals, wikipediLinks);
individuals = ontologyGenerator.AddShortAbstracts(individuals, shortAbstracts);
Console.WriteLine("Adding Wikipedia links and short abstracts... - Done");
Console.WriteLine("Generating ontology...");
XmlDocument owl = new OwlGenerator().Generate(entities, individuals);
Console.WriteLine("Generating ontology... - Done");
string fileName = "generatedOntology_" + DateTime.Now.ToString("yyyyMMdd_HHmmss") + ".owl";
owl.Save(fileName);
Console.WriteLine("Ontology file [" + fileName + "]");
Console.WriteLine("Done");
}
catch (Exception e)
{
Console.WriteLine("Error: " + e.Message);
Console.Write(e.StackTrace);
}
Console.ReadKey();
}
public NTripleCollection GetTriplesRecursvely(string path, List <string> filters)
{
NTripleCollection nTripleCollection = GetTriples(path, filters, NtripleSearchType.Object);
bool newTriplesFound = true;
while (newTriplesFound)
{
List <string> subjects = nTripleCollection.Subjects;
int before = nTripleCollection.Triples.Count;
nTripleCollection.AddTriples(GetTriples(path, subjects, NtripleSearchType.Object).Triples);
int after = nTripleCollection.Triples.Count;
newTriplesFound = after > before;
}
return(nTripleCollection);
}
public NTripleCollection GetTriples(string path, List <string> nTriples, NtripleSearchType searchType)
{
int searchTypeIndex = GetSearchTripeIndex(searchType);
Console.WriteLine("Reading triples from " + path);
NTripleCollection nTripleCollection = new NTripleCollection();
long currentLine = 0;
StreamReader reader = new StreamReader(path);
string[] nTriplesFormatted = nTriples.Select(x => "<" + x + ">").ToArray();
while (!reader.EndOfStream)
{
currentLine++;
Console.Write("\rLine {0}", currentLine);
string line = reader.ReadLine();
if (nTriplesFormatted.Length == 0)
{
string[] values = line.Split('>');
nTripleCollection.Triples.Add(new NTriple(RemoveUnwantedChars(values[0]), RemoveUnwantedChars(values[1]), RemoveUnwantedChars(values[2])));
}
else
{
for (int i = 0; i < nTriplesFormatted.Length; i++)
{
if (line.Contains(nTriplesFormatted[i]))
{
string[] values = line.Split('>');
if (nTriples.Count == 0 || nTriples.Any(x => values[searchTypeIndex].Contains(x)))
{
nTripleCollection.Triples.Add(new NTriple(RemoveUnwantedChars(values[0]), RemoveUnwantedChars(values[1]), RemoveUnwantedChars(values[2])));
break;
}
}
}
}
}
Console.WriteLine(Environment.NewLine + "Reading triples from " + path + " - Done");
reader.Close();
return(nTripleCollection);
}
