internal LinkedUtterance LinkUtterance(string utterance)
{
var parts = new List <LinkedUtterancePart>();
var words = utterance.Split(' ');
foreach (var word in words)
{
if (_indexedNodes.ContainsKey(word))
{
var entityInfo = new EntityInfo(word, word, 1, 1);
parts.Add(LinkedUtterancePart.Entity(word, new[] { entityInfo }));
}
else
{
parts.Add(LinkedUtterancePart.Word(word));
}
}
return(new LinkedUtterance(parts));
}
internal LinkedUtterance LinkedUtterance_Hungry()
{
var usedMatches = new List <EntityMatch>();
while (true)
{
EntityMatch bestMatch = null;
foreach (var match in _matches.SelectMany(m => m))
{
var hasCollision = usedMatches.Any(u => testCollision(match, u));
if (hasCollision)
{
continue;
}
// var bestScore = bestMatch == null ? 0 : bestMatch.Entity.InBounds + bestMatch.Entity.OutBounds;
// var currentScore = match.Entity.InBounds + match.Entity.OutBounds;
var bestScore = bestMatch == null ? 0 : bestMatch.Entity.Score;
var currentScore = match.Entity.Score;
if (bestScore < currentScore)
{
bestMatch = match;
}
}
if (bestMatch == null)
{
break;
}
usedMatches.Add(bestMatch);
}
var linkedParts = new List <LinkedUtterancePart>();
for (var i = 0; i < _words.Length; ++i)
{
var match = getIndexedMatch(i, usedMatches);
if (match == null)
{
linkedParts.Add(LinkedUtterancePart.Word(_words[i]));
}
else
{
var ngram = string.Join(" ", _words.Skip(i).Take(match.Length));
var entities = new List <EntityInfo>();
foreach (var ambigMatch in _matches[i])
{
if (ambigMatch.Length == match.Length)
{
entities.Add(ambigMatch.Entity);
}
}
linkedParts.Add(LinkedUtterancePart.Entity(ngram, entities.ToArray()));
i += match.Length - 1;
}
}
return(new LinkedUtterance(linkedParts));
}
public LinkedUtterance LinkUtterance(string utterance, IEnumerable <EntityInfo> context = null)
{
_context.Clear();
if (context != null)
{
foreach (var entity in context)
{
_context[Db.GetFreebaseId(entity.Mid)] = entity;
}
}
var linkedUtterance = base.LinkUtterance(utterance, 20).FirstOrDefault();
if (!_useDisambiguation || linkedUtterance == null)
{
return(linkedUtterance);
}
var entityClusters = new List <EntityInfo[]>();
var contextIndex = new HashSet <string>();
if (context != null)
{
contextIndex.UnionWith(context.Select(e => e.Mid));
}
foreach (var part in linkedUtterance.Parts)
{
if (part.Entities.Any())
{
//force context entities
var entities = part.Entities.ToArray();
EntityInfo contextEntity = null;
foreach (var entity in entities)
{
if (!contextIndex.Contains(entity.Mid))
{
continue;
}
contextEntity = entity;
break;
}
if (contextEntity == null)
{
entityClusters.Add(entities);
}
else
{
entityClusters.Add(new[] { contextEntity });
}
}
}
var disambiguatedClusters = new List <HashSet <EntityInfo> >();
for (var clusterIndex = 0; clusterIndex < entityClusters.Count; ++clusterIndex)
{
disambiguatedClusters.Add(new HashSet <EntityInfo>());
}
for (var i = 0; i < Nbest; ++i)
{
var disambiguated = disambiguateClusters(entityClusters);
for (var clusterIndex = 0; clusterIndex < entityClusters.Count; ++clusterIndex)
{
disambiguatedClusters[clusterIndex].Add(disambiguated[clusterIndex]);
if (entityClusters[clusterIndex].Length == 1)
{
//keep the last entity to prevent empty components
continue;
}
entityClusters[clusterIndex] = entityClusters[clusterIndex].Except(new[] { disambiguated[clusterIndex] }).ToArray();
}
}
var entityQueue = new Queue <HashSet <EntityInfo> >(disambiguatedClusters);
var disambiguatedParts = new List <LinkedUtterancePart>();
foreach (var part in linkedUtterance.Parts)
{
if (part.Entities.Any())
{
disambiguatedParts.Add(LinkedUtterancePart.Entity(part.Token, entityQueue.Dequeue().ToArray()));
}
else
{
disambiguatedParts.Add(part);
}
}
return(new LinkedUtterance(disambiguatedParts));
}
