override public SuggestionResult[] GetSuggestions(string query)
{
// to bigrams
var bigrams = GetBigrams(query);
var aggregated = new Dictionary <int, SuggestionResult>();
foreach (var gram in bigrams)
{
var gramLabel = gram.Key;
// Console.WriteLine("label: " + gramLabel);
HashSet <int> docs;
if (bigramToSuggestions.TryGetValue(gramLabel, out docs))
{
foreach (var docId in docs)
{
SuggestionResult tempSuggestionResult;
// add to aggregated result
if (!aggregated.TryGetValue(docId, out tempSuggestionResult))
{
aggregated[docId] = new SuggestionResult(documents[docId], 1);
}
else
{
tempSuggestionResult.score++;
}
continue;
}
}
}
return(aggregated.Values.OrderBy(x => - x.score).ToArray());
}
public SuggestionResult[] SuggestToken(string pollutedToken, int limit)
{
// naive method
var grams = GetBigrams(pollutedToken);
var result = new Dictionary <int, SuggestionResult>();
foreach (var gram in grams)
{
SuggestionResult suggest;
HashSet <int> tokenIndexes;
if (!bigrams.TryGetValue(gram.Key, out tokenIndexes))
{
continue;
}
foreach (var tokenId in tokenIndexes)
{
// add candidate words to dict
if (!result.TryGetValue(tokenId, out suggest))
{
result[tokenId] = new SuggestionResult(invertedDict[tokenId], 1);
}
else
{
suggest.score++;
}
}
}
return(result.Values.OrderBy(x => - x.score).Take(limit).ToArray());
}
override public SuggestionResult[] GetSuggestions(string query)
{
query = query.Trim();
// tokenizer
var tokens = query.Split(' ').ToList();
var result = new Dictionary <int, SuggestionResult>();
for (int i = 0; i < tokens.Count; ++i)
{
int tokenIndex;
if (dict.TryGetValue(tokens[i], out tokenIndex))
{
foreach (var candidate in invertedIndex[tokenIndex])
{
SuggestionResult suggestion;
if (!result.TryGetValue(candidate, out suggestion))
{
result[candidate] = new SuggestionResult(idToSuggestion[candidate].content, 1);
}
else
{
suggestion.score++;
}
// if (result.Count >= 10) {
// Console.WriteLine("Break due to limit exceeding");
// break;
// }
}
continue;
}
// else go for a suggestion for 3 small word:
var suggestions = SuggestToken(tokens[i], 3);
for (int j = 0; j < suggestions.Length; ++j)
{
tokens.Add(suggestions[j].value);
}
}
return(result.Values.OrderBy(x => - x.score).Take(10).ToArray());
}
public SuggestionResult[] GetFastSuggestions(string query, int tolerance = 100)
{
// tokenizer
var tokens = query.Split(' ').ToList();
var result = new Dictionary <int, SuggestionResult>();
for (int i = 0; i < tokens.Count; ++i)
{
int tokenIndex;
if (dict.TryGetValue(tokens[i], out tokenIndex))
{
foreach (var candidate in invertedIndex[tokenIndex])
{
SuggestionResult suggestion;
if (!result.TryGetValue(candidate, out suggestion))
{
result[candidate] = new SuggestionResult(idToSuggestion[candidate].content, 1);
}
else
{
suggestion.score++;
}
}
continue;
}
var watch = new Stopwatch();
watch.Start();
// else go for a suggestion for matches small words:
var suggestions = SuggestToken(tokens[i], 2);
watch.Stop();
Console.WriteLine("Suggestion ellapsed time: " + watch.Elapsed.TotalMilliseconds.ToString("0.0") + "ms");
for (int j = 0; j < suggestions.Length; ++j)
{
tokens.Add(suggestions[j].value);
}
}
return(result.Values.OrderBy(x => - x.score).Take(10).ToArray());
}
