public IEnumerable<Tuple<int, InvertedIndexUnit>> GetMedicamentsFuzzyIndex()
{
var medNames = idToTradeName.Select(med => Tuple.Create(med.Key, med.Value.Name)).ToList();
var stemmer = new MyStemmer("Mkb10TradeNames.stemmed.txt", medNames.Select(it => it.Item2));
var stemmedTnNames =
medNames.Select(it => Tuple.Create(it.Item1, String.Join(" ", it.Item2.SplitIntoWords().Select(stemmer.Stem)))).ToList();
var tnWords = stemmedTnNames.Select(it => it.Item2).SelectMany(name => name.SplitIntoWords());
var answerTexts = medicalQuestions.SelectMany(q => q.GetAnswers()).Select(a => Tuple.Create(a.QuestionId, a.Text));
var wordsIndex = new FuzzyIndex(answerTexts, tnWords);
return (from it in medNames
let medId = it.Item1
let name = it.Item2
let medIds = GetUnitsIntersection(wordsIndex.GetIndex(), name.SplitIntoWords().ToArray()).ToList()
where medIds.Any()
select Tuple.Create(medId, new InvertedIndexUnit(idToTradeName[medId].Name, medIds))).ToList();
}
public static IEnumerable<InvertedIndexUnit> GetFuzzyIndex()
{
return DataActualityChecker.Check(
new Lazy<InvertedIndexUnit[]>(() =>
{
var questionList = Program.DefaultQuestionList;
var medicaments = new Medicaments(Program.MedicamentsFileName);
var idAnswerText = questionList.GetAllAnswers().Select(a => Tuple.Create(a.QuestionId, a.Text));
var fuzzyIndex = new FuzzyIndex(idAnswerText, medicaments.GetMedicamentNames());
return fuzzyIndex.GetIndex().OrderByDescending(k => k.Ids.Count).ToArray();
}),
InvertedIndexUnit.FormatStringWrite,
InvertedIndexUnit.FormatStringParse,
new FileDependencies(
Program.FilesDirectory + "MedicamentsFuzzyIndex.txt",
Program.MedicamentsFileName));
}
public IEnumerable<Tuple<int, InvertedIndexUnit>> GetDeseasesFuzzyIndex()
{
var deseaseNames = idToDesease.Select(des => { var names = des.Value.Synonyms; names.Add(des.Value.Name);
return Tuple.Create(des.Key, names);
}).ToList();
var stemmer = new MyStemmer("Mkb10Deseases.stemmed.txt", deseaseNames.SelectMany(it => it.Item2));
var stemmedDeseaseNames =
deseaseNames.Select(it => Tuple.Create(it.Item1, it.Item2.Select(name => String.Join(" ", name.SplitIntoWords().Select(stemmer.Stem))))).ToList();
var desWords = stemmedDeseaseNames.SelectMany(it => it.Item2).SelectMany(name => name.SplitIntoWords());
var questionTexts = medicalQuestions.Select(q => Tuple.Create(q.Id, q.WholeText));
var wordsIndex = new FuzzyIndex(questionTexts, desWords);
var deseaseIndex = new List<Tuple<int, InvertedIndexUnit>>();
foreach (var it in stemmedDeseaseNames)
{
var desId = it.Item1;
var names = it.Item2;
var deseaseIds = new List<long>();
foreach (var name in names)
{
deseaseIds.AddRange(GetUnitsIntersection(wordsIndex.GetIndex(), name.SplitIntoWords().ToArray()));
}
if(deseaseIds.Any())
deseaseIndex.Add(Tuple.Create(desId, new InvertedIndexUnit(idToDesease[desId].Name, deseaseIds)));
}
return deseaseIndex;
}