FrequencyAnalysisTask.cs - master - katherinesiv/CSharp_Practice5

FrequencyAnalysisTask.cs
84 строки3 KB
Сиваева Екатерина
create FrequencyAnalysisTask.cs
14 ноя 2025, 09:41
14 ноя 2025, 09:4197f5b73
namespace TextAnalysis;

static class FrequencyAnalysisTask
{
    public static Dictionary<string, string> GetMostFrequentNextWords(List<List<string>> text)
    {
        var countBigrams = new Dictionary<string, Dictionary<string, int>>();
        var countTrigrams = new Dictionary<string, Dictionary<string, int>>();
        foreach (var sentence in text)
        {
            if (sentence.Count < 2) continue;

            MakeBigrams(sentence, countBigrams);
            MakeTrigrams(sentence, countTrigrams);
        }
        var result = BuildFinalModel(countBigrams, countTrigrams);
        return result;
    }

    private static void MakeBigrams(List<string> sentence,
                                     Dictionary<string, Dictionary<string, int>> bigramCounts)
    {
        for (var i = 0; i < sentence.Count - 1; i++)
        {
            var firstWord = sentence[i];
            var secondWord = sentence[i + 1];

            if (!bigramCounts.ContainsKey(firstWord))
                bigramCounts[firstWord] = new Dictionary<string, int>();

            if (!bigramCounts[firstWord].ContainsKey(secondWord))
                bigramCounts[firstWord][secondWord] = 0;

            bigramCounts[firstWord][secondWord]++;
        }
    }

    private static void MakeTrigrams(List<string> sentence,
                                      Dictionary<string, Dictionary<string, int>> trigramCounts)
    {
        for (var i = 0; i < sentence.Count - 2; i++)
        {
            var firstTwoWords = $"{sentence[i]} {sentence[i + 1]}";
            var thirdWord = sentence[i + 2];

            if (!trigramCounts.ContainsKey(firstTwoWords))
                trigramCounts[firstTwoWords] = new Dictionary<string, int>();

            if (!trigramCounts[firstTwoWords].ContainsKey(thirdWord))
                trigramCounts[firstTwoWords][thirdWord] = 0;

            trigramCounts[firstTwoWords][thirdWord]++;
        }
    }

    private static Dictionary<string, string> BuildFinalModel(
        Dictionary<string, Dictionary<string, int>> bigramCounts,
        Dictionary<string, Dictionary<string, int>> trigramCounts)
    {
        var result = new Dictionary<string, string>();

        foreach (var bigram in bigramCounts)
            result[bigram.Key] = FindMostFrequentNextWord(bigram.Value);

        foreach (var trigram in trigramCounts)
            result[trigram.Key] = FindMostFrequentNextWord(trigram.Value);

        return result;
    }

    private static string FindMostFrequentNextWord(Dictionary<string, int> nextWordCounts)
    {
        var maxFrequency = nextWordCounts.Values.Max();
        var mostFrequentWords = nextWordCounts
            .Where(pair => pair.Value == maxFrequency)
            .Select(pair => pair.Key)
            .ToList();

        if (mostFrequentWords.Count == 1)
            return mostFrequentWords[0];
        else
            return mostFrequentWords.OrderBy(word => word, StringComparer.Ordinal).First();
    }
}