带权重的词典设计

一、词典类

public class WeightedDictionary
{
    public List<WeightedWord> Words { get; set; } = new List<WeightedWord>();
    private const int MAX_WORDS = 5000;

    public void AddWord(WeightedWord word)
    {
        var existing = Words.FirstOrDefault(w => w.Text == word.Text);
        if (existing != null)
        {
            existing.Weight += word.Weight;
        }
        else
        {
            Words.Add(word);
        }

        // 维护最大数量限制
        if (Words.Count > MAX_WORDS)
        {
            Words = Words
                .OrderByDescending(w => w.Weight)
                .Take(MAX_WORDS)
                .ToList();
        }
    }
}

public class WeightedWord
{
    [JsonProperty("text")]
    public string Text { get; set; }

    [JsonProperty("pinyin")]
    public string Pinyin { get; set; }

    [JsonProperty("weight")]
    public float Weight { get; set; } = 1.0f;

    [JsonProperty("replacements")]
    public List<string> CommonErrors { get; set; } = new List<string>();
}

二、智能权重管理

每用到一个词，*1.2来增加权重，其他词*0.98来衰减，新增的词自带1权重。

public class WeightManager
{
    // 动态权重调整
    public void UpdateWeights(string recognizedText)
    {
        foreach (var word in _dictionary.Words)
        {
            if (recognizedText.Contains(word.Text))
            {
                word.Weight *= 1.2f; // 增加当前词权重
            }
            else
            {
                word.Weight *= 0.98f; // 衰减未使用词的权重
            }
        }
    }

    // 自动清理低权重词
    public void PruneDictionary(float threshold = 0.1f)
    {
        _dictionary.Words.RemoveAll(w => w.Weight < threshold);
    }
}

三、将词典保存到本地

private void SaveDictionary() 
{ 
var timestamp = DateTime.Now.ToString("yyyyMMddHHmmss"); 
File.Copy("active_dict.json", $"archive/dict_{timestamp}.json"); 
var json = JsonConvert.SerializeObject(_dictionary);
 File.WriteAllText("active_dict.json", json); 
}