24. Filtrado por Relevancia Semántica en Búsquedas

public class SemanticRelevanceFilter
{
    private const double MinVectorScore = 0.35;
    private const double MinLexicalOverlap = 0.2;

    public async Task<List<SearchResult>> FilterRelevantResultsAsync(
        string query,
        List<SearchResult> candidates,
        SearchContext context)
    {
        var queryTokens = TokenizeQuery(query);
        var filtered = new List<SearchResult>();

        foreach (var candidate in candidates)
        {
            var relevanceScore = CalculateRelevance(
                query,
                queryTokens,
                candidate,
                context);

            if (relevanceScore.IsRelevant)
            {
                candidate.RelevanceScore = relevanceScore.Score;
                candidate.RelevanceReason = relevanceScore.Reason;
                filtered.Add(candidate);
            }
        }

        return filtered.OrderByDescending(r => r.RelevanceScore).ToList();
    }

    private RelevanceScore CalculateRelevance(
        string query,
        HashSet<string> queryTokens,
        SearchResult candidate,
        SearchContext context)
    {
        // 1. Vector score check
        if (candidate.VectorScore < MinVectorScore)
        {
            return RelevanceScore.NotRelevant("Vector score too low");
        }

        // 2. Lexical overlap check
        var candidateTokens = TokenizeText(candidate.Description);
        var overlap = CalculateLexicalOverlap(queryTokens, candidateTokens);

        if (overlap < MinLexicalOverlap)
        {
            return RelevanceScore.NotRelevant("Insufficient lexical overlap");
        }

        // 3. Domain match check
        if (context.Domain != null)
        {
            if (!MatchesDomain(candidate, context.Domain))
            {
                return RelevanceScore.NotRelevant("Domain mismatch");
            }
        }

        // Calculate final score
        var finalScore = (candidate.VectorScore * 0.6) + 
                        (overlap * 0.3) + 
                        (DomainBonus(candidate, context) * 0.1);

        return RelevanceScore.Relevant(finalScore, "Passed all checks");
    }
}