test(backend): add matchmaking relevance and ranking tests

tests/TarninaHub.Tests/Controllers/InnovationsControllerTests.cs

@@ -4,6 +4,7 @@ using Microsoft.AspNetCore.Mvc;
 using TarninaHub.Application.Abstractions;
 using TarninaHub.Application.Dtos;
 using TarninaHub.Controllers;
+using TarninaHub.Tests.Matchmaking;
 
 namespace TarninaHub.Tests.Controllers;
 
@@ -11,7 +12,7 @@ public class InnovationsControllerTests
 {
     private static InnovationsController CreateController(IInnovationService service)
     {
-        var controller = new InnovationsController(service);
+        var controller = new InnovationsController(service, new StubMatchmakingService());
         var identity = new ClaimsIdentity(
             [new Claim("sub", "user:tester"), new Claim("role", "admin")],
             authenticationType: "test");

tests/TarninaHub.Tests/Controllers/SubmissionsControllerTests.cs

@@ -4,6 +4,7 @@ using Microsoft.AspNetCore.Mvc;
 using TarninaHub.Application.Abstractions;
 using TarninaHub.Application.Dtos;
 using TarninaHub.Controllers;
+using TarninaHub.Tests.Matchmaking;
 
 namespace TarninaHub.Tests.Controllers;
 
@@ -11,7 +12,7 @@ public class SubmissionsControllerTests
 {
     private static SubmissionsController CreateController(ISubmissionService service, string role = "operator")
     {
-        var controller = new SubmissionsController(service);
+        var controller = new SubmissionsController(service, new StubMatchmakingService());
         var identity = new ClaimsIdentity(
             [new Claim("sub", "user:tester"), new Claim("role", role)],
             authenticationType: "test");

tests/TarninaHub.Tests/Matchmaking/EmbeddingAndRankingTests.cs

@@ -0,0 +1,70 @@
+using TarninaHub.Application.Matchmaking;
+
+namespace TarninaHub.Tests.Matchmaking;
+
+public class EmbeddingAndRankingTests
+{
+    private static double Magnitude(float[] vector) =>
+        Math.Sqrt(vector.Sum(value => (double)value * value));
+
+    [Fact]
+    public void HashingEmbeddingProvider_ProducesNormalizedVectorOfConfiguredDimension()
+    {
+        var provider = new HashingEmbeddingProvider();
+
+        var vector = provider.Embed("klub sąsiedzki dla seniorów");
+
+        Assert.Equal(EmbeddingConstants.Dimension, vector.Length);
+        Assert.Equal(1.0, Magnitude(vector), precision: 5);
+    }
+
+    [Fact]
+    public void HashingEmbeddingProvider_IsDeterministic()
+    {
+        var provider = new HashingEmbeddingProvider();
+
+        var first = provider.Embed("praca wspierana niepełnosprawność");
+        var second = provider.Embed("praca wspierana niepełnosprawność");
+
+        Assert.Equal(first, second);
+    }
+
+    [Fact]
+    public void CosineSimilarity_IsHigherForSharedVocabulary()
+    {
+        var provider = new HashingEmbeddingProvider();
+
+        var query = provider.Embed("seniorzy spotkania samotność");
+        var related = provider.Embed("klub sąsiedzki seniorzy spotkania samotność");
+        var unrelated = provider.Embed("hostel młodzież bezdomność");
+
+        var relatedScore = VectorMath.CosineSimilarity(query, related);
+        var unrelatedScore = VectorMath.CosineSimilarity(query, unrelated);
+
+        Assert.True(relatedScore > unrelatedScore);
+    }
+
+    [Fact]
+    public void SemanticRanker_OrdersByScoreAndRespectsTopK()
+    {
+        var provider = new HashingEmbeddingProvider();
+        var query = provider.Embed("cyfryzacja seniorzy internet");
+        var candidates = new (string Id, IReadOnlyList<float> Vector)[]
+        {
+            ("cyfryzacja", provider.Embed("cyfryzacja seniorzy internet szkolenie")),
+            ("praca", provider.Embed("praca wspierana doradca")),
+            ("hostel", provider.Embed("hostel młodzież bezdomność")),
+        };
+
+        var ranked = SemanticRanker.Rank(query, candidates, topK: 2);
+
+        Assert.Equal(2, ranked.Count);
+        Assert.Equal("cyfryzacja", ranked[0].Item);
+    }
+
+    [Fact]
+    public void CosineSimilarity_ReturnsZeroForMismatchedDimensions()
+    {
+        Assert.Equal(0, VectorMath.CosineSimilarity([1f, 0f], [1f]));
+    }
+}

tests/TarninaHub.Tests/Matchmaking/MatchmakingQualityTests.cs

@@ -0,0 +1,73 @@
+using TarninaHub.Application.Matchmaking;
+
+namespace TarninaHub.Tests.Matchmaking;
+
+/// <summary>
+/// Relevance test set for semantic matchmaking. Measures top-1 accuracy,
+/// precision@3 and recall@3 over a small synthetic corpus (matches the
+/// demo dataset vocabulary).
+/// </summary>
+public class MatchmakingQualityTests
+{
+    private static readonly (string Id, string Text)[] Documents =
+    [
+        ("klub", "klub sąsiedzki seniorzy spotkania samotność izolacja"),
+        ("praca", "praca wspierana niepełnosprawność zatrudnienie doradca"),
+        ("hostel", "hostel aktywizacyjny młodzież bezdomność usamodzielnianie"),
+        ("psych", "wsparcie psychiczne młodzież szkoła profilaktyka"),
+        ("ubostwo", "ubóstwo energetyczne żywność spizarnia seniorzy"),
+        ("integracja", "asysta kulturowa cudzoziemcy integracja język"),
+        ("rodzina", "rodzina zastępcza rodzeństwo piecza wsparcie"),
+        ("cyfryzacja", "cyfryzacja seniorzy internet szkolenie obsługa"),
+    ];
+
+    private static readonly (string Query, string Expected)[] Queries =
+    [
+        ("seniorzy spotkania samotność", "klub"),
+        ("praca niepełnosprawność doradca", "praca"),
+        ("hostel młodzież bezdomność", "hostel"),
+        ("psychiczne młodzież szkoła", "psych"),
+        ("ubóstwo energetyczne spizarnia", "ubostwo"),
+        ("asysta kulturowa integracja", "integracja"),
+        ("rodzina zastępcza rodzeństwo", "rodzina"),
+        ("cyfryzacja seniorzy internet", "cyfryzacja"),
+    ];
+
+    [Fact]
+    public void Matchmaking_MeetsRelevanceThresholds()
+    {
+        var provider = new HashingEmbeddingProvider();
+        var vectors = Documents
+            .Select(document => (document.Id, Vector: (IReadOnlyList<float>)provider.Embed(document.Text)))
+            .ToList();
+
+        var top1Correct = 0;
+        var recallAt3 = 0;
+        var precisionSum = 0.0;
+
+        foreach (var (query, expected) in Queries)
+        {
+            var ranked = SemanticRanker.Rank(provider.Embed(query), vectors, topK: 3);
+
+            if (ranked.Count > 0 && ranked[0].Item == expected)
+            {
+                top1Correct++;
+            }
+
+            if (ranked.Any(item => item.Item == expected))
+            {
+                recallAt3++;
+            }
+
+            precisionSum += ranked.Count(item => item.Item == expected) / 3.0;
+        }
+
+        var top1Accuracy = (double)top1Correct / Queries.Length;
+        var recall = (double)recallAt3 / Queries.Length;
+        var meanPrecision = precisionSum / Queries.Length;
+
+        Assert.True(top1Accuracy >= 0.75, $"top-1 accuracy was {top1Accuracy:P0}");
+        Assert.True(recall >= 0.9, $"recall@3 was {recall:P0}");
+        Assert.True(meanPrecision >= 0.3, $"precision@3 was {meanPrecision:P0}");
+    }
+}

tests/TarninaHub.Tests/Matchmaking/StubMatchmakingService.cs

@@ -0,0 +1,25 @@
+using TarninaHub.Application.Abstractions;
+using TarninaHub.Application.Dtos;
+
+namespace TarninaHub.Tests.Matchmaking;
+
+public sealed class StubMatchmakingService : IMatchmakingService
+{
+    public Task IndexInnovationAsync(string innovationId, CancellationToken cancellationToken) =>
+        Task.CompletedTask;
+
+    public Task IndexProblemAsync(string problemId, CancellationToken cancellationToken) =>
+        Task.CompletedTask;
+
+    public Task IndexSubmissionAsync(string submissionId, CancellationToken cancellationToken) =>
+        Task.CompletedTask;
+
+    public Task<int> ReindexAsync(CancellationToken cancellationToken) =>
+        Task.FromResult(0);
+
+    public Task<MatchResponse?> MatchAsync(
+        string submissionId,
+        MatchRequest request,
+        CancellationToken cancellationToken) =>
+        Task.FromResult<MatchResponse?>(new MatchResponse([], [], true));
+}