Pass confidences in AWP; other minor fixes

sillsdev · Feb 24, 2025 · a020acf · a020acf
1 parent 3fdf92e
commit a020acf
Show file tree

Hide file tree

Showing 25 changed files with 43 additions and 76 deletions.
diff --git a/src/Echo/src/EchoEngine/WordAlignmentEngineServiceV1.cs b/src/Echo/src/EchoEngine/WordAlignmentEngineServiceV1.cs
@@ -51,7 +51,6 @@ ServerCallContext context
             {
                 SourceTokens = { sourceTokens },
                 TargetTokens = { targetTokens },
-                Confidences = { Enumerable.Repeat(1.0, minLength) },
                 Alignment = { GenerateAlignedWordPairs(minLength) }
             }
         };
@@ -93,8 +92,7 @@ await _parallelCorpusPreprocessingService.PreprocessAsync(
                                             .Select(
                                                 (_, i) => new AlignedWordPair() { SourceIndex = i, TargetIndex = i }
                                             )
-                                    },
-                                    Confidences = { row.SourceSegment.Split().Select(_ => 1.0) }
+                                    }
                                 }
                             );
                             return Task.CompletedTask;

diff --git a/src/Machine/src/Serval.Machine.Shared/Configuration/ThotWordAlignmentModelOptions.cs b/src/Machine/src/Serval.Machine.Shared/Configuration/ThotWordAlignmentModelOptions.cs
@@ -2,7 +2,7 @@
 
 public class ThotWordAlignmentModelOptions
 {
-    public const string Key = "WordAlignmentModel";
+    public const string Key = "ThotWordAlignmentModel";
 
     public ThotWordAlignmentModelOptions()
     {

diff --git a/src/Machine/src/Serval.Machine.Shared/Models/WordAlignment.cs b/src/Machine/src/Serval.Machine.Shared/Models/WordAlignment.cs
@@ -7,6 +7,5 @@ public record WordAlignment
     public required IReadOnlyList<string> Refs { get; init; }
     public required IReadOnlyList<string> SourceTokens { get; set; }
     public required IReadOnlyList<string> TargetTokens { get; set; }
-    public required IReadOnlyList<double> Confidences { get; set; }
     public required IReadOnlyList<AlignedWordPair> Alignment { get; set; }
 }
diff --git a/src/Machine/src/Serval.Machine.Shared/Serval.Machine.Shared.csproj b/src/Machine/src/Serval.Machine.Shared/Serval.Machine.Shared.csproj
@@ -42,6 +42,7 @@
 		<PackageReference Include="SIL.Machine.Translation.Thot" Version="3.5.2" Condition="!Exists('..\..\..\..\..\machine\src\SIL.Machine.Translation.Thot\SIL.Machine.Translation.Thot.csproj')" />
 		<PackageReference Include="SIL.WritingSystems" Version="14.1.1" />
 		<PackageReference Include="System.Linq.Async" Version="6.0.1" />
+		<PackageReference Include="YamlDotNet" Version="11.2.1" />
 	</ItemGroup>
 
 	<ItemGroup>
@@ -50,7 +51,6 @@
 		<ProjectReference Include="..\..\..\..\..\machine\src\SIL.Machine\SIL.Machine.csproj" Condition="Exists('..\..\..\..\..\machine\src\SIL.Machine\SIL.Machine.csproj')" />
 		<ProjectReference Include="..\..\..\..\..\machine\src\SIL.Machine.Morphology.HermitCrab\SIL.Machine.Morphology.HermitCrab.csproj" Condition="Exists('..\..\..\..\..\machine\src\SIL.Machine.Morphology.HermitCrab\SIL.Machine.Morphology.HermitCrab.csproj')" />
 		<ProjectReference Include="..\..\..\..\..\machine\src\SIL.Machine.Translation.Thot\SIL.Machine.Translation.Thot.csproj" Condition="Exists('..\..\..\..\..\machine\src\SIL.Machine.Translation.Thot\SIL.Machine.Translation.Thot.csproj')" />
-		<ProjectReference Include="..\..\..\..\..\machine\src\SIL.Machine.Tool\SIL.Machine.Tool.csproj" Condition="Exists('..\..\..\..\..\machine\src\SIL.Machine.Tool\SIL.Machine.Tool.csproj')" />
 		<ProjectReference Include="..\..\..\ServiceToolkit\src\SIL.ServiceToolkit\SIL.ServiceToolkit.csproj" />
 		<EmbeddedResource Include="data\flores200languages.csv" />
 	</ItemGroup>

diff --git a/src/Machine/src/Serval.Machine.Shared/Services/ClearMLMonitorService.cs b/src/Machine/src/Serval.Machine.Shared/Services/ClearMLMonitorService.cs
@@ -1,4 +1,6 @@
-namespace Serval.Machine.Shared.Services;
+using System.ComponentModel;
+
+namespace Serval.Machine.Shared.Services;
 
 public class ClearMLMonitorService(
     IServiceProvider services,
@@ -105,9 +107,20 @@ await _clearMLService.GetTasksByIdAsync(
             var dataAccessContext = scope.ServiceProvider.GetRequiredService<IDataAccessContext>();
             foreach (ITrainingEngine engine in engineToBuildServiceDict.Keys)
             {
-                IPlatformService platformService = scope.ServiceProvider.GetKeyedService<IPlatformService>(
-                    engine.Type.ToEngineGroup()
-                )!;
+                IPlatformService platformService = scope.ServiceProvider.GetRequiredKeyedService<IPlatformService>(
+                    engine.Type switch
+                    {
+                        EngineType.SmtTransfer => EngineGroup.Translation,
+                        EngineType.Nmt => EngineGroup.Translation,
+                        EngineType.Statistical => EngineGroup.WordAlignment,
+                        _
+                            => throw new InvalidEnumArgumentException(
+                                nameof(engine.Type),
+                                (int)engine.Type,
+                                typeof(EngineType)
+                            )
+                    }
+                );
                 if (engine.CurrentBuild is null || !tasks.TryGetValue(engine.CurrentBuild.JobId, out ClearMLTask? task))
                     continue;
 

diff --git a/src/Machine/src/Serval.Machine.Shared/Services/EngineType.cs b/src/Machine/src/Serval.Machine.Shared/Services/EngineType.cs
@@ -12,17 +12,3 @@ public enum EngineGroup
     Translation,
     WordAlignment
 }
-
-public static class EngineTypeExtensions
-{
-    public static EngineGroup ToEngineGroup(this EngineType engineType)
-    {
-        return engineType switch
-        {
-            EngineType.SmtTransfer => EngineGroup.Translation,
-            EngineType.Nmt => EngineGroup.Translation,
-            EngineType.Statistical => EngineGroup.WordAlignment,
-            _ => throw new ArgumentOutOfRangeException(nameof(engineType), engineType, null)
-        };
-    }
-}
diff --git a/...ine/src/Serval.Machine.Shared/Services/ServalWordAlignmentPlatformOutboxMessageHandler.cs b/...ine/src/Serval.Machine.Shared/Services/ServalWordAlignmentPlatformOutboxMessageHandler.cs
@@ -74,7 +74,6 @@ await call.RequestStream.WriteAsync(
                                 Refs = { wordAlignment.Refs },
                                 SourceTokens = { wordAlignment.SourceTokens },
                                 TargetTokens = { wordAlignment.TargetTokens },
-                                Confidences = { wordAlignment.Confidences },
                                 Alignment = { Map(wordAlignment.Alignment) }
                             },
                             cancellationToken
@@ -125,7 +124,6 @@ JsonSerializerOptions options
             }
             string corpusId = "",
                 textId = "";
-            IReadOnlyList<double> confidences = [];
             IReadOnlyList<string> refs = [],
                 sourceTokens = [],
                 targetTokens = [];
@@ -145,10 +143,6 @@ JsonSerializerOptions options
                             reader.Read();
                             textId = reader.GetString()!;
                             break;
-                        case "confidences":
-                            reader.Read();
-                            confidences = JsonSerializer.Deserialize<IList<double>>(ref reader, options)!.ToArray();
-                            break;
                         case "refs":
                             reader.Read();
                             refs = JsonSerializer.Deserialize<IList<string>>(ref reader, options)!.ToArray();
@@ -178,7 +172,6 @@ JsonSerializerOptions options
                 TextId = textId,
                 Refs = refs,
                 Alignment = alignedWordPairs,
-                Confidences = confidences,
                 SourceTokens = sourceTokens,
                 TargetTokens = targetTokens
             };

diff --git a/src/Machine/src/Serval.Machine.Shared/Services/StatisticalEngineService.cs b/src/Machine/src/Serval.Machine.Shared/Services/StatisticalEngineService.cs
@@ -80,8 +80,7 @@ public async Task<WordAlignmentResult> AlignAsync(
                 {
                     SourceTokens = { sourceTokens },
                     TargetTokens = { targetTokens },
-                    Alignment = { wordPairs.Select(Map) },
-                    Confidences = { wordPairs.Select(wp => wp.AlignmentScore).ToList() }
+                    Alignment = { wordPairs.Select(Map) }
                 };
             },
             cancellationToken: cancellationToken

diff --git a/src/Machine/src/Serval.Machine.Shared/Services/StatisticalTrainBuildJob.cs b/src/Machine/src/Serval.Machine.Shared/Services/StatisticalTrainBuildJob.cs
@@ -177,8 +177,7 @@ CancellationToken cancellationToken
                     targetWriter,
                     wordAlignment with
                     {
-                        Alignment = alignedWordPairs,
-                        Confidences = alignedWordPairs.Select(wp => wp.AlignmentScore).ToArray()
+                        Alignment = alignedWordPairs
                     },
                     JsonSerializerOptions
                 );

diff --git a/src/Machine/test/Serval.Machine.Shared.Tests/Services/StatisticalEngineServiceTests.cs b/src/Machine/test/Serval.Machine.Shared.Tests/Services/StatisticalEngineServiceTests.cs
@@ -132,7 +132,6 @@ public async Task AlignAsync()
         using var env = new TestEnvironment();
         WordAlignmentResult result = await env.Service.AlignAsync(EngineId1, "esto es una prueba.", "this is a test.");
         Assert.That(string.Join(' ', result.TargetTokens), Is.EqualTo("this is a test ."));
-        Assert.That(result.Confidences, Has.Count.EqualTo(5));
         Assert.That(result.Alignment.First().SourceIndex, Is.EqualTo(0));
         Assert.That(result.Alignment.First().TargetIndex, Is.EqualTo(0));
     }

diff --git a/src/Serval/src/Serval.Client/Client.g.cs b/src/Serval/src/Serval.Client/Client.g.cs
@@ -7137,8 +7137,7 @@ public partial interface IWordAlignmentEnginesClient
         /// <br/>  * An auto-generated reference of `[TextId]:[lineNumber]`, 1 indexed.
         /// <br/>* **SourceTokens**: the tokenized source segment
         /// <br/>* **TargetTokens**: the tokenized target segment
-        /// <br/>* **Confidences**: the confidence of the alignment on a scale from 0 to 1
-        /// <br/>* **Alignment**: a list of aligned word pairs
+        /// <br/>* **Alignment**: a list of aligned word pairs with associated scores
         /// <br/>            
         /// <br/>Word alignments can be filtered by text id if provided.
         /// <br/>Only word alignments for the most recent successful build of the engine are returned.
@@ -8406,8 +8405,7 @@ public string BaseUrl
         /// <br/>  * An auto-generated reference of `[TextId]:[lineNumber]`, 1 indexed.
         /// <br/>* **SourceTokens**: the tokenized source segment
         /// <br/>* **TargetTokens**: the tokenized target segment
-        /// <br/>* **Confidences**: the confidence of the alignment on a scale from 0 to 1
-        /// <br/>* **Alignment**: a list of aligned word pairs
+        /// <br/>* **Alignment**: a list of aligned word pairs with associated scores
         /// <br/>            
         /// <br/>Word alignments can be filtered by text id if provided.
         /// <br/>Only word alignments for the most recent successful build of the engine are returned.
@@ -10547,10 +10545,6 @@ public partial class WordAlignmentResult
         [System.ComponentModel.DataAnnotations.Required]
         public System.Collections.Generic.IList<string> TargetTokens { get; set; } = new System.Collections.ObjectModel.Collection<string>();
 
-        [Newtonsoft.Json.JsonProperty("confidences", Required = Newtonsoft.Json.Required.Always)]
-        [System.ComponentModel.DataAnnotations.Required]
-        public System.Collections.Generic.IList<double> Confidences { get; set; } = new System.Collections.ObjectModel.Collection<double>();
-
         [Newtonsoft.Json.JsonProperty("alignment", Required = Newtonsoft.Json.Required.Always)]
         [System.ComponentModel.DataAnnotations.Required]
         public System.Collections.Generic.IList<AlignedWordPair> Alignment { get; set; } = new System.Collections.ObjectModel.Collection<AlignedWordPair>();
@@ -10644,10 +10638,6 @@ public partial class WordAlignment
         [System.ComponentModel.DataAnnotations.Required]
         public System.Collections.Generic.IList<string> TargetTokens { get; set; } = new System.Collections.ObjectModel.Collection<string>();
 
-        [Newtonsoft.Json.JsonProperty("confidences", Required = Newtonsoft.Json.Required.Always)]
-        [System.ComponentModel.DataAnnotations.Required]
-        public System.Collections.Generic.IList<double> Confidences { get; set; } = new System.Collections.ObjectModel.Collection<double>();
-
         [Newtonsoft.Json.JsonProperty("alignment", Required = Newtonsoft.Json.Required.Always)]
         [System.ComponentModel.DataAnnotations.Required]
         public System.Collections.Generic.IList<AlignedWordPair> Alignment { get; set; } = new System.Collections.ObjectModel.Collection<AlignedWordPair>();

diff --git a/src/Serval/src/Serval.Grpc/Protos/serval/word_alignment/v1/engine.proto b/src/Serval/src/Serval.Grpc/Protos/serval/word_alignment/v1/engine.proto
@@ -61,12 +61,12 @@ message GetQueueSizeResponse {
 message AlignedWordPair {
     int32 source_index = 1;
     int32 target_index = 2;
+    double score = 3;
 }
 
 message WordAlignmentResult {
     repeated string source_tokens = 1;
     repeated string target_tokens = 2;
-    repeated double confidences = 3;
     repeated AlignedWordPair alignment = 4;
 }
 

diff --git a/src/Serval/src/Serval.Grpc/Protos/serval/word_alignment/v1/platform.proto b/src/Serval/src/Serval.Grpc/Protos/serval/word_alignment/v1/platform.proto
@@ -63,8 +63,7 @@ message InsertWordAlignmentsRequest {
     repeated string refs = 4;
     repeated string source_tokens = 5;
     repeated string target_tokens = 6;
-    repeated double confidences = 7;
-    repeated AlignedWordPair alignment = 8;
+    repeated AlignedWordPair alignment = 7;
 }
 
 message UpdateBuildExecutionDataRequest {

diff --git a/src/Serval/src/Serval.Shared/Models/AlignedWordPair.cs b/src/Serval/src/Serval.Shared/Models/AlignedWordPair.cs
@@ -4,4 +4,5 @@ public record AlignedWordPair
 {
     public required int SourceIndex { get; set; }
     public required int TargetIndex { get; set; }
+    public double? Score { get; set; }
 }
diff --git a/src/Serval/src/Serval.Translation/Services/PretranslationService.cs b/src/Serval/src/Serval.Translation/Services/PretranslationService.cs
@@ -100,7 +100,7 @@ await GetAllAsync(engineId, modelRevision, corpusId, textId, cancellationToken)
                             textId,
                             pretranslations.ToList(),
                             fullName: targetSettings.FullName,
-                            behavior: UpdateUsfmBehavior.PreferExisting
+                            textBehavior: UpdateUsfmTextBehavior.PreferExisting
                         ) ?? "";
                     break;
                 case PretranslationUsfmTextOrigin.PreferPretranslated:
@@ -109,7 +109,7 @@ await GetAllAsync(engineId, modelRevision, corpusId, textId, cancellationToken)
                             textId,
                             pretranslations.ToList(),
                             fullName: targetSettings.FullName,
-                            behavior: UpdateUsfmBehavior.PreferNew
+                            textBehavior: UpdateUsfmTextBehavior.PreferNew
                         ) ?? "";
                     break;
                 case PretranslationUsfmTextOrigin.OnlyExisting:
@@ -118,7 +118,7 @@ await GetAllAsync(engineId, modelRevision, corpusId, textId, cancellationToken)
                             textId,
                             [], // don't put any pretranslations, we only want the existing text.
                             fullName: targetSettings.FullName,
-                            behavior: UpdateUsfmBehavior.PreferNew
+                            textBehavior: UpdateUsfmTextBehavior.PreferNew
                         ) ?? "";
                     break;
                 case PretranslationUsfmTextOrigin.OnlyPretranslated:
@@ -127,7 +127,7 @@ await GetAllAsync(engineId, modelRevision, corpusId, textId, cancellationToken)
                             textId,
                             pretranslations.ToList(),
                             fullName: targetSettings.FullName,
-                            behavior: UpdateUsfmBehavior.StripExisting
+                            textBehavior: UpdateUsfmTextBehavior.StripExisting
                         ) ?? "";
                     break;
             }
@@ -151,14 +151,14 @@ await GetAllAsync(engineId, modelRevision, corpusId, textId, cancellationToken)
                             textId,
                             pretranslations.ToList(),
                             fullName: targetSettings.FullName,
-                            behavior: UpdateUsfmBehavior.StripExisting
+                            textBehavior: UpdateUsfmTextBehavior.StripExisting
                         ) ?? "";
                 case PretranslationUsfmTextOrigin.OnlyExisting:
                     return updater.UpdateUsfm(
                             textId,
                             [], // don't pass the pretranslations, we only want the existing text.
                             fullName: targetSettings.FullName,
-                            behavior: UpdateUsfmBehavior.StripExisting
+                            textBehavior: UpdateUsfmTextBehavior.StripExisting
                         ) ?? "";
             }
         }

diff --git a/src/Serval/src/Serval.WordAlignment/Contracts/WordAlignmentDto.cs b/src/Serval/src/Serval.WordAlignment/Contracts/WordAlignmentDto.cs
@@ -6,6 +6,5 @@ public record WordAlignmentDto
     public required IReadOnlyList<string> Refs { get; init; }
     public required IReadOnlyList<string> SourceTokens { get; init; }
     public required IReadOnlyList<string> TargetTokens { get; init; }
-    public required IReadOnlyList<double> Confidences { get; init; }
     public required IReadOnlyList<AlignedWordPairDto> Alignment { get; init; }
 }
diff --git a/src/Serval/src/Serval.WordAlignment/Contracts/WordAlignmentResultDto.cs b/src/Serval/src/Serval.WordAlignment/Contracts/WordAlignmentResultDto.cs
@@ -4,6 +4,5 @@ public record WordAlignmentResultDto
 {
     public required IReadOnlyList<string> SourceTokens { get; init; }
     public required IReadOnlyList<string> TargetTokens { get; init; }
-    public required IReadOnlyList<double> Confidences { get; init; }
     public required IReadOnlyList<AlignedWordPairDto> Alignment { get; init; }
 }
diff --git a/src/Serval/src/Serval.WordAlignment/Controllers/WordAlignmentEnginesController.cs b/src/Serval/src/Serval.WordAlignment/Controllers/WordAlignmentEnginesController.cs
@@ -387,8 +387,7 @@ CancellationToken cancellationToken
     ///   * An auto-generated reference of `[TextId]:[lineNumber]`, 1 indexed.
     /// * **SourceTokens**: the tokenized source segment
     /// * **TargetTokens**: the tokenized target segment
-    /// * **Confidences**: the confidence of the alignment on a scale from 0 to 1
-    /// * **Alignment**: a list of aligned word pairs
+    /// * **Alignment**: a list of aligned word pairs with associated scores
     ///
     /// Word alignments can be filtered by text id if provided.
     /// Only word alignments for the most recent successful build of the engine are returned.
@@ -984,7 +983,6 @@ private WordAlignmentResultDto Map(WordAlignmentResult source)
         {
             SourceTokens = source.SourceTokens.ToList(),
             TargetTokens = source.TargetTokens.ToList(),
-            Confidences = source.Confidences.Select(c => Math.Round(c, 8)).ToList(),
             Alignment = source.Alignment.Select(Map).ToList(),
         };
     }
@@ -1002,7 +1000,6 @@ private static WordAlignmentDto Map(Models.WordAlignment source)
             Refs = source.Refs,
             SourceTokens = source.SourceTokens.ToList(),
             TargetTokens = source.TargetTokens.ToList(),
-            Confidences = source.Confidences.Select(c => Math.Round(c, 8)).ToList(),
             Alignment = source
                 .Alignment.Select(c => new AlignedWordPairDto()
                 {

diff --git a/src/Serval/src/Serval.WordAlignment/Models/WordAlignment.cs b/src/Serval/src/Serval.WordAlignment/Models/WordAlignment.cs
@@ -11,6 +11,5 @@ public class WordAlignment : IEntity
     public required IReadOnlyList<string> Refs { get; init; }
     public required IReadOnlyList<string> SourceTokens { get; set; }
     public required IReadOnlyList<string> TargetTokens { get; set; }
-    public required IReadOnlyList<double> Confidences { get; set; }
     public required IReadOnlyList<AlignedWordPair> Alignment { get; set; }
 }
diff --git a/src/Serval/src/Serval.WordAlignment/Models/WordAlignmentResult.cs b/src/Serval/src/Serval.WordAlignment/Models/WordAlignmentResult.cs
@@ -4,6 +4,5 @@ public record WordAlignmentResult
 {
     public required IReadOnlyList<string> SourceTokens { get; set; }
     public required IReadOnlyList<string> TargetTokens { get; set; }
-    public required IReadOnlyList<double> Confidences { get; set; }
     public required IReadOnlyList<AlignedWordPair> Alignment { get; set; }
 }
diff --git a/src/Serval/src/Serval.WordAlignment/Services/EngineService.cs b/src/Serval/src/Serval.WordAlignment/Services/EngineService.cs
@@ -437,14 +437,18 @@ private Models.WordAlignmentResult Map(V1.WordAlignmentResult source)
         {
             SourceTokens = source.SourceTokens.ToList(),
             TargetTokens = source.TargetTokens.ToList(),
-            Confidences = source.Confidences.ToList(),
             Alignment = source.Alignment.Select(Map).ToList(),
         };
     }
 
     private Shared.Models.AlignedWordPair Map(V1.AlignedWordPair source)
     {
-        return new Shared.Models.AlignedWordPair { SourceIndex = source.SourceIndex, TargetIndex = source.TargetIndex };
+        return new Shared.Models.AlignedWordPair
+        {
+            SourceIndex = source.SourceIndex,
+            TargetIndex = source.TargetIndex,
+            Score = source.Score
+        };
     }
 
     private V1.ParallelCorpus Map(

diff --git a/src/Serval/src/Serval.WordAlignment/Services/WordAlignmentPlatformServiceV1.cs b/src/Serval/src/Serval.WordAlignment/Services/WordAlignmentPlatformServiceV1.cs
@@ -307,12 +307,12 @@ ServerCallContext context
                     Refs = request.Refs.ToList(),
                     SourceTokens = request.SourceTokens.ToList(),
                     TargetTokens = request.TargetTokens.ToList(),
-                    Confidences = request.Confidences.ToList(),
                     Alignment = request
                         .Alignment.Select(a => new Shared.Models.AlignedWordPair
                         {
                             SourceIndex = a.SourceIndex,
-                            TargetIndex = a.TargetIndex
+                            TargetIndex = a.TargetIndex,
+                            Score = a.Score
                         })
                         .ToList()
                 }