More documentation

src/RankLib/Learning/IRanker.cs

@@ -14,8 +14,13 @@ public interface IRankerParameters
 }
 
 /// <summary>
-/// A ranker
+/// A ranker that can score <see cref="DataPoint"/> and rank data points in a <see cref="RankList"/>.
+/// A ranker can be trained by a <see cref="RankerTrainer"/>, or a trained ranker model can be loaded
+/// from file
 /// </summary>
+/// <remarks>
+/// Use <see cref="EvaluatorFactory"/> to create an <see cref="Evaluator"/> for training and evaluation.
+/// </remarks>
 public interface IRanker
 {
 	/// <summary>
@@ -120,15 +125,20 @@ public interface IRanker
 }
 
 /// <summary>
-/// A generic ranker
+/// A generic ranker that can score <see cref="DataPoint"/> and rank data points in a <see cref="RankList"/>.
+/// A ranker can be trained by a <see cref="RankerTrainer"/>, or a trained ranker model can be loaded
+/// from file
 /// </summary>
-/// <typeparam name="TParameters">The type of rank parameters</typeparam>
-public interface IRanker<TParameters> : IRanker
-	where TParameters : IRankerParameters
+/// <typeparam name="TRankerParameters">The type of ranker parameters</typeparam>
+/// <remarks>
+/// Use <see cref="EvaluatorFactory"/> to create an <see cref="Evaluator"/> for training and evaluation.
+/// </remarks>
+public interface IRanker<TRankerParameters> : IRanker
+	where TRankerParameters : IRankerParameters
 {
 	/// <summary>
 	/// Gets or sets the parameters for the ranker.
 	/// The ranker uses parameters for training
 	/// </summary>
-	new TParameters Parameters { get; set; }
+	new TRankerParameters Parameters { get; set; }
 }

src/RankLib/Learning/LinearRegression.cs

@@ -36,22 +36,25 @@ public class LinearRegression : Ranker<LinearRegressionParameters>
 	private readonly ILogger<LinearRegression> _logger;
 	private double[] _weight = [];
 
-	public LinearRegression(ILogger<LinearRegression>? logger = null) : base() =>
+	public LinearRegression(ILogger<LinearRegression>? logger = null) =>
 		_logger = logger ?? NullLogger<LinearRegression>.Instance;
 
 	public LinearRegression(List<RankList> samples, int[] features, MetricScorer scorer,
 		ILogger<LinearRegression>? logger = null)
 		: base(samples, features, scorer) =>
 		_logger = logger ?? NullLogger<LinearRegression>.Instance;
 
+	/// <inheritdoc />
 	public override string Name => RankerName;
 
+	/// <inheritdoc />
 	public override Task InitAsync()
 	{
 		_logger.LogInformation("Initializing...");
 		return Task.CompletedTask;
 	}
 
+	/// <inheritdoc />
 	public override Task LearnAsync()
 	{
 		_logger.LogInformation("Training starts...");
@@ -110,17 +113,18 @@ public override Task LearnAsync()
 
 		TrainingDataScore = SimpleMath.Round(Scorer.Score(Rank(Samples)), 4);
 		_logger.LogInformation("Finished successfully.");
-		_logger.LogInformation("{ScorerName} on training data: {ScoreOnTrainingData}", Scorer.Name, TrainingDataScore);
+		_logger.LogInformation("{Scorer} on training data: {TrainingScore}", Scorer.Name, TrainingDataScore);
 
 		if (ValidationSamples != null)
 		{
 			ValidationDataScore = Scorer.Score(Rank(ValidationSamples));
-			_logger.LogInformation("{ScorerName} on validation data: {BestScoreOnValidationData}", Scorer.Name, SimpleMath.Round(ValidationDataScore, 4));
+			_logger.LogInformation("{Scorer} on validation data: {ValidationScore}", Scorer.Name, SimpleMath.Round(ValidationDataScore, 4));
 		}
 
 		return Task.CompletedTask;
 	}
 
+	/// <inheritdoc />
 	public override double Eval(DataPoint dataPoint)
 	{
 		var score = _weight[^1];
@@ -130,6 +134,7 @@ public override double Eval(DataPoint dataPoint)
 		return score;
 	}
 
+	/// <inheritdoc />
 	public override string GetModel()
 	{
 		var output = new StringBuilder()
@@ -147,6 +152,7 @@ public override string GetModel()
 		return output.ToString();
 	}
 
+	/// <inheritdoc />
 	public override void LoadFromString(string model)
 	{
 		try

src/RankLib/Learning/Ranker.cs

@@ -1,20 +1,26 @@
 using System.Text;
+using RankLib.Eval;
 using RankLib.Metric;
 using RankLib.Utilities;
 
 namespace RankLib.Learning;
 
 /// <summary>
-/// Base class for a generic ranker with typed ranker parameters.
+/// Base class for a generic ranker that can score <see cref="DataPoint"/> and rank data points in a <see cref="RankList"/>.
+/// A ranker can be trained by a <see cref="RankerTrainer"/>, or a trained ranker model can be loaded
+/// from file
 /// </summary>
 /// <typeparam name="TRankerParameters">The type of ranker parameters</typeparam>
+/// <remarks>
+/// Use <see cref="EvaluatorFactory"/> to create an <see cref="Evaluator"/> for training and evaluation.
+/// </remarks>
 public abstract class Ranker<TRankerParameters> : IRanker<TRankerParameters>
 	where TRankerParameters : IRankerParameters, new()
 {
 	private MetricScorer? _scorer;
 
-	protected double TrainingDataScore = 0.0;
-	protected double ValidationDataScore = 0.0;
+	protected double TrainingDataScore = 0;
+	protected double ValidationDataScore = 0;
 
 	/// <inheritdoc />
 	public List<RankList> Samples { get; set; } = [];

src/RankLib/Learning/RankerTrainer.cs

@@ -10,9 +10,9 @@ public class RankerTrainer
 	private readonly RankerFactory _rankerFactory;
 
 	/// <summary>
-	/// Initializes a new instance of <see cref="RankerTrainer"/>
+	/// Instantiates a new instance of <see cref="RankerTrainer"/>
 	/// </summary>
-	/// <param name="rankerFactory"></param>
+	/// <param name="rankerFactory">The ranker factory used to create rankers</param>
 	public RankerTrainer(RankerFactory rankerFactory) => _rankerFactory = rankerFactory;
 
 	/// <summary>

src/RankLib/Learning/Sampler.cs

@@ -2,10 +2,6 @@
 
 namespace RankLib.Learning;
 
-using System;
-using System.Collections.Generic;
-using System.Linq;
-
 /// <summary>
 /// Samples rank lists from a pool of rank lists
 /// </summary>

src/RankLib/Parsing/ModelLineProducer.cs

@@ -20,9 +20,7 @@ private bool ReadUntil(string fullTextChar, int beginOfLineCursor, int endOfLine
 		if (fullTextChar[beginOfLineCursor] != '#')
 		{
 			for (var j = beginOfLineCursor; j <= endOfLineCursor; j++)
-			{
 				model.Append(fullTextChar[j]);
-			}
 		}
 
 		// Check for ensemble tag