Making MF to be a part of ML.NET

Microsoft.ML.sln

@@ -133,6 +133,8 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Microsoft.ML.Samples", "doc
 EndProject
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Microsoft.ML.SamplesUtils", "src\Microsoft.ML.SamplesUtils\Microsoft.ML.SamplesUtils.csproj", "{11A5210E-2EA7-42F1-80DB-827762E9C781}"
 EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Microsoft.ML.Recommend", "src\Microsoft.ML.Recommend\Microsoft.ML.Recommend.csproj", "{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}"
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU
@@ -501,6 +503,14 @@ Global
 		{11A5210E-2EA7-42F1-80DB-827762E9C781}.Release|Any CPU.Build.0 = Release|Any CPU
 		{11A5210E-2EA7-42F1-80DB-827762E9C781}.Release-Intrinsics|Any CPU.ActiveCfg = Release-Intrinsics|Any CPU
 		{11A5210E-2EA7-42F1-80DB-827762E9C781}.Release-Intrinsics|Any CPU.Build.0 = Release-Intrinsics|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Debug-Intrinsics|Any CPU.ActiveCfg = Debug-Intrinsics|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Debug-Intrinsics|Any CPU.Build.0 = Debug-Intrinsics|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Release|Any CPU.Build.0 = Release|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Release-Intrinsics|Any CPU.ActiveCfg = Release-Intrinsics|Any CPU
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3}.Release-Intrinsics|Any CPU.Build.0 = Release-Intrinsics|Any CPU
 	EndGlobalSection
 	GlobalSection(SolutionProperties) = preSolution
 		HideSolutionNode = FALSE
@@ -556,6 +566,7 @@ Global
 		{4B101D58-E7E4-4877-A536-A9B41E2E82A3} = {AED9C836-31E3-4F3F-8ABC-929555D3F3C4}
 		{ECB71297-9DF1-48CE-B93A-CD969221F9B6} = {DA452A53-2E94-4433-B08C-041EDEC729E6}
 		{11A5210E-2EA7-42F1-80DB-827762E9C781} = {09EADF06-BE25-4228-AB53-95AE3E15B530}
+		{C8E1772B-DFD9-4A4D-830D-6AAB1C668BB3} = {09EADF06-BE25-4228-AB53-95AE3E15B530}
 	EndGlobalSection
 	GlobalSection(ExtensibilityGlobals) = postSolution
 		SolutionGuid = {41165AF1-35BB-4832-A189-73060F82B01D}

src/Microsoft.ML.Console/Microsoft.ML.Console.csproj

@@ -27,12 +27,14 @@
     <ProjectReference Include="..\Microsoft.ML.Sweeper\Microsoft.ML.Sweeper.csproj" />
     <ProjectReference Include="..\Microsoft.ML.TensorFlow\Microsoft.ML.TensorFlow.csproj" />
     <ProjectReference Include="..\Microsoft.ML.Transforms\Microsoft.ML.Transforms.csproj" />
+    <ProjectReference Include="..\Microsoft.ML.Recommend\Microsoft.ML.Recommend.csproj" />
 
     <NativeAssemblyReference Include="FastTreeNative" />
     <NativeAssemblyReference Include="CpuMathNative" />
     <NativeAssemblyReference Include="FactorizationMachineNative" />
     <NativeAssemblyReference Include="LdaNative" />
     <NativeAssemblyReference Include="SymSgdNative" />
+    <NativeAssemblyReference Include="LibMFWrapper" />
   </ItemGroup>
 
-</Project>
+</Project>

src/Microsoft.ML.Recommend/IMIValueMapper.cs

@@ -0,0 +1,93 @@
+//------------------------------------------------------------------------------
+// <copyright company="Microsoft Corporation">
+//     Copyright (c) Microsoft Corporation. All rights reserved.
+// </copyright>
+//------------------------------------------------------------------------------
+
+using System;
+using Microsoft.ML.Runtime.Data;
+
+namespace Microsoft.ML.Runtime.Recommend
+{
+    /// <summary>
+    /// Interface for mapping a two input value (of the two indicated ColumnTypes) to
+    /// an output value (of an indicated ColumnType). This interface is commonly implemented
+    /// by predictors. Note that the input and output ColumnTypes determine the proper
+    /// type arguments for GetMapper, but typically contain additional information like
+    /// vector lengths.
+    /// </summary>
+    public interface IMIValueMapper
+    {
+        /// <summary>
+        /// The type of the input value for the matrix column index.
+        /// </summary>
+        ColumnType InputXType { get; }
+
+        /// <summary>
+        /// The type of the input value for the matrix row index.
+        /// </summary>
+        ColumnType InputYType { get; }
+
+        /// <summary>
+        /// The type of the output value, that is, the type of the matrix elements.
+        /// </summary>
+        ColumnType OutputType { get; }
+
+        /// <summary>
+        /// Get a delegate used for mapping from inputs to output values. Note that the delegate
+        /// should only be used on a single thread - it should NOT be assumed to be safe for concurrency.
+        /// </summary>
+        ValueMapper<TXIn, TYIn, TOut> GetMapper<TXIn, TYIn, TOut>();
+    }
+
+    /// <summary>
+    /// This is the interface for the 'items selection' mapper. It takes the user's ID, features and per-item affinities,
+    /// and produces ranked list of candidates.
+    ///
+    /// The interface is extensive, but the actual models will ignore some (or all) of the inputs. For example:
+    /// - the 'always suggest most popular items' model will ignore all inputs.
+    /// - the matrix factorization will ignore all the inputs except the user ID.
+    /// - SAR will ignore user ID and features and only read the item history.
+    /// - Matchbox in 'user to item' mode will use user ID and features and ignore the item history.
+    /// </summary>
+    /// <param name="userId">The user ID</param>
+    /// <param name="userFeatures">The user feature vector.</param>
+    /// <param name="items">The vector of items purchased by the user.</param>
+    /// <param name="weights">For each item in <paramref name="items"/>, the user's 'affinity' to the item.</param>
+    /// <param name="recommendedItems">Output. The ranked list of candidates.</param>
+    /// <param name="scores">Output, parallel to <paramref name="recommendedItems"/>. The scores of the ranked candidates.</param>
+    public delegate void UserHistoryToItemsMapper(
+        ref uint userId, ref VBuffer<Single> userFeatures, ref VBuffer<uint> items, ref VBuffer<Single> weights,
+        ref VBuffer<uint> recommendedItems, ref VBuffer<Single> scores);
+
+    /// <summary>
+    /// The interface for the 'user history -> items' recommender.
+    /// In the end-to-end recommendation pipeline, this predictor can be used both as a standalone recommender, and as a
+    /// candidate selection model. In the second case, the recommended items will then be featurized and ranked by downstream
+    /// component.
+    /// </summary>
+    public interface IUserHistoryToItemsRecommender: IPredictor
+    {
+        /// <summary>
+        /// The number of user features. 0 means variable size, -1 means that the algorithm ignores the user features.
+        /// </summary>
+        int UserFeaturesSize { get; }
+
+        /// <summary>
+        /// The type of the user ID. This will be a U4 key or null, if the algorithm doesn't care about the user ID.
+        /// </summary>
+        KeyType UserIdType { get; }
+
+        /// <summary>
+        /// The type of the item ID, both inputs and outputs. This will be a U4 key.
+        /// </summary>
+        KeyType ItemIdType { get; }
+
+        /// <summary>
+        /// Returns a mapper to invoke for item recommendations.
+        /// </summary>
+        /// <param name="recommendationCount">Maximum number of recommendation to return.</param>
+        /// <param name="includeHistory">Whether to recommend items in the user's history.</param>
+        UserHistoryToItemsMapper GetRecommendMapper(int recommendationCount, bool includeHistory);
+    }
+}