├── .github └── workflows │ └── sbt-publish.yml ├── .gitignore ├── .travis.yml ├── LICENSE ├── README.md ├── data ├── .gitignore ├── annoy │ └── sample-glove-25-angular.txt ├── download.sh ├── dump.py ├── fasttext │ └── cc.ko.300.vec.sample.gz └── mllib │ └── als │ ├── LICENSE │ └── sample_movielens_ratings.txt ├── dev ├── accuracy_test.py └── hdf5_to_parquet.py ├── project ├── build.properties └── plugins.sbt ├── src ├── main │ └── scala │ │ ├── org │ │ └── apache │ │ │ └── spark │ │ │ └── ml │ │ │ └── nn │ │ │ ├── Annoy.scala │ │ │ ├── NodeUDT.scala │ │ │ └── VectorUDT.scala │ │ └── sparkannoy │ │ ├── AnnoyUtil.scala │ │ ├── Distance.scala │ │ ├── Index.scala │ │ ├── IndexAggregator.scala │ │ ├── Node.scala │ │ └── Vectors.scala └── test │ ├── java │ └── sparkannoy │ │ └── spark │ │ └── Kiss32Random.java │ └── scala │ └── sparkannoy │ ├── AccuracyTest.scala │ ├── ArithmeticTest.scala │ ├── IndexAggregatorTest.scala │ ├── KissRandomTest.scala │ ├── LocalBuilds.scala │ └── spark │ ├── CollectTest.scala │ ├── LocalSparkApp.scala │ ├── NodeUDTTest.scala │ ├── PrepareDataset.scala │ └── example │ ├── ALSBasedUserItemIndexing.scala │ ├── DistributedBuilds.scala │ └── FastTextIndexing.scala └── version.sbt /.github/workflows/sbt-publish.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/.github/workflows/sbt-publish.yml -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/.gitignore -------------------------------------------------------------------------------- /.travis.yml: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/.travis.yml -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/LICENSE -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/README.md -------------------------------------------------------------------------------- /data/.gitignore: -------------------------------------------------------------------------------- 1 | *.bin 2 | *.hdf5 3 | train 4 | test 5 | -------------------------------------------------------------------------------- /data/annoy/sample-glove-25-angular.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/data/annoy/sample-glove-25-angular.txt -------------------------------------------------------------------------------- /data/download.sh: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/data/download.sh -------------------------------------------------------------------------------- /data/dump.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/data/dump.py -------------------------------------------------------------------------------- /data/fasttext/cc.ko.300.vec.sample.gz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/data/fasttext/cc.ko.300.vec.sample.gz -------------------------------------------------------------------------------- /data/mllib/als/LICENSE: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/data/mllib/als/LICENSE -------------------------------------------------------------------------------- /data/mllib/als/sample_movielens_ratings.txt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/data/mllib/als/sample_movielens_ratings.txt -------------------------------------------------------------------------------- /dev/accuracy_test.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/dev/accuracy_test.py -------------------------------------------------------------------------------- /dev/hdf5_to_parquet.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/dev/hdf5_to_parquet.py -------------------------------------------------------------------------------- /project/build.properties: -------------------------------------------------------------------------------- 1 | sbt.version = 0.13.15 -------------------------------------------------------------------------------- /project/plugins.sbt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/project/plugins.sbt -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/ml/nn/Annoy.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/org/apache/spark/ml/nn/Annoy.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/ml/nn/NodeUDT.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/org/apache/spark/ml/nn/NodeUDT.scala -------------------------------------------------------------------------------- /src/main/scala/org/apache/spark/ml/nn/VectorUDT.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/org/apache/spark/ml/nn/VectorUDT.scala -------------------------------------------------------------------------------- /src/main/scala/sparkannoy/AnnoyUtil.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/sparkannoy/AnnoyUtil.scala -------------------------------------------------------------------------------- /src/main/scala/sparkannoy/Distance.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/sparkannoy/Distance.scala -------------------------------------------------------------------------------- /src/main/scala/sparkannoy/Index.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/sparkannoy/Index.scala -------------------------------------------------------------------------------- /src/main/scala/sparkannoy/IndexAggregator.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/sparkannoy/IndexAggregator.scala -------------------------------------------------------------------------------- /src/main/scala/sparkannoy/Node.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/sparkannoy/Node.scala -------------------------------------------------------------------------------- /src/main/scala/sparkannoy/Vectors.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/main/scala/sparkannoy/Vectors.scala -------------------------------------------------------------------------------- /src/test/java/sparkannoy/spark/Kiss32Random.java: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/java/sparkannoy/spark/Kiss32Random.java -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/AccuracyTest.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/AccuracyTest.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/ArithmeticTest.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/ArithmeticTest.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/IndexAggregatorTest.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/IndexAggregatorTest.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/KissRandomTest.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/KissRandomTest.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/LocalBuilds.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/LocalBuilds.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/spark/CollectTest.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/spark/CollectTest.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/spark/LocalSparkApp.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/spark/LocalSparkApp.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/spark/NodeUDTTest.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/spark/NodeUDTTest.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/spark/PrepareDataset.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/spark/PrepareDataset.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/spark/example/ALSBasedUserItemIndexing.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/spark/example/ALSBasedUserItemIndexing.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/spark/example/DistributedBuilds.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/spark/example/DistributedBuilds.scala -------------------------------------------------------------------------------- /src/test/scala/sparkannoy/spark/example/FastTextIndexing.scala: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/em3s/spark-annoy/HEAD/src/test/scala/sparkannoy/spark/example/FastTextIndexing.scala -------------------------------------------------------------------------------- /version.sbt: -------------------------------------------------------------------------------- 1 | version in ThisBuild := "0.1.6-SNAPSHOT" 2 | --------------------------------------------------------------------------------