[GitHub] spark pull request: [SPARK-2199] [mllib] topic modeling

[karlhigley <gi...@git.apache.org>]

Github user karlhigley commented on a diff in the pull request:

    https://github.com/apache/spark/pull/1269#discussion_r18312090
  
    --- Diff: mllib/src/main/scala/org/apache/spark/mllib/clustering/topicmodeling/topicmodels/RobustPLSA.scala ---
    @@ -0,0 +1,180 @@
    +/*
    + * Licensed to the Apache Software Foundation (ASF) under one or more
    + * contributor license agreements.  See the NOTICE file distributed with
    + * this work for additional information regarding copyright ownership.
    + * The ASF licenses this file to You under the Apache License, Version 2.0
    + * (the "License"); you may not use this file except in compliance with
    + * the License.  You may obtain a copy of the License at
    + *
    + *    http://www.apache.org/licenses/LICENSE-2.0
    + *
    + * Unless required by applicable law or agreed to in writing, software
    + * distributed under the License is distributed on an "AS IS" BASIS,
    + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    + * See the License for the specific language governing permissions and
    + * limitations under the License.
    + */
    +
    +package org.apache.spark.mllib.clustering.topicmodeling.topicmodels
    +
    +
    +import java.util.Random
    +
    +import org.apache.spark.broadcast.Broadcast
    +import org.apache.spark.mllib.clustering.topicmodeling.documents.Document
    +import org.apache.spark.mllib.clustering.topicmodeling.topicmodels.regulaizers.{DocumentOverTopicDistributionRegularizer, TopicsRegularizer, UniformDocumentOverTopicRegularizer, UniformTopicRegularizer}
    +import org.apache.spark.rdd.RDD
    +import org.apache.spark.{Logging, SparkContext}
    +
    +
    +/**
    + * distributed topic modeling via RobustPLSA (Hofmann (1999), Vorontsov, Potapenko (2014) )
    + *
    + * @param sc  spark context
    + * @param numberOfTopics number of topics
    + * @param numberOfIterations number of iterations
    + * @param random java.util.Random need for initialisation
    + * @param documentOverTopicDistributionRegularizer
    + * @param topicRegularizer
    + * @param computePpx boolean. If true, model computes perplexity and prints it puts in the log at
    + *                   INFO level. it takes some time and memory
    + * @param gamma weight of background
    + * @param eps   weight of noise
    + */
    +class RobustPLSA(@transient protected val sc: SparkContext,
    +                 protected val numberOfTopics: Int,
    +                 protected val numberOfIterations: Int,
    +                 protected val random: Random,
    +                 private val documentOverTopicDistributionRegularizer:
    +                  DocumentOverTopicDistributionRegularizer =
    +                          new UniformDocumentOverTopicRegularizer,
    +                 @transient protected val topicRegularizer: TopicsRegularizer =
    +                                                        new UniformTopicRegularizer,
    +                 private val computePpx: Boolean = true,
    +                 private val gamma: Float = 0.3f,
    +                 private val eps: Float = 0.01f)
    +  extends AbstractPLSA[RobustDocumentParameters, RobustGlobalParameters, RobustGlobalCounters]
    +  with Logging
    +  with Serializable {
    +
    +
    +  /**
    +   *
    +   * @param documents  -- document collection
    +   * @return a pair of rdd of document parameters global parameters
    +   */
    +  override def infer(documents: RDD[Document])
    +      : (RDD[RobustDocumentParameters], RobustGlobalParameters) = {
    +    val alphabetSize = getAlphabetSize(documents)
    +    EM(documents, getInitialTopics(alphabetSize), alphabetSize, foldingIn = false)
    +  }
    +
    +
    +  /**
    +   *
    +   * @param documents  docs to be folded in
    +   * @param globalParams global parameters that were produced by infer method (stores topics)
    +   * @return
    +   */
    +  override def foldIn(documents: RDD[Document],
    +                      globalParams: RobustGlobalParameters): RDD[RobustDocumentParameters] = {
    +    EM(documents, sc.broadcast(globalParams.phi), globalParams.alphabetSize, foldingIn = true)._1
    +  }
    +
    +  private def EM(documents: RDD[Document],
    +                 topicBC: Broadcast[Array[Array[Float]]],
    +                 alphabetSize : Int,
    +                 foldingIn : Boolean) :(RDD[RobustDocumentParameters], RobustGlobalParameters) = {
    +    val collectionLength = getCollectionLength(documents)
    +
    +    val parameters = documents.map(doc => RobustDocumentParameters(doc,
    +                                                  numberOfTopics,
    +                                                  gamma,
    +                                                  eps,
    +                                                  documentOverTopicDistributionRegularizer))
    +
    +    val background = Array.fill(alphabetSize)(1f / alphabetSize)
    +
    +    val (result, topics, backgound) = newIteration(parameters,
    --- End diff --
    
    The name of the third return value here should be `background`.  This appears to change the background in the global parameters returned by this method.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---

---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org