Examples of edu.umd.cloud9.io.map.HMapIFW

edu.umd.cloud9.io.map.HMapIFW

Writable representing a map where keys are ints and values are floats.

One notable feature of this class is the ability to support lazy decoding, controlled by the {@link #setLazyDecodeFlag(boolean)} method. In lazydecoding mode, when an object of this type is deserialized, key-value pairs are not inserted into the map, but rather held in arrays. The reduces memory used in cases where random access to values is not required. In lazy decoding mode, the raw keys and values may be fetched by the {@link #getKeys()} and{@link #getValues()} methods, respectively. The map can be subsequentlypopulated with the {@link #decode()} method.
@author Jimmy Lin

   *     mapping from F-terms to their df values
   * @return
   *     mapping from E-terms to their computed df values
   */
  public static HMapIFW translateDFTable(Vocab eVocabSrc, Vocab fVocabTrg, TTable_monolithic_IFAs e2f_probs, HMapSIW dfs){
    HMapIFW transDfTable = new HMapIFW();
    for(int e=1;e<eVocabSrc.size();e++){
      int[] fS = null;
      try {
        fS = e2f_probs.get(e).getTranslations(0.0f);
      } catch (Exception e1) {
        e1.printStackTrace();
      }
      float df=0;
      for(int f : fS){
        float probEF = e2f_probs.get(e, f);
        String fTerm = fVocabTrg.get(f);
        if(!dfs.containsKey(fTerm)){  //only if word is in the collection, can it contribute to the df values.
          continue;
        }      
        float df_f = dfs.get(fTerm);
        df+=(probEF*df_f);
      }
      transDfTable.put(e, df);
    }
    return transDfTable;
  }

View Full Code Here


    List<HMapSFW> transDocs = new ArrayList<HMapSFW>();


    // translate doc texts here
    for (HMapSIW deDoc : docs) {
      HMapIFW tfS = new HMapIFW();
      int docLen = 0;
      try {
        Tokenizer tokenizer = TokenizerFactory.createTokenizer(eLang, null, true, eStopwordsFile, eStopwordsFile + ".stemmed", null);
        docLen = CLIRUtils.translateTFs(deDoc, tfS, eVocabSrc, eVocabTrg, fVocabSrc,
            fVocabTrg, e2f_Probs, f2e_Probs, tokenizer , null);   // tokenizer just for stopword list

View Full Code Here

    }


    public void map(IntWritable docno, WeightedIntDocVector docvectorIn,
        OutputCollector<IntWritable, MinhashSignature> output, Reporter reporter)
        throws IOException {
      HMapIFW docvector = docvectorIn.getWeightedTerms();
      signature.clear();


      for (int i = 0; i < randomOrderings.size(); i++) {
        int minTerm = getMinHashTerm(docvector, (ArrayListOfIntsWritable) randomOrderings.get(i));
        signature.add(minTerm);

View Full Code Here

    for(String term : terms){
      term2Tf.increment(term);
    }


    //translated tf values
    HMapIFW transTermTf = new HMapIFW();
    for(Entry<String> entry : term2Tf.entrySet()){
      String fTerm = entry.getKey();
      int tf = entry.getValue();
      // transTermTf won't be updated if fTerm not in vocab
      transTermTf = CLIRUtils.updateTFsByTerm(fTerm, tf, transTermTf, eVocabSrc, eVocabTrg, fVocabSrc, fVocabTrg, e2f_Probs, f2e_Probs, eTok, sLogger);

View Full Code Here

      return s;
    }


    public void map(IntWritable docno, WeightedIntDocVector docvectorIn,
        OutputCollector<IntWritable, NBitSignature> output, Reporter reporter) throws IOException {
      HMapIFW docvector = docvectorIn.getWeightedTerms();
      FloatAsBytesWritable value;


      for (int i = 0; i < randomUnitVectors.size(); i++) {
        value = (FloatAsBytesWritable) randomUnitVectors.get(i);
        double dprod = dotProduct(docvector, value);

View Full Code Here

        if (e1.getDocno() < mBlockStart)
          continue;
        if (e1.getDocno() >= mBlockEnd)
          break;


        HMapIFW map = new HMapIFW();


        sLogger.debug(key + ": " + e1);


        PostingsReader reader2 = postings.getPostingsReader();


        while (reader2.nextPosting(e2)) {


          sLogger.debug(key + ": " + e1 + ", " + e2);


          if (e1.getDocno() == e2.getDocno())
            continue;


          // compute partial score of similarity for a pair of
          // documents
          float weight = mModel.computeScore(e1.getTf(), e2.getTf(),
              mDocLengthTable.getDocLength(e1.getDocno()), mDocLengthTable
                  .getDocLength(e2.getDocno()));


          map.put(e2.getDocno(), weight);
        }
        output.collect(new IntWritable(e1.getDocno()), map);
      }
    }

View Full Code Here

  private int docLength;
  private HMapIFW weightedTerms;


  public WeightedIntDocVector () {
    docLength = 0;
    weightedTerms = new HMapIFW ();
  }

View Full Code Here

    weightedTerms.write (out);
  }


  public void readFields (DataInput in) throws IOException {
    docLength = WritableUtils.readVInt (in);
    weightedTerms = new HMapIFW ();
    weightedTerms.readFields (in);
  }

View Full Code Here

   *     FileSystem object
   * @return
   *     mapping from term ids to df values
   */
  public static HMapIFW readTransDfTable(Path path, FileSystem fs) {
    HMapIFW transDfTable = new HMapIFW();
    try {
      SequenceFile.Reader reader = new SequenceFile.Reader(fs, path, fs.getConf());


      IntWritable key = (IntWritable) reader.getKeyClass().newInstance();
      FloatWritable value = (FloatWritable) reader.getValueClass().newInstance();


      while (reader.next(key, value)) {
        transDfTable.put(key.get(), value.get());
        //        logger.info(key.get()+"-->"+value.get());
        key = (IntWritable) reader.getKeyClass().newInstance();
        value = (FloatWritable) reader.getValueClass().newInstance();
      }
      reader.close();

View Full Code Here

   *     ttable E-->F (i.e., Pr(f|e))
   * @return
   *     mapping from E-terms to their computed df values
   */
  public static HMapIFW translateDFTable(Vocab eVocabSrc, Vocab fVocabTrg, TTable_monolithic_IFAs e2f_probs, FrequencySortedDictionary dict, DfTableArray dfTable){
    HMapIFW transDfTable = new HMapIFW();
    for(int e=1;e<eVocabSrc.size();e++){
      int[] fS = e2f_probs.get(e).getTranslations(0.0f);
      float df=0;
      for(int f : fS){
        float probEF = e2f_probs.get(e, f);
        String fTerm = fVocabTrg.get(f);
        int id = dict.getId(fTerm); 
        if(id != -1){
          float df_f = dfTable.getDf(id);        
          df += (probEF*df_f);
        }else{
          logger.debug(fTerm+" not in dict");
        }
      }
      transDfTable.put(e, df);
    }
    return transDfTable;
  }

View Full Code Here

0 1 2 3 4 5 6 7 8 9

TOP

Related Classes of edu.umd.cloud9.io.map.HMapIFW

bak.pcj.IntIterator

edu.umd.cloud9.example.bfs.BfsNodeTest

edu.umd.cloud9.example.pagerank.PageRankNodeTest

edu.umd.cloud9.io.array.ArrayListOfIntsWritable

edu.umd.cloud9.io.fastutil.Int2FloatOpenHashMapWritableTest

edu.umd.cloud9.io.fastutil.Int2IntOpenHashMapWritableTest

edu.umd.cloud9.io.fastutil.String2FloatOpenHashMapWritableTest

edu.umd.cloud9.io.fastutil.String2IntOpenHashMapWritableTest

edu.umd.cloud9.io.map.HMapIDW

edu.umd.cloud9.io.map.HMapIDWTest

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.