Author Age Prediction

This is a author age categorizer that leverages the Apache OpenNLP Maximum Entropy Classifier. It takes a text sample and classifies it into the following age categories: xx-18|18-24|25-34|35-49|50-64|65-xx.

Pre-Requisites

1. Download Apache Spark 2.0.0 and place in the local directory for this checkout. Once downloaded remember to run tar xvzf spark-2.0.0-bin-hadoop2.7.tgz
2. export SPARK_HOME="spark-2.0.0-bin-hadoop2.7"
3. Run bin/download-opennlp.sh to download Apache OpenNLP models referenced below.
4. Run mvn clean install to build the assembly jars. The key one you need is age-predictor-assembly/target/age-predictor-assembly-1.1-SNAPSHOT-jar-with-dependencies.jar. If you do not see this jar, investigate your Maven and Java issues. It should build fine with

openjdk version "13.0.2" 2020-01-14
OpenJDK Runtime Environment (build 13.0.2+8)
OpenJDK 64-Bit Server VM (build 13.0.2+8, mixed mode, sharing)
MT-202397:AgePredictor mattmann$ mvn --V
Apache Maven 3.6.1 (d66c9c0b3152b2e69ee9bac180bb8fcc8e6af555; 2019-04-04T12:00:29-07:00)
Maven home: /usr/local/Cellar/maven/3.6.1/libexec
Java version: 13.0.2, vendor: Oracle Corporation, runtime: /Library/Java/JavaVirtualMachines/openjdk-13.0.2.jdk/Contents/Home
Default locale: en_US, platform encoding: UTF-8
OS name: "mac os x", version: "10.15.7", arch: "x86_64", family: "mac"     

QuickStart

1. Follow the instructions to perform training, and build yourself a model/en-ageClassify.bin file
   • bin/authorage AgeClassifyTrainer -model model/en-ageClassify.bin -lang en -data data/sample_train.txt -encoding UTF-8
2. Run the Age prediction with the sample data
   • bin/authorage AgePredict ./model/classify-unigram.bin ./model/regression-global.bin data/sample_test.txt < data/sample_test.txt
3. Run the Age prediction and grep out the predictions from the sample data
   • bin/authorage AgePredict ./model/classify-unigram.bin ./model/regression-global.bin data/sample_test.txt < data/sample_test.txt 2>&1 | grep "Prediction"
   • If you see as output from the above command you're good!

Prediction: 33.25378998833527
Prediction: 31.67628280063772

Usage

How to train an Age Classifier

Note: The training data should be a line-by-line, with each line starting with the age, or age category, followed by a tab and the text associated with the age.

Usage: bin/authorage AgeClassifyTrainer [-factory factoryName] [-featureGenerators featuregens] [-tokenizer tokenizer] -model modelFile [-params paramsFile] -lang language -data sampleData [-encoding charsetName]

Arguments description:
	-factory factoryName
        a sub-class of DoccatFactory where to get implementation and resources.
	-featureGenerators featuregens
	    comma separated feature generator classes. Bag of words default.
	-tokenizer tokenizer
        tokenizer implementation. WhitespaceTokenizer is used if not specified.
	-model modelFile
        output model file.
	-params paramsFile
	    training parameters file.
	-lang language
	    language which is being processed.
	-data sampleData
	    data to be used, usually a file name.
	-encoding charsetName
	    encoding for reading and writing text, if absent the system default is used.

Example Usage:

bin/authorage AgeClassifyTrainer -model model/en-ageClassify.bin -lang en -data data/sample_train.txt -encoding UTF-8

Training data format - Age and text seperated by tab in each line like <AGE><Tab><TEXT>
Sample training data-

12	I am just 12 year old
25	I am little bigger
35	I am mature
45	I am getting old
60	I am old like wine

How to evaluate an Age Classifier Model

Usage: bin/authorage AgeClassifyEvaluator -model model [-misclassified true|false] -data sampleData [-encoding charsetName]

Arguments description:
	-model model
		the model file to be evaluated.
	-misclassified true|false
		if true will print false negatives and false positives.
	-data sampleData
		data to be used, usually a file name.
	-encoding charsetName
		encoding for reading and writing text, if absent the system default is used.

Example Usage:

bin/authorage AgeClassifyEvaluator -model model/en-ageClassify.bin -data data/sample_test.txt -encoding UTF-8

How to run the Age Classifier

Note: Each document must be followed by an empty line to be detected as a separate case from the others.

Usage: bin/authorage AgeClassify model < documents

Usage: bin/authorage AgePredict ./model/classify-unigram.bin ./model/regression-global.bin  data/sample_test.txt < data/sample_test.txt

Downloads

For AgePredict to work you need to download en-pos-maxent.bin, en-sent.bin and en-token.bin from http://opennlp.sourceforge.net/models-1.5/ to model/opennlp/

Citation:

If you use this work, please cite:

@article{hong2017ensemble,
  title={Ensemble Maximum Entropy Classification and Linear Regression for Author Age Prediction},
  author={Hong, Joey and Mattmann, Chris and Ramirez, Paul},
  booktitle={Information Reuse and Integration (IRI), 2017 IEEE 18th International Conference on},
  organization={IEEE}
  year={2017}
}

Contributors

• Chris A. Mattmann, JPL & USC
• Joey Hong, Caltech
• Madhav Sharan, JPL & USC

License

Apache License, version 2