Skip to content

A generic and semantic profiling of 1159 New York City open datasets using Apache Spark. The generic profiling has been performed using the Spark RDDs and Dataframes. Semantic profiling has been performed using Named Entity Recognition, Soundex, Regex, Ontologies and Clustering.

Notifications You must be signed in to change notification settings

ankushjain2001/NYC-Open-Data-Profiling

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

55 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

The Generic and Semantic Profiling of Big Datasets


We call ourselves CreaTAK
T is for Theodore
A is for Ankush
K is for Kat
...and we all like to create stuff.

In this study, we ran Apache Spark over NYU’s 48-node Hadoop cluster, running Cloudera CDH 5.15.0, to generically and semantically profile 1159 datasets from NYC Open Data. We refer to these two profiling methods as Task 1 and Task 2, respectively.

Of the 1159 files we profiled in Task 1, we found 11674 integer columns, 13646 text columns, 1137 date/time columns, and 4527 real number columns. For Task 2, we analyzed 260 columns and we were able to identify the semantic types for 210 columns with a precision of 72.40%.


Instructions

  • Log into NYU's DUMBO.
  • Load the correct versions of Python and Spark:
    • module load python/gnu/3.6.5
    • module load spark/2.4.0

Task 1: Generic Profiling

  • Navigate to task1/src/
  • Type the following to run:
    • spark-submit --conf spark.pyspark.python=$PYSPARK_PYTHON task1.py

Task 2: Generic Profiling

  • Navigate to task2/src/
  • Type the following to run:
    • spark-submit --conf spark.pyspark.python=$PYSPARK_PYTHON task2.py
  • Use task2_md.py to execute with 'en_web_core_md' NLP model. This model was used to produce the final results.
  • This model is not available on dumbo by default. Use the following command to install the package.

python -m spacy download en_core_web_md

About

A generic and semantic profiling of 1159 New York City open datasets using Apache Spark. The generic profiling has been performed using the Spark RDDs and Dataframes. Semantic profiling has been performed using Named Entity Recognition, Soundex, Regex, Ontologies and Clustering.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published