Skip to content

Getting Started

Jump to bottom
Nithin Krishna edited this page Aug 10, 2016 · 9 revisions

#Tutorial

The following tutorial explains how polar deep insights can be used to gain insights about climate change and climate change indicators over the trec polar dataset.

##1) Define an ontology We start off with defining an ontology based on climate change indicators based on the EPA. We also look subtrees in the JPL sweet ontology which describes system state change and chemical process.

Once we've defined this ontology we persist it. We then export this ontology onto a file (entities.txt).

##2) Extract meta information

  • We need to prepare our elastic search index for extraction.
  • We replace the generated entities.txt file with the one in ./insight-generator/data.
  • We then begin the extraction script over our data set.

##3) Start analysis Once the elastic search indices have been populated we can perform the following kinds of analysis with the insight visualizer.

###Time variance In all the documents talking about 'wild fire', what is the variation of occurrence of time?

###Geographic Distribution In all the documents talking about 'wild fire', what is the geographic distribution by document count?

###Concept Correlation(Co-occurrence) In all the documents talking about 'glacier retreat', what are some of the other frequent occurring concepts?

###Measurement Distribution In all the documents talking about 'glacier retreat', what are the measurements extracted and their ranges?

###Geo Diversity In all the documents talking about 'glacier retreat', what are some of the prominent places?

Information Retrieval and Data Science (IRDS) research group, University of Southern California.

Clone this wiki locally