Content Enrichment in the Text Retrieval Conference (TREC) Polar Dynamic Domain Dataset

Performing Content Extraction, Extracting Metadata, Information Clustering and Similarity, and Named Entity Recognition – to scientifically enrich the Polar dataset and to contribute to http://polar.usc.edu , a new website being created to provide Polar Data Insights to the scientific community

1. Context Extraction Enrichment – Applying the Tag Ratios algorithm to identify text, and you will construct a Tika parser to extract Measurement mentions from text automatically.
   

2. Metadata Enrichment – Applying the GROBID journal parser with Tika, and extract TEI metadata, and also scientific publication metadata using the Google Scholar API to develop a network of related scientific publications to your Polar dataset, and to map publications to the data. In addition, classifying the data using a common Earth science domain model, ontology, called SWEET, for Semantic Web for Earth and Environmental Terminology (http://sweet.jpl.nasa.gov/). Also creating Digital Object Identifiers (DOIs) for your data.
   

3. Information Similarity and Clustering – Creating clusters of your Polar data using the enriched measurements extracted, and using the enriched metadata, and browse and expose your information using Data-Driven-Documents visualizations after ingesting data into Apache Solr and/or ElasticSearch.
   

4. Named Entity Recognition (NER) – Applying geospatial NER using the GeoTopicParser in Apache Tika and using the MEMEX GeoParser tools

Charan Shampur
Manisha Kampasi

Apache Tika https://tika.apache.org/license.html