Description

This project applies LDA topic modeling over SPARQL query results, in order to recognize if one dataset contains the same topics as another. More spesificly in this case we are look for topic simalerity between semantic vocabularies.

Requirements

R version 3.3.2
Download the following packages: SPARQL, mallet, rjava, rvest, ggplot2
Can be installed in Rstudio

Procedure

"Main" contains all scrips needed.
To run the program run
Main.r
or
track the scripts one by one in ascending order:
1-Functions.r
2-WebScraper.r
2-SPARQL.r
3-SparqlMassage.r
3-WebScraperMassage.r
4-FinalDataMassage.r
5-TopicModeling.r
6-TermCorrelation.r
6-Visualization.r

If you want to change the search query yourself
-Open 4-FinalDataMassage.r
-Look for naturalLanguagelVocabularySearch("X",corpus.df)
-Replace the Text where X is, with your own.
-Then run the whole program again.
-How ever the ontologies harvested will be the first 100 found on linken open vocabularies
and foaf terms in more details.

Depricated

The SPARQL endpoint originally used seems to have changed output : http://lov.okfn.org/dataset/lov/sparql I have yet to fix this, but inputs and configurations can be changed to suit other SPARQL endpoints.