Querying SPARQL endpoint with R and plotting results with R graphics package

gistfile1.r

######################################################
#
# Example of querying Linked Open Aalto Data Service
# SPARQL endpoint with the SPARQL R client, and
# plotting data with the R graphics package.
#
# http://data.aalto.fi/
#
# Tuija Sonkkila, 12.11.2012
#
######################################################

# http://cran.r-project.org/web/packages/SPARQL/index.html
library(SPARQL)

endpoint <- "http://data.aalto.fi/sparql"

# Prepare the query 
#
# Example is from http://data.aalto.fi/vizplay
# adapted with namespace handling introduced at
# http://linkedscience.org/tools/sparql-package-for-r/tutorial-on-sparql-package-for-r/

q <- "SELECT ?date (COUNT(?member) AS ?contributors )(COUNT(DISTINCT ?publication) AS ?publications ) 
WHERE { 
 ?publication <http://purl.org/ontology/bibo/authorList> ?list ; 
              <http://purl.org/dc/terms/date> ?date. 
 ?list <http://www.w3.org/1999/02/22-rdf-syntax-ns#rest> */
       <http://www.w3.org/1999/02/22-rdf-syntax-ns#first> ?member . 
} 
 GROUP BY ?date 
 ORDER BY ?date"

# Query the endpoint, and store results in a data frame
res <- SPARQL(url=endpoint, q)$results

# Find out about e.g. min and max values
summary(res)

# Reduce font size
par(cex = 0.8) 

# Plot data about publications in blue
plot(res$date,             
     res$publications, 
     main = "Aalto University publication statistics",
     type = "b",                # join symbols with lines
     pch = 2,                   # plot symbol
     lty = 2,                   # plot line type
     col = "blue",              
     ylim = c(0, 21000),        # Y axis limits
     xaxt = "n", yaxt = "n",    # but don't draw axes yet
     xlab = "", ylab = "")      # nor axis labels

# In the same plot, add data about contributors
lines(res$date, 
      res$contributors, 
      type = "b", 
      pch = 2, 
      lty = 2, 
      col = "red")

# Add legend
legend("topleft", 
       inset = c(.05,.05), 
       c("Publications", "Contributors"), 
       pch = c(2, 2),
       col = c("blue","red"))

# Draw X axis tick marks without labels
axis(1, 
     at = seq(1975, 2012, by = 2), 
     labels = FALSE, 
     col.axis = "black", 
     lty = 1)

# Character vector of X axis labels
xlablist <- as.vector(seq(1975, 2012, by = 2))

# Draw X axis labels in 45 degree angle
# http://jnlnet.wordpress.com/2009/05/20/x-axis-labels-on-a-45-degree-angle-using-r/
text(seq(1975, 2012, by = 2), 
     par("usr")[3], 
     adj = c(1.5,1.5), 
     labels = xlablist, 
     srt = 45, 
     xpd = TRUE)

# Draw the Y axis
axis(2)

# Add horizontal and vertical lines at tick points of both axes
abline(h = c(seq(1, 21000, by = 1000)), lty = 3, col = "gray")
abline(v = c(seq(1975, 2012, by = 2)), lty = 3, col = "gray")

# Add some explanatory text in italics without a box
par(font = 3)
legend("bottomleft", 
       inset = c(.18,.08), 
       bty = "n", 
       "Data officially since 1992")

# Restore the default font size and type
par(cex = 1, font = 1)