digg2clickstream.R

#---------load data------------#
setwd("F:/digg/")
ft = read.csv("./final_front_zero_mean_time.csv", head=T, na.string='NA', stringsAsFactors=T)
dat = read.csv("./digg_votes_threshold.csv", header=T, stringsAsFactors = F)
# storyid   user       time   threshold
# 1 oay as a session

time = dat$time
time = as.POSIXct(time, origin="1970-01-01")
time = as.numeric(as.Date(time)) - 14394

get_daily_data = function(i){
  datd = dat[which(time == i), 1:3]
  # sort by uer, time
  datd = datd[with(datd, order(user, time)), ]
  return(datd)
}

get_individual_stream =  function(i){
  day_user = data[which(data$user == unique_user[i]),]
  story_list = c("source", day_user$storyid, "sink")
  stream = embed(story_list, 2)[,2:1]
  return(stream)
}

get_daily_clickstream = function(unique_user){
  streams = lapply(1:length(unique_user), get_individual_stream)
  streams = do.call(rbind, streams)
  return(streams)
}


plot_clickstream = function(net){
  require(igraph)
  g =graph.data.frame(net,directed=TRUE )
  E(g)$weight = net[,3]
  g = simplify(g,  remove.loops = TRUE)
  set.seed(34)   
  l=layout.fruchterman.reingold(g)
  V(g)$size =  log(centralization.degree(g)$res+ 1) 
  E(g)$width = (E(g)$weight-min(E(g)$weight))/1000
  node_name = V(g)$name
  node_name[which(!node_name%in%c("source", "sink"))] = ""
  plot(g, vertex.label= node_name,
       edge.curved =  FALSE, vertex.frame.color="#FFFFFF",
       vertex.label.cex =1.2,  edge.arrow.size=0.1, layout=l )
}




day_id = 10
data = get_daily_data(day_id )
data = unique(data)
unique_user = unique(data$user)
net = data.frame(get_daily_clickstream(unique_user)); dim(net)
net$weight = 1
net = aggregate(weight ~ X1 + X2, FUN = sum, data=net); dim(net)
plot_clickstream(net)