Traven_stylometry/04_imposters_analysis.R at main · SimoneRebora/Traven_stylometry · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
### Imposters_analysis

library(stylo)

methods_combination <- read.csv("Methods_combination_parallel.csv", stringsAsFactors = F)

load(paste("Preprocessed_dataset_", methods_combination$unit[1], ".RData", sep = ""))

args <- commandArgs(trailingOnly=TRUE)
args_split <- unlist(strsplit(args, " "))

start_loop <- as.numeric(args_split[1])
end_loop <- as.numeric(args_split[2])
validation_rounds <- as.numeric(args_split[3])

filename <- paste("Final_results_", methods_combination$unit[1], "_", start_loop, ".csv", sep = "")

### 0. Functions

# new minmax function
dist.minmaxfast = function(x){

  # test if the input dataset is acceptable
  if(is.matrix(x) == FALSE & is.data.frame(x) == FALSE) {
    stop("cannot apply a distance measure: wrong data format!")
  }
  # then, test whether the number of rows and cols is >1
  if(length(x[1,]) < 2 | length(x[,1]) < 2) {
    stop("at least 2 cols and 2 rows are needed to compute a distance!")
  }

  # getting the size of the input table
  rows = length(x[,1])
  # starting a new matrix
  y = matrix(nrow = rows, ncol = rows)
  rownames(y) = rownames(x)
  colnames(y) = rownames(x)
  # iterating over rows and columns
  for(i in 1:rows) {
    for(j in i:rows ) {
      y[j,i] = 1 - sum(pmin(x[i,], x[j,])) / sum(pmax(x[i,], x[j,]))
    }
  }

  # converting the matrix to the class 'dist'
  y = as.dist(y)

  return(y)
}

### 1. First phase: testing

full_results <- data.frame()

for(method in start_loop:end_loop){

  # cat("\n######\n######\n###### Configuration", method, "\n\n")

  # cat("\n######\n######\n Right author\n\n")

  for(candidate_id in unique(full_corpus_metadata$candidate_id)){

    # cat("\n######\n no.", candidate_id, "\n\n")

    available_corpora <- which(full_corpus_metadata$analysis_type == "verifications" &
                                 full_corpus_metadata$impostors == methods_combination$n_imposters[method] &
                                 full_corpus_metadata$candidate_id == candidate_id)

    for(validation in 1:validation_rounds){

      # select random text
      selected_id <- sample(available_corpora, 1)

      # load preprocessed dataset
      data <- full_corpus_list[[selected_id]]

      # culling
      data <- perform.culling(data, culling.level = methods_combination$culling[method])
      if(dim(data)[2] > methods_combination$MFU[method])
        data <- data[,1:methods_combination$MFU[method]]

      # apply the imposters
      imposters_results <- imposters(reference.set = data[-c(1,2),], test = data[1,], candidate.set = data[2,], distance = methods_combination$distance[method])

      # save incrementally
      full_results <- rbind(full_results, data.frame(configuration = methods_combination$configuration[method], repetition = ((candidate_id-1)*validation_rounds)+validation, analysis_type = "green", value = imposters_results))

    }

  }

  # cat("\n######\n######\n Wrong author\n\n")

  for(candidate_id in unique(full_corpus_metadata$candidate_id)){

    # cat("\n######\n no.", candidate_id, "\n\n")

    available_corpora <- which(full_corpus_metadata$analysis_type == "convalidations" &
                                 full_corpus_metadata$impostors == methods_combination$n_imposters[method] &
                                 full_corpus_metadata$candidate_id == candidate_id)

    for(validation in 1:validation_rounds){

      # select random text
      selected_id <- sample(available_corpora, 1)

      # load preprocessed dataset
      data <- full_corpus_list[[selected_id]]

      # culling
      data <- perform.culling(data, culling.level = methods_combination$culling[method])
      if(dim(data)[2] > methods_combination$MFU[method])
        data <- data[,1:methods_combination$MFU[method]]

      # apply the imposters
      imposters_results <- imposters(reference.set = data[-c(1,2),], test = data[1,], candidate.set = data[2,], distance = methods_combination$distance[method])

      # save incrementally
      full_results <- rbind(full_results, data.frame(configuration = methods_combination$configuration[method], repetition = ((candidate_id-1)*validation_rounds)+validation, analysis_type = "red", value = imposters_results))

    }

  }

  # cat("\n######\n######\n Actual analysis\n\n")

  available_corpus <- which(full_corpus_metadata$analysis_type == "actual" &
                                 full_corpus_metadata$impostors == methods_combination$n_imposters[method])

  for(validation in 1:validation_rounds){

    # load preprocessed dataset
    data <- full_corpus_list[[available_corpus]]

    # culling
    data <- perform.culling(data, culling.level = methods_combination$culling[method])
    if(dim(data)[2] > methods_combination$MFU[method])
      data <- data[,1:methods_combination$MFU[method]]

    # apply the imposters
    imposters_results <- imposters(reference.set = data[-c(1,2),], test = data[1,], candidate.set = data[2,], distance = methods_combination$distance[method])

    # save incrementally
    full_results <- rbind(full_results, data.frame(configuration = methods_combination$configuration[method], repetition = validation, analysis_type = "actual", value = imposters_results))

  }

  # Save results incrementally

  write.csv(full_results, file = filename, row.names = F)

}