Zack Meier
58 lines
1.8 KiB
PowerShell
58 lines
1.8 KiB
PowerShell
$transcript = Get-Content D:\Downloads\transcript.txt
|
|
|
|
$RawResult = [System.Collections.ArrayList]@()
|
|
|
|
$x = 0
|
|
|
|
While ($x -lt (@($transcript).count)) {
|
|
|
|
$obj = [pscustomobject]@{
|
|
timestamp = $transcript[$x + 0]
|
|
speaker = $transcript[$x + 1]
|
|
words = $transcript[$x + 2]
|
|
}
|
|
|
|
$null = $RawResult.Add($obj)
|
|
$x = $x + 3
|
|
}
|
|
|
|
$WordsResult = [System.Collections.ArrayList]@()
|
|
ForEach ($dialog in $RawResult) {
|
|
ForEach ($word in $dialog.words.split(' ')) {
|
|
$word = [regex]::Replace($word, "[^a-zA-Z0-9\s]", "")
|
|
$obj = [pscustomobject]@{
|
|
timestamp = $dialog.timestamp
|
|
speaker = $dialog.speaker
|
|
word = $word
|
|
}
|
|
$null = $WordsResult.Add($obj)
|
|
}
|
|
}
|
|
|
|
|
|
# Top 300 Words
|
|
$WordsResult | Group-Object Word | Sort-Object -Descending Count | select -First 300
|
|
|
|
# Top words by speaker
|
|
$WordsBySpeakerResult = [System.Collections.ArrayList]@()
|
|
$DialogBySpeaker = $WordsResult | Group-Object speaker
|
|
ForEach ($speaker in $DialogBySpeaker) {
|
|
$Top10SpeakerWords = $speaker.Group | Group-Object Word | Sort-Object -Descending Count | select -First 10
|
|
$obj = [pscustomobject]@{
|
|
speaker = $speaker.name
|
|
word1 = $Top10SpeakerWords[0].Name
|
|
word2 = $Top10SpeakerWords[1].Name
|
|
word3 = $Top10SpeakerWords[2].Name
|
|
word4 = $Top10SpeakerWords[3].Name
|
|
word5 = $Top10SpeakerWords[4].Name
|
|
word6 = $Top10SpeakerWords[5].Name
|
|
word7 = $Top10SpeakerWords[6].Name
|
|
word8 = $Top10SpeakerWords[7].Name
|
|
word9 = $Top10SpeakerWords[8].Name
|
|
word10 = $Top10SpeakerWords[9].Name
|
|
}
|
|
$null = $WordsBySpeakerResult.Add($obj)
|
|
}
|
|
|
|
$TopNouns = (Import-Csv "D:\Downloads\english-word-list-nouns.csv" -Delimiter ';').word
|
|
$FilteredWordsResult = $WordsResult |