getting sentimental about ‘sentiment’ analysis

Wanted to read War and Peace, since always… finally starting to do it in R:

Let’s feed first few pages of the text to R:

myTolstoy <- readLines("https://github.com/ghose-gh/text-r/edit/main/War_n_Peace.txt")
TolstoyLines<- get_sentences(myTolstoy)

Runtime show:

231 lines…

Good time to check the sentiment for each…

sentiment <- get_sentiment(TolstoyLines)

And… so much to see, so much to feel…

Now some quick numbers…

> sum(sentiment)
[1] 127.2
> mean(sentiment)
[1] 0.5506494

Being Tolstoy was not easy… from -3 to 5.5!

> summary(sentiment)
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
-2.9500  0.0000  0.5000  0.5506  1.2500  5.5000

And some finale graphs…

plot(
    sentiment, 
    type="l", 
    main="Example Plot Trajectory", 
    xlab = "Narrative Time", 
    ylab= "Emotional Valence"
)

let’s get the percentage values…

ssn<- get_percentage_values(sentiment)

making it a bit more readable:

Now to plot the percentage values:

plot(
    ssn, 
    type="l", 
    main="O Tolstoy!", 
    xlab = "Narrative Time", 
    ylab= "Emotional Valence", 
    col="red"
)

which give us…

Now before I go, let’s try transformed values instead of percentage, to get a more grained picture of the ups and downs:

transformed <- get_transformed_values(
    sentiment, 
    low_pass_size = 3, 
    x_reverse_len = 100,
    scale_vals = TRUE,
    scale_range = FALSE
)

And the plot:

plot(
    transformed, 
    type="l", 
    main="O Tolstoy!", 
    xlab = "Narrative Time", 
    ylab= "Emotional Valence", 
    col="red"
)

That was a good day R in the Syuzhet library!