SentiStrength is a software
program that analyzes sentiment. SentiStrength gives sentiment results in three
types. First, it gives us the results in binary form. In other words, it gives
the sentiment results as positive or negative. Second, it gives it as a
trinary. That is, the results are presented as positive, negative or neutral. Finally,
it gives sentiment results as a single scale. In other words, sentiment results
are given on a scale between -5 and +5. SentiStrength supports many languages
such as German, English, Finnish and Turkish and is capable of sentiment
analysis of texts written in these languages. The SentiStrength program is
completely free.
10108 tweets about online food
ordering services are analyzed with the SentiStrength program to explain this
software in depth. The main problem we encounter at this stage is that the
SentiStrength program only accepts documents with the ".txt"
extension. If the text you want to analyze is ".xlsx" extended, you
should deal with this issue first. Unfortunately, the file I would to analyse
has the ".xlsx" extension. For this reason, I will first convert the
section I want to examine in the excel file to .txt format. In this context,
delete the sections except the section we want to analyze, and only the column
we want to analyze remains in our file. In the next step, we need to remove the
duplicates to avoid recounting the same tweets. In Excel, we can remove
duplicates using the “Data-> Remove Duplicates” function. The logo you see
marked is the logo that belongs to the "remove duplicates" function.
Then save our excel file, which
we cleaned from repetitions, as unicode text (.txt). In the next step, open the
SentiStrength program and select the "Analyze ALL Texts in File (each line
separately)" function in the "Sentiment Strength Analysis"
section.
Next, let's select the .txt file
we want to analyze. Before the analysis starts, SentiStrength will ask if you
want to add a header line to the resulting file, and we need to select
"Yes". SentiStrength will ask us which column to use for analysis and
we can directly write "1" here.
If this image occurs, it means
that our analysis is complete. However, at this stage, the problem we
encountered at the beginning reoccurs. The results are saved on our computer
with the .txt extension.
It will be useful to copy and
paste the results with .txt extension to the excel file so that we can see the
results more clearly.
At the last stage, we will have
an excel file with 5 columns such as positive, negative and EmotionRationale. When
we look at the results, it is seen that there are many -1 and +1. -1 means not
negative or neutral. +1 means not positive or neutral. Between 2 and 4 is
positive and 5 is extremly positive. On the other hand, between -2 and -4 is
negative and -5 is extremely negative.
Finally, we can add another
column named "score" next to the negative and positive columns in the
excel file and see the general sentiment result. For this process, we need to
use the following formula; "=C2+D2". The results we get with this formula will
give the overall result of each tweet we analyzed.
