Note: the instructions are tested on a PC computer with Windows 10 and Microsoft Office 365.
- In Netlytic (under My Datasets), download one of your Twitter datasets from Netlytic to your computer as a CSV file:
- Open the downloaded dataset in Excel and delete all columns except the “description” column.
- In Excel, remove duplicates (retweets or tweets that have been posted multiple times by the same user) to avoid double counting of the same messages. In Excel, you can remove duplicates using its built-in function under “Data-> Remove Duplicates”
- Save the Excel file as a “Unicode TXT” file with a new name; let’s call it “tweets.txt”
- Download and install SentiStrength from http://sentistrength.wlv.ac.uk/#Download
- In SentiStrength, uncheck the following two reporting options: “Classification Rationale” and “Translation” as shown below:
- In SentiStrength, run a sentiment analysis with the default options under “Sentiment Strength Analysis” -> “Analyse ALL Texts in File (each line separately)”.
Note: Before the analysis starts, SentiStrength will ask you if you wish to add a header row in the resulting file; select “Yes”:
Note: SentiStrength will also ask you which column to use for the analysis; enter “1”:
- Once the analysis is complete, open the resulting text file in Excel. If Excel opens the Text Import wizard, specify that you are working a tab-delimited file that has a header, as shown below:
- If everything worked as intended, you will see a table with three columns: the original tweet, the positive polarity score (from 1 to 5) and the negative polarity score (from -1 to -5). Please note that tweets with scores (1 and -1) are considered to be neutral.
- For additional information on how SentiStrength works and how to interpret results, see the About section of the SentiStrength website at http://sentistrength.wlv.ac.uk/#About