Sentiment analysis is an important natural language processing method. The method analyzes texts in English or German to determine its polarity as positive, negative or neutral. For sentiment analysis in English, the method uses 4 popular sentiment analysis tools that are 1. SentiStrength, 2. VADER, 3. TextBlob, and 4. SentiWordNet. Among these, SentiStrength, VADER and TextBlob employ a combination of lexicon and rule-based approaches, having lexicons focused on specific words and phrases for sentiment analysis while the rule observe sentiments in sentence structures. While SentiWordNet relies on lexicons only for the word sense used the text. These tools are specifically optimized for social media based content i.e., shorter texts with emojis etc. For sentiment analysis in German, the German variants of 1. SentiStrength_de 2. GerVADER, and 3. TextBlobDE are used.
The method can be used to analyze sentiments of textual content such as social media posts and customer reviews.
The method reads input as text file. The sample data files for English and German are data/input_text_en and data/input_text_de.
| Text |
|---|
| I absolutely loved the performance — it was breathtaking from start to finish! |
| The movie was enjoyable, though a bit slow in parts. |
| The package arrived on Tuesday as expected. |
| I was slightly disappointed with the quality of the material. |
| That was the worst experience I’ve ever had — completely unacceptable! |
| Text | SentiStrength | VADER | TextBlob | SentiWordNet |
|---|---|---|---|---|
| I absolutely loved the performance — it was breathtaking from start to finish! | {'pos': 5, 'neg': -1, 'neu': 1} | {'neg': 0.0, 'neu': 0.595, 'pos': 0.405, 'compound': 0.8169} | {'polarity': 0.85, 'subjectivity': 0.9} | {'pos': 0.625, 'neg': 0.0} |
| The movie was enjoyable, though a bit slow in parts. | {'pos': 3, 'neg': -1, 'neu': 1} | {'neg': 0.0, 'neu': 0.756, 'pos': 0.244, 'compound': 0.4404} | {'polarity': 0.09999999999999998, 'subjectivity': 0.5} | {'pos': 0.5, 'neg': 0.375} |
| The package arrived on Tuesday as expected. | {'pos': 1, 'neg': -1, 'neu': 0} | {'neg': 0.0, 'neu': 1.0, 'pos': 0.0, 'compound': 0.0} | {'polarity': -0.1, 'subjectivity': 0.4} | {'pos': 0.0, 'neg': 0.0} |
| I was slightly disappointed with the quality of the material. | {'pos': 1, 'neg': -2, 'neu': -1} | {'neg': 0.238, 'neu': 0.762, 'pos': 0.0, 'compound': -0.4228} | {'polarity': -0.75, 'subjectivity': 0.75} | {'pos': 0.375, 'neg': 0.5} |
| That was the worst experience I ’ve ever had — completely unacceptable! | {'pos': 1, 'neg': -5, 'neu': -1} | {'neg': 0.411, 'neu': 0.589, 'pos': 0.0, 'compound': -0.8264} | {'polarity': -0.4375, 'subjectivity': 0.7} | {'pos': 0.75, 'neg': 1.5} |
The method runs on a small virtual machine provided by a cloud computing company (2 x86 CPU cores, 4 GB RAM, 40GB HDD).
Execute the following command with conda to setup the virtual environmnet.
conda env create -f environment.yml
Run the following command to perform sentiment analysis on English texts and German input texts in data/input_text_en.txt and data/input_text_de.txt files, respectively. The output is written into the files data/output_sentiments_en.txt and data/output_sentiments_de.txt
python main.py
For any queries, please contact taimoor.khan@gesis.org