Build a resume sentiment analyzer

To build a sentiment analyzer for resumes, we can break it down into a few logical steps. Here’s a simplified approach to creating a resume sentiment analyzer using Python and natural language processing (NLP) techniques.

1. Set Up Your Environment

First, you’ll need to install some Python libraries. These will help with text processing and sentiment analysis.

bash
CopyEdit
pip install nltk pandas scikit-learn textblob

2. Import Libraries

python
CopyEdit
import nltk
import pandas as pd
from textblob import TextBlob
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report, accuracy_score

3. Load Data and Preprocessing

For the purpose of this sentiment analysis, we will use a sample dataset containing resume descriptions. In practice, you would train on a dataset that contains labeled sentiments (positive, neutral, negative).

Sample Dataset (You can expand this with more samples)

Here is a sample format for the dataset:

python
CopyEdit
data = {
    'Resume Text': [
        'Dynamic professional with a proven track record in software development.',
        'Highly skilled in managing teams and ensuring efficient project deliveries.',
        'A passionate problem solver with a strong focus on coding practices.',
        'Experienced in project management with excellent communication skills.',
        'A hardworking individual with a lot of technical expertise in AI and machine learning.'
    ],
    'Sentiment': ['Positive', 'Positive', 'Neutral', 'Positive', 'Positive']
}

df = pd.DataFrame(data)

4. Preprocess the Text

Before analyzing the sentiment, you need to clean and prepare the text:

python
CopyEdit
# Convert to lowercase, remove punctuation, etc.
nltk.download('stopwords')
from nltk.corpus import stopwords
import re

stop_words = set(stopwords.words('english'))

def preprocess(text):
    # Remove special characters and digits
    text = re.sub(r'[^a-zA-Zs]', '', text)
    
    # Convert text to lowercase
    text = text.lower()
    
    # Remove stop words
    text = ' '.join([word for word in text.split() if word not in stop_words])
    
    return text

# Apply preprocessing
df['Cleaned Resume Text'] = df['Resume Text'].apply(preprocess)

5. Feature Extraction

We need to convert text into numerical features. A common method is Count Vectorization.

python
CopyEdit
vectorizer = CountVectorizer()
X = vectorizer.fit_transform(df['Cleaned Resume Text'])

6. Train a Model

We’ll use a simple Naive Bayes classifier to classify sentiments.

python
CopyEdit
# Encode labels (Positive -> 1, Negative -> 0)
df['Sentiment_Label'] = df['Sentiment'].map({'Positive': 1, 'Neutral': 0})

# Split data into training and testing
X_train, X_test, y_train, y_test = train_test_split(X, df['Sentiment_Label'], test_size=0.2, random_state=42)

# Train a Naive Bayes classifier
model = MultinomialNB()
model.fit(X_train, y_train)

# Predict on the test set
y_pred = model.predict(X_test)

# Evaluate the model
print(classification_report(y_test, y_pred))
print("Accuracy:", accuracy_score(y_test, y_pred))

7. Sentiment Analysis Function

After training the model, you can use it to analyze the sentiment of new resumes.

python
CopyEdit
def analyze_sentiment(resume_text):
    cleaned_text = preprocess(resume_text)
    vectorized_text = vectorizer.transform([cleaned_text])
    prediction = model.predict(vectorized_text)
    
    if prediction == 1:
        return 'Positive'
    else:
        return 'Negative'

# Example usage
resume = "Experienced software developer with expertise in building web applications."
sentiment = analyze_sentiment(resume)
print(f"Sentiment of the resume: {sentiment}")

8. Improving the Model

• Larger Dataset: To get more accurate results, you should train on a larger dataset with more varied resume data.

• Advanced NLP Models: You can also use more advanced models like BERT or GPT for better text understanding, but for simple sentiment analysis, Naive Bayes can work well.

• Fine-tuning: You can further improve the model by fine-tuning hyperparameters and using other feature extraction methods like TF-IDF.

Summary

This implementation gives you a basic sentiment analyzer that can determine whether a resume is positive, negative, or neutral based on its content. While this example uses a simple machine learning approach (Naive Bayes), you can enhance it with more complex NLP models and a larger dataset for better accuracy.