Week 7: Slight Change of Plans
Krish S -
As a recap from last week’s blog, I was going to be doing two things to enhance my sentiment analysis model: 1) adding gender differentiation to the sentiment analysis to identify potential biases in how male and female actors are described, and 2) implementing frequency analysis to account for how often each actor is mentioned in a review. My original plan was to take one week for gender differentiation and another week for frequency analysis. Turns out, I actually finished both, so the lock in went quite well.
I did however change my approach. I didn’t necessarily add anything to my sentiment analysis model. I instead made a new model that would produce gender bias scores, which range from -5 to +5. This model would cover both of the things I was talking about in the previous paragraph. So, by the end of this project, I will be analyzing sentiment scores and gender bias scores for the ‘Best Actor’ and ‘Best Actress’ reviews from 2005-2024. Although I have completed the coding for both models, I haven’t yet processed all 120 film reviews through them. I’ve only processed one film review through the sentiment model and the gender bias model to make sure it works. And it does! Now, let’s talk more about what these gender bias scores actually mean.
For gender bias analysis, a custom approach was implemented to assess the presence of gendered language within the reviews. This involved identifying gender-specific terms and pronouns associated with male and female descriptors. A gender bias score was calculated based on the frequency and context of these terms, with adjustments made according to the review category (e.g., Best Actor vs. Best Actress). The score indicates the extent to which a review aligns with or diverges from traditional gender stereotypes. My goal for this upcoming week is to have the sentiment scores and gender bias scores for all 120 reviews. I’ll be talking about that in the next post, so see you then!
Comments:
All viewpoints are welcome but profane, threatening, disrespectful, or harassing comments will not be tolerated and are subject to moderation up to, and including, full deletion.