如何从sentimentr包词典移除特定词汇，使其不影响情感得分

Hey there! If you want to remove certain words (like "appreciate" in your example) from sentimentr's dictionary so they don't impact your sentiment scores, here's a straightforward, reproducible way to do it:

Step 1: Load Packages and Grab the Default Dictionary #

First, we need to access the default sentiment dictionary that sentimentr uses under the hood. It pulls from the lexicon package, so we'll load both:

library(sentimentr)
library(lexicon)

# Fetch the default sentiment dictionary (jockers-rinker)
custom_dictionary <- lexicon::hash_sentiment_jockers_rinker

Step 2: Remove Your Target Words #

The dictionary is a named vector where each name is a word, and the value is its sentiment polarity (positive/negative). We can filter out the words we don't want using basic vector indexing:

# List of words to exclude (add more if needed)
words_to_remove <- c("appreciate")

# Remove the words from the dictionary
custom_dictionary <- custom_dictionary[!names(custom_dictionary) %in% words_to_remove]

Step 3: Calculate Sentiment with the Modified Dictionary #

Now pass this modified dictionary to the sentiment() function using the polarity_dt parameter. Let's test it with your example text:

# Your sample text
sample_text <- "i appreciate the help"

# Get sentiment scores using the adjusted dictionary
result <- sentiment(sample_text, polarity_dt = custom_dictionary)

# View the output
print(result)

What You'll See #

Before removing "appreciate", this text would have a positive sentiment score because "appreciate" is a positive term in the default dictionary. After removal, the remaining words ("the", "help") aren't in the filtered dictionary, so the sentiment score will be 0—exactly what you want!

Bonus: Reusable Function for Multiple Words #

If you need to exclude words frequently, wrap this logic into a function to save time:

# Function to create a filtered sentiment dictionary
create_filtered_dict <- function(exclude_words) {
  base_dict <- lexicon::hash_sentiment_jockers_rinker
  base_dict[!names(base_dict) %in% exclude_words]
}

# Example usage: exclude multiple words
my_filtered_dict <- create_filtered_dict(c("appreciate", "thank"))
sentiment("i appreciate your help, thank you!", polarity_dt = my_filtered_dict)

Note on Other Dictionaries #

If you're using a different sentiment dictionary (like the Loughran-McDonald financial lexicon), just replace lexicon::hash_sentiment_jockers_rinker with the dictionary you're using (e.g., lexicon::hash_sentiment_loughran_mcdonald).

内容的提问来源于stack exchange，提问作者Michael