This example uses the td_sentiment_extractor_mle() sentiment extraction function from the tdplyr package to evaluate and classify a set of restaurant reviews.
The input data table "restaurant_reviews" table is listed here.
| id | review_text |
|---|---|
| 1 | This restaurant was great. The food was amazing. Our waiter was excellent. The appetizers in particular were very creative and well-thought-out. |
| 2 | I really enjoyed my meal, and my daughter's steak was perfectly prepared. The chocolate torte was superb. |
| 3 | The service was terrible! The food was ok, but the bread was stale and the drinks were very weak. |
| 4 | Not a must-do. Perfectly adequate but nothing special for the price. |
| 5 | I can't recommend this place. Service was slow and unfriendly. Food so-so. |
| 6 | Definitely a good choice for a special occasion. Highly recommended! |
- Create a tibble from the input data table "restaurant_reviews".
tddf_restaurant_reviews <- tbl(con, "restaurant_reviews")
This example uses the default values for many arguments of the td_sentiment_extractor_mle() function. One of those is the object argument which specifies the source used to assign sentiment values to words. The default value of object argument is a built-in dictionary based on the WordNet lexical database. - Call the sentiment extraction function.
td_sentiment_extractor_out <- td_sentiment_extractor_mle( object = "dictionary", newdata = tddf_restaurant_reviews, level = "document", text.column = "review_text", accumulate = c("id") ) - Inspect the result.
print(td_sentiment_extractor_out$result)