I recently just finished a paper with my partner in my data mining course. We were assigned the topic of Naive Bayes classifier and we decided to apply them to sets of Amazon user reviews in a bunch of categories. I've uploaded the paper on this blog in case you are interested in reading the details of the data mining.
In summary, the main results and implications of our research were:
- It is possible to accurately classify consumer sentiments through analyzing and identifying key words in the review.
- We can get more accurate predictions and meaningful data by examining reviews of one single product than a category of similar products.
- Get a large data set to train the naive-bayes classifier
- Predictions can take a long time.
- Stemming helps reduce the feature matrix and make better predictions
- Firms can leverage this to help predict success of certain products
- Firms can use text mining as a way to identify and address issues within certain products
- Firms can find relevant attributes in sub-categories of products and strategically position theirs.
- Are we able to model word-of-mouth and recommendation marketing in its effects on earning surprises?
More specifically, I wonder if we are able to take these results based on positive negative reviews and see if it can affect subsequent earning surprises for certain large firms with enough reviews. Note that this doesn't have to necessarily deal with text mining (although it can), but rather examining positive/negative reviews and subsequent earnings announcements. Large-cap firms usually have high analyst coverage and earning predictions are often posted up well before announcements. The surprise is calculated as the difference between the expected and actual earnings.
One scenario I can think of is that, given the new Apple Watch coming out, are we able to use our classifiers trained on Watch reviews and see if we can pull hidden information from opinion articles? Moreover, can we use initial reviews on Amazon, find the overall sentiment, and see if the Apple Watch will help Apple smash earnings in the next quarter? 2 quarters?
I'm skeptical about this hypothesis but I feel that it is certainly something worth examining...