Cognitive Enterprise

Watson the word lover

Share this post:

Key points:

    • Watson read the author’s favorite passage from “The Great Gatsby” by F. Scott Fitzgerald
    • Watson identified Daisy, Tom, and Gatsby as the three most relevant entities
    • Watson interpreted the sentiment and emotion associated with each of these entities

Watch a video on our NLU service

I love books. Or, more precisely, I love words and stories. I often find myself poring over specific passages and savoring them, searching for nuance and pattern. In many ways, Watson is a word lover too, finding connections and insights within unstructured data. Usually, Watson reads medical journals or legal regulations, but I have often wondered what Watson might find in some of my favorite novels. So, I decided to see how Watson would ‘read’ a scene from The Great Gatsby by F. Scott Fitzgerald.

I put a passage into Natural Language Understanding to see what Watson would uncover. This service identifies entities, keywords, and concepts, and interprets emotion and sentiment. I analyzed the results in Excel, focusing on patterns that resonated with my reading of the text. Now, I thought I knew this book: as one of my all-time favorite novels, I’ve read it three times and I studied it in college. Nevertheless, Watson’s findings still brought new ideas to my attention while also confirming my own interpretation of the novel.

Before we dig into the results, however, first we need a recap of the novel. The Great Gatsby tells the story of Jay Gatsby’s love of Daisy Buchanan. He fell for her years ago, and at the time she seemed to have reciprocated his affections, but since then she has married another man, Tom Buchanan. Gatsby desperately wants to win back Daisy, and to do so he has reinvented himself by amassing immense wealth (through some not entirely moral means). He throws lavish parties in the hopes that Daisy will attend one and fall back in love with him.

That plan never works out, but eventually Gatsby’s neighbor Nick Carraway, the narrator of the novel, arranges a reintroduction. From there, Gatsby tries to win back Daisy and rediscover his love for her so they can relive and rewrite their past. Notice all of those returns and re-dos? That’s a central theme of this book – that Gatsby wants to get back to where he started, but he never does.

The passage I chose for analysis focuses on this very theme of return and repetition. In it, Gatsby asserts that he can “repeat the past.” The passage occurs at the end of Chapter 6, and it starts with Nick Carraway and Gatsby talking after another wild soirrée. Then, the scene slips into a memory of Gatsby and Daisy’s first kiss. This passage always sends a shiver down my spine, and although Watson doesn’t really shiver, Natural Language Understanding identified intriguing patterns in the text. I’ve included some of my favorite findings in graphs throughout this post.

Figure 1: This graph shows the emotion scores of the top three entities in the passage. Each bar represents a different emotion. Scores over 0.5 are especially significant.

Watson identifies Daisy, Tom, and Gatsby as the three most relevant entities, which makes sense. They are the three main characters, after all. Watson also interprets the sentiment and emotion associated with each of these entities. Now, remember that Nick Carraway narrates The Great Gatsby, and so he communicates the emotion and sentiment of the language.

Watson finds a negative sentiment associated with Tom and a neutral sentiment associated with both Daisy and Gatsby. My interpretation of Nick’s sentiment matches Watson’s view. While Nick provides complex descriptions and multi-faceted characterizations of Daisy and Gatsby, he really hates Tom. Throughout the novel, Nick calls him a “brute” and declares that his life peaked during his college football career. Nick definitely expresses a negative sentiment towards Tom, while his judgement of Daisy and Gatsby remains mixed.

Watson also scores the emotion associated with these characters. Watson detects sadness with a score of .57 for Tom, which reflects Nick’s aforementioned pity for Daisy’s husband. By contrast, Watson detects fear as the primary emotion associated with Gatsby, assigning a .51 score to the emotion. Indeed, Nick worries for Gatsby, reminding him, “You can’t repeat the past.” But, Nick knows that Gatsby wants “to recover something, some idea of himself perhaps, that had gone into loving Daisy” by returning to a previous “certain starting place.” Watson detects Nick’s expression of fear for Gatsby.

With Daisy, on the other hand, Watson’s emotional read is more divided, with no emotion earning a particularly high score. This distribution might reflect Nick’s conflicted feelings about Daisy. She is wonderful and pitiable yet also unlikeable. Perfect example: Daisy is absolutely charming and beautiful, and her most striking attribute is her melodious voice. But, when Nick asks Gatsby what makes Daisy’s voice so beautiful, Gatsby responds that her voice is “full of money.” Nick agrees with him, and this conflation is rather off-putting. Her beauty is intrinsically linked with immense wealth or even greed.

Perhaps Watson’s mix of emotional scores reflect Nick’s conflicting feelings towards Daisy, but, on the other hand, a longer section of the book might yield a more significant output.

Figure 2 : This graph shows the relevance, emotion, and sentiment scores of several of the keywords in the passage. I couldn’t include all of the keywords identified by Watson due to the sheer number of them, so I focused on those with high relevancy scores as well as some that I found especially interesting or resonant with my reading of the novel. I also included those that I discussed in the post.

Natural Language Understanding also extracts keywords, and this portion of the output brought some new insights to my attention. For example, Watson identifies “inevitable swimming party” as the most relevant phrase. Now, this finding might have something to do with the location of the phrase in the passage – it occurs within the first sentence. However, this phrase also eerily foreshadows the end of the novel when Gatsby drowns himself in his swimming pool.

I have never noticed this foreshadowing! Watson’s keyword extraction, however, highlights it immediately. Plus, not only does the phrase reference the tragic end of the book, but it also adds new significance to it; Gatsby’s death was “inevitable.” He was doomed from the start. I never would have noticed that without Watson’s help.

Watson also identifies one of my favorite phrases from the novel “perishable breath” as highly relevant, with a score of 0.71. This phrase occurs within the description of Daisy and Gatsby’s first kiss: Gatsby knows that he has “forever wed his unutterable visions to her perishable breath” and that “his mind [will] never romp again like the mind of God.” During their embrace, Daisy “blossom[s] for him like a flower and the incarnation [is] complete.”

Now, even though kisses are generally joyful occasions, this one feels a little different. Gatsby loses something as a result of the smooch, and Daisy transforms, incarnates, into his “visions” and ambitions. Ultimately, this loss and transformation won’t work out for Daisy or for Gatsby. It will prove futile and fatal, and Watson detects the sadness in this phrase “perishable breath” and reads its sentiment as negative, which fits with the outcome of the novel. By extracting “perishable breath” as a highly relevant phrase, Watson also draws my attention to its connotation: mortality.

Figure 3 : Instead of looking at specific parts of the passage, this chart displays the emotion scores for the passage as a whole. Again, values over 0.5 are especially significant.

As a whole, Watson calculates the passage’s sentiment as negative, and rates the document emotion rather morosely as well, identifying primarily sadness, fear, and disgust. This is a beautiful passage complete with parties, beaches, and “the incomparable milk of wonder” and yet Watson’s results seem, well, tragic. Watson’s findings match my own experience of the novel; it is about dreams and hopes and love, but in the end none of these aspirations come to fruition. Instead, the novel ends with death, destruction, and heartbreak.

Not only did Watson confirm my reading of this passage, but Natural Language Understanding also gave me an additional perspective on that reading, drawing my attention to new connotations and significance. To me, fiction is fundamentally a human endeavor. It is written and read with creativity and imagination. But patterns and connections calculated by Watson can deepen and broaden that experience, uncovering a new layer of meaning.

Learn more about Watson’s Natural Language Understanding service


Add Comment

Leave a Reply

Your email address will not be published.Required fields are marked *

Fernando R Abreu

Great analysis!


Sashi. G

This is awesome. So much of potential for the future. Can be used in a number of ways in the education world!

More Cognitive Enterprise Stories
July 26, 2017

How cognitive computing will revolutionize the retail industry

Cognitive computing is already redefining the retail industry worldwide. 91% of retail executives familiar with cognitive computing believe it will play a disruptive role in their organization. Opportunities for cognitive insights in the retail industry will continue to grow in the near future. Learn more.

Continue reading

June 15, 2017

The code of ethics for AI and chatbots that every brand should follow

While chatbots continue to grow in popularity, businesses often overlook important issues related to morals and ethics of chatbots and AI. Customers need to know when they are communicating with a machine, and that brands will protect their privacy and data in today’s interconnected world.

Continue reading

May 17, 2017

Watson Knowledge Studio: Deploy custom models and teach Watson at a lower cost

Watson Knowledge Studio is a cloud-based application that allows developers and industry experts to create custom annotator components. Collaborate and create custom annotator components and identify mentions and relations in unstructured text and analyze text specific to an industry. The price to deploy a custom model has been reduced by nearly 80%, from $3500 monthly to $800.

Continue reading