I used the Wildcard advanced search to see how many times the word "climate" is mentioned throughout the years as well as the most popular words that follow "climate." I typed "climate *" into the search box. As the graph shows, the use of "climate change" skyrocketed since 1985. The usage of "change" even exceeded the most common words such as "of," "and," "in." This really shows that the phrase "climate change" has become inseparable in modern writings, and the issue is discussed more and more frequently.
I combined features of wildcards and finding the dependencies of words together in the second experiment. By typing "intelligence=>*_ADJ" into the searchbox, I tried to find what adjective is most commonly used to modify intelligence throughout the years. Not surprisingly, the adjectives used for intelligence before 1960 shared similar frequencies. However, after the phrase "artificial intelligence" was coined, the adjective "artificial" suddenly dominated.
I picked "The Great Gatsby" by F. Scott Fitzgerald again for this Voyant system exploration. The word cloud shows that the most frequent words in the corpus are "said (235); gatsby (201); tom (177); daisy (150); like (122)." The document has a total of 51,892 words and 6,317 unique word forms. The average words per sentence are 15.2.
This trend graph tells the relative frequency of words as they appear in different segments of the book. It shows that most mentions of the characters' name appear in segment 7, which is about the climax of the novel. The most mentions of "cars" appear at the end of the novel because it is when the tragic car accident happens.
This chart shows that the context that the word "money" is embedded in the novel. By examining the sentences with contexts, we can really see that money is a central theme in The Great Gatsby. Money is the motivation for many characters and their actions. This chart shows the society's morbid fascination and yearning for money in the 1920s.
The display on the map is very interesting to me because it links all the locations in the story together. It becomes very clear that the major plot of the story happens around the Long Islands of New York. Occassionally, there are mentions to the west coast and parts of Europe because Europe is where some characters trace their lineage and history.
A word that could be either positive or negative is "empty." For example, when it is used to describe a subway station that used to be very busy, it is positive because less people can enjoy more room in the station. However, when it is used to describe a state of mind or a lacking in meaning, it carries a negative connotation. Another ambiguous word is "stubborn." It is used to describe someone who show strong determination not to change their attitude. Usually, it is used as a derogatory term to criticize people. However, sometimes, depending on contexts, it can be used to praise one's persistence and resolution.
In the source code of Sentimood, the word "sentence" is given a -2 weighting. Even though the word is negative when used as the punishment for a defendant, it is also used a lot of times as "a statement," which is neutral. I think Sentimood fails to account for the multiple meanings of the same word, so the weighting is seriously wrong. Another word that is weighted seriously wrong is "ominous." It is given +3 points. However, there is no doubt that it carries a negative connotation that hints at a threatening and inauspicious situation. Therefore, the weighting for this word is seriously wrong.
Two sentences that the two analyzers agree and both appear to be correct:
"Mr. Pontellier, unable to read his newspaper with any degree of comfort, arose with an expression and an exclamation of disgust." --The Awakening by Kate Chopin
(Sentimood gives this sentence -1 point, while the commercial interpreter assesses it as negative.)
"Alongside echoing enthusiasm for the transformative research award, Murphy expressed gratitude for also receiving the Pioneer Award in an email to the Prince." --Daily Princetonian
(Sentimood gives this sentence 6 points, while the commercial interpreter assesses it as very positive.)
Two sentences that are judged very differently by the two analyzers:
"There was an ominous silence." This is because Sentimood characterizes "ominous" as positive, while commercial analyzer does not.
Another sentence is: "This sentence is gramtically correct." While Sentimood gives -2 points because of the word "sentence," the commercial analyzer gives the correct response that it is positive.
Two sentences that the two analyzers agree on the sentiment, but both are wrong:
This quote from Jane Austen's "Pride and Prejudice" is meant to be an irony. In the context, Miss Bingley sarcastically praises Mrs. Bennet in order to attack Elizabeth Bennet's family. However, without understanding of the contexts, both Sentimood and the commercial analyzer fail to recognize the irony and rate the sentence as positive because of the word "charming."
Both analyzers interpret this sentence as negative probably because of words such as "risk" and "low." However, this news article actually tells a good news because a low risk status is a good thing.
Examples that are spectacularly wrong:
1. "Gatsby turned out all right at the end; it is what preyed on Gatsby, what foul dust floated in the wake of his dreams that temporarily closed out my interest in the abortive sorrows and short-winded elations of men." --The Great Gatsby
Google translates it back to "Gatsby did well in the end; it was it that plundered Gatsby, the filthy dust floating in his dreams, and temporarily shut off my sadness and short-lived interest in men's miscarriage." Bing translates it back to "Gatsby's lastthing was normal: it was Gatsby's food, the dust that floated after his dream, that temporarily ended my euphoric interest in the grief and short wind of men's miscarriages."
2. "Sweet are the uses of adversity which, like the toad, ugly and venomous, wears yet a precious jewel in his head." -Shakespeare
Google translates it as "Make good use of adversity, like a toad, ugly and vicious, and wear treasures on your head." Bing translates it as "Sweetness is the function of adversity, like clams, ugly and poisonous, with a precious gem on its head."
The translation did poorly on my chosen language, Mandarin. Even though the meaning is sometimes preserved when the language is translated back to English again, the intermediate step/sentence makes no pragmatic sense. Therefore, the translation would be very useless in practice.
I believe there are meaningful differences between the different translation services. In my opinion, Google focuses more on syntax because the intermediate Mandarin translation always sounds natural syntactically. Google's translation is also more concise. However, a disadvantage is that the meanings are not preserved very well. On the other hand, Bing translates every word in a sentence literally and focuses on the exact meanings. Therefore, it results in much longer sentences and does not always sound natural. However, the translation captures the original literal meanings quite well.
I used the camera in my computer to capture two versions of my portraits, one with mask on and one with a hand covering nose and mouth. I chose this because masking is relevant in our pandemic age. My question is that whether cameras can detect us covering our faces with hands and pretending to wear a mask. This experiment worked very well. The machine is able to pretty accurately identify whether I am wearing my mask. However, it seems to do so by checking if the color blue is in the image or not. When the mask is shown in the image but not worn, the machine still claims that I am wearing the mask. Only a few training examples (less than 5) are necessary to distinguish between class 1 and class 2, and it did improve with more training.
Second project
For this experiment, I placed two stuffed animal toys on top of each other and switched their positions. I did this because the two toys look somewhat similar, so I wanted to see if the machine is able to distinguish their differences. It worked well, and it also only requires a few training examples for the machine to make the correct decision. In addition, it requires more examples that shows different angles of the toys. It improved when more examples of the toys' different angles are used.