ms6863, Miles Sugarman

PART 1








Here we have two graphs. The first one is a search on Bachelor degrees to see which one is the most used in text. I used the "Wildcard" advanced search function for this first graph, which displays the top ten substitutions that comes after the words inputted. As we can see, Bachelor of Arts is the most used as of 2019. But Bachelor of Science was the most used in last 200 years between 1911 to 2019.
This next graph displays two different ways the word "smoke" is used and how often they are used. I used "Part-of-speech Tags" to get the noun and verb versions of the word "smoke". We can see that the noun version of smoke is much more used than the verb version of smoke.

PART 2

In this next part, we move onto a webpage called Voyant Tools. It scans all the words for a certain document and reveals information about it, like how often certain words are used. I downloaded The Great Gatsby and got back this information.



This first image displays a cirrus cloud, which shows the most frequently used words. As we can see, the words said, tom, gatsby, and daisy are quite popular. This cloud only displays 150 of the most popular words. Even though words like "the" and "and" are most popular in most novels, this site displays the most popular words unique to that book.

Now let's look at this second image, which is quite similar to the first. It also displays common words used throughout the novel. But this time, it shows how often these words were used in different parts of the book. As we can see, the name "tom" is used much more often in the latter half of the book. This is very interesting to look at because now we can infer that Tom's character is not very relevant for a couple segments of the book.

PART 3

In Part 3 we go over Sentiment Analysis. We will be utilizing two different website tools. One being Sentimood, and the other being Meaning Cloud. Both are tools that analyze text and attempts to determine if the text is positive or negative. But these tools can make mistakes. First, we look at Sentimood. Let's look at the word "great". Many people would likely think of great as a positive word. But there are certain contexts where the word "great" can be meant in not such a positive way. For instance, in the sentence, "He brought me a great deal of pain" great describes how much pain was endured. But Sentimood sees the word on its own and deems it positive. It also says the word "no" is negative.In certain examples, that is true, it is typically known as a negative word. But there are many instances where the word can mean something positive. Like in this sentence: "There is no one in this world that I hate."

A word that I believe Sentimood gets wrong regarding its weighing is the word "silly". Sentimood deems the word to be negative. But I believe that there are many instances where the word silly can be used in a more positive way. Like when you describe a funny cat: "Your cat is so funny and cute, she's such a silly cat." I do not think many people would read that and believe the word silly to be negative. Another word that I think is wrongly weighted is the word "kind". While "kind" in certain contexts, like, "he is so kind" is very positive, you can also use the word in this context, "I kind of think he is funny". It still deems kind to be very positive. But as we know, kind doesn't really mean anything positive in this sentence.


Now, here are two sentences where both Sentimood and Meaning Cloud give similar scores and seem to be correct:

"Do not just seek happiness for yourself. Seek happiness for all. Through kindness. Through mercy." -- Both Sentimood and Meaning Cloud say that this passage is positive, which is correct.
"I hate being alone." -- Both Sentimood and Meaning Cloud say that this passage is negative, which is correct.

Here are two sentences where Sentimood and Meaning Cloud give different, contradicting scores:

"I guess that's what saying goodbye is always like, like jumping off an edge. The worst part is making the choice to do it. Once you're in the air, there's nothing you can do but let go" -- Sentimood says this an overall positive passage, while Meaning Cloud more correctly says this passage is mostly negative.
"You see I usually find myself among stranger because I drift here and there trying to forget the sad things that happened to me." -- Meaning Cloud says that this passage is positive while Sentimood says that it's negative.


And finally, here are two sentences where both Sentimood and Meaning Cloud give similar scores but seem to be clearly incorrect:

"When someone you love dies, and you're not expecting it, you don't lost her all at once; you lose her in pieces over a long time -- the way the mail stops coming, and her scent fades from the pillows and even from the clothes in her closet and drawers." -- Both Sentimood and Meaning Cloud says that this passage is overall positive, but we can see that this passage is quite sad.
"The one you love and the one who loves you are never, ever the same person." -- Both Sentimood and Meaning Cloud say that this passage is very positive. But as we can see, this is a quite sad passage.

PART 4

In this part, we will be looking over translation services. We will be experimenting with Google Translate and Bing Translate. All of our passages will be from the book The Great Gatsby.

It reads, "Then wear the gold hat, if that will move her; If you can bounce high, bounce for her too, Till she cry "Lover, gold-hatted, high-bouncing lover, I must have you!" -- This passage translated to Arabic and then back into English using Google Translate reads back,

"Then put on the golden hat, if that will move it; If you can jump high, so jump for her, so she cries "Baby, gold-hatted, high-jump darling, I gotta be with you!" -- As we can see, this is quite wrong. Let's see how Bing Translate does.

"Then wear a gold hat, if it will move her. If you can bounce high, bounce her too, so cry "Lover, hated gold, high bouncing lover, must be yours!" -- As we can see, both translation services are close but definitely not getting a perfect translation. Let's try another example.

Our initial passage is this: "It was lonely for a day or so until one morning some man, more recently arrived than I, stopped me on the road." -- After putting this into Google Translate and getting a translation back from Arabic, we get, "I was alone for a day or so until one morning a man, who had arrived more recently than me, stopped me on the road." -- This is much closer and not too bad of a translation.

Again, let's try a different passage and use Bing Translate. The initial passage is, "The rain cooled about half-past three to a damp mist, through which occasional thin drops swam like dew." -- After inputting this into Bing and getting an Arabic Translation, we translate it once more into English and get, "The rain cooled about half past three to a wet mist, through which occasional thin drops swam like dew."-- This translation was near perfect! There was only one word that changed after the translation. Pretty remarkable

Let's try one more example and use Bing again. The initial passage is, "A man in a long duster had dismounted from the wreck and now stood in the middle of the road, looking from the car to the tyre and from the tyre to the observers in a pleasant, puzzled way." After being translated back into English from Arabic, we get "A man in long dust took off from the wreckage and now stood in the middle of the road, looking from car to tyre and from tyre to onlookers in a pleasant, puzzled way." -- A lot of the words are different and the meaning is a little off.


Now the question is, wow well did these translation services do? And would they be useful in practice? My answer is yes. While in most cases, you do not get a perfect translation and certain words and phrase get miss used, the general idea of the sentence does not become lost. So I would use these services if I need to get an idea of what a phrase means in a different language, but not so much if I need a perfect translation. Also, I did not see a very meaningful difference between Google Translate and Bing Translate. Both had very similar results in my examples.

PART 5

For this final part of the lab, we will be using a very cool website called Teachable Machines provided by Google. This is by far one of the coolest websites we have utilized so far in this class. I did two different projects both using photos. Here are screenshots of the two projects.



Let me quickly walk through both of these projects. In the first one (photo on the left), I had the computer analyze over 100 photos of me wearing a mask, and over 100 photos of me not wearing a mask. I got the inspiration to do this because sometimes when I book an Uber, the app asks me to take a photo of my face to prove that I am wearing a mask. It had me thinking, how is the app able to tell? Now I know. I also thought this would be topical due to the pandemic we are currently living through, and how there are many ways where it could be useful to detect if someone is wearing a mask from a photo. The computer was able to distinguish between the two easily and with a high success rate.

For the second project (photo on the left), I wanted the computer to be able to detect if there were 2 people in the Frame or just 1. Big shoutout to my brother (Arno Sugarman, Princeton Class of '23) for volunteering to help with this! (Side-note, he never volunteered, we are both on a train right now and he is the only person close to me). I thought this would be an interesting one to test. It worked super well and the computer was very successful in its detections. I wanted to see if the computer would be able to tell how many people were in frame. I tested this a little where my brother and I were in a slightly different background but both still in frame, and it was still sometimes able to pick up that there were 2 people in frame. I also had over 100 photos of just me, and over 100 photos of both my brother and I. This was such a fun experiment to play around with. Very interesting!