CS1026: Assignment 3 – Sentiment Analysis
Weight: 11%
Learning Outcome:
By completing this assignment, you will gain skills relating to
• using functions,
• complex data structures,
• nested loops,
• text processing,
• file input and output,
• exceptions in Python,
• writing code that is used by other programs.
Background:
With the emergence of Internet companies such as Google, Facebook, and Twitter, more and more data accessible online is comprised of text. Text ual data and the computational means of processing it and extracting information is also increasingly more important in areas such as business, humanities, social sciences, etc. In this assignment, you will deal with textual analysis.
Twitter has become very popular, with many people “tweeting” aspects of their daily lives. This “flow of tweets” has recently become a way to study or guess how people feel about various aspects of the world or their own life. For example, analysis of tweets has been used to try to determine how certain geographical regions may be voting – this is done by analyzing the content, the words, and phrases, in tweets. Similarly, analysis of keywords or phrases in tweets can be used to determine how popular or unpopular a movie might be. This is often referred to as sentiment analysis.
Task:
In this assignment, you will write a Python module, called sentiment_analysis.py (this is the name of the file that you should use) and a main program, main.py, that uses the module to analyze Twitter information. In this module, you will create a function that will perform simple sentiment analysis on Twitter data. The Twitter data contains comments from individuals about how they feel about their lives and comes from individuals across the continental United States. The objective is to determine which timezone (Eastern, Central, Mountain, Pacific; see below for more information on how to do this) is
the “happiest”. To do this, your program will need to:
• Analyze each individual tweet to determine a score – a “happiness score”.
• The “happiness score” for a single tweet is found by looking for certain keywords (which are given) in a tweet and for each keyword found in that tweet totaling their “sentiment values”. In this assignment, each value is an integer from 1 to 10.
The happiness score for the tweet is simply the sum of the “sentiment values” for
keywords found in the tweet divided by the number of keywords found in the tweet.
If there are none of the given keywords in a tweet, it is just ignored, i.e., you do NOT
count it.
To determine the words in a tweet, you should do the following:
o Separate a tweet into words based on white space. A “word” is any sequence of
characters surrounded by white space (blank, tab, end of line, etc.).
o You should remove any punctuation from the beginning or end of the word. So,
“#lonely” would become “lonely” and “happy!!” would become “happy”.
o You should convert the “word” into just lower case letters. This gives you a “word”
from the tweet.
o If you match the “word” to any of the sentiment keywords (see below), you add the
score of that sentiment keyword to a total for the tweet; you can just do exact
matches.
• A “counted tweet” is a tweet in which there was at least one matched keyword.
• The “happiness score” for a timezone is just the total of the scores for all the counted
tweets in that region divided by the number of counted tweets in that region; again, if a
tweet has NO keywords, then it is NOT counted as a tweet in that timezone.
A file called tweets.txt contains the tweets and a file called keywords.txt contains keywords and scores for determining the “sentiment” of an individual tweet. These files are described in more detail below.
File tweets.txt
The file tweets.txt contains the tweets; one per line (some lines are quite long). The format of a tweet is:
[lat, long] value date time text
where:
• [lat, long] – the latitude and longitude of where the tweet originated. You will need
these values to determine the time zone in which the tweet originated.
• value – not used; this can be skipped.
• date – the date of the tweet; not used, this can be skipped.
• time – the time of day that the tweet was sent; not used this can be skipped.
• text – the text in the tweet.
File keywords.txt
The file keywords.txt contains sentiment keywords and their “happiness scores”; one per line.
The format of a line is:
keyword, value where:
• keyword – the keyword to look for.
• value – the value of the keyword; values are limited to 1, 5, 7 and 10, where 1
represents very “unhappy” and 10 represents “very happy”.
(you are free to explore different sets of keywords and values at your leisure for the sheer fun of it!).
Determining timezones across the continental United States
Given a latitude and longitude, the task of determining exactly the location that it corresponds to can be very challenging given the geographical boundaries of the United States. For this assignment, we simply approximate the regions corresponding to the timezones by rectangular areas defined by latitude and longitude points. Our approximation looks like:
p9 p7 p5 p3 p1
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Click HERE to order a unique plagiarism free paper done by professional writers and delivered before your deadline
Recent Comments