top of page

CSE 5334 Programming Assignment 1 (P1) Solution Python

Updated: Aug 16, 2023

In this assignment, you will implement a toy "search engine" in Python. You code will read a corpusand produce TF-IDF vectors for documents in the corpus. Then, given a query string, you code willreturn the query answer--the document with the highest cosine similarity score for the query.

The instructions on this assignment are written in an .ipynb fi le. You can use the following commands to install the Jupyter notebook viewer. You can use the following commands to installthe Jupyter notebook viewer. "pip" is a command for installing Python packages. You are required touse Python 3.5.1 or more recent versions of Python in this project.

pip install jupyter

pip install notebook (You might have to use "sudo"

if you are installing them at system level)

To run the Jupyter notebook viewer, use the following command:

jupyter notebook P1.ipynb

The above command will start a webservice at


and display the instructions inthe '.ipynb' fi le.


This assignment must be done individually. You must implement the whole assignment byyourself. Academic dishonety will have serious consequences.

You can discuss topics related to the assignment with your fellow students. But you are notallowed to discuss/share your solution and code.


We use a corpus of 15 Inaugural addresses of different US presidents. We processed the corpusand provided you a .zip fi le, which includes 15 .txt fi les.

Programming Language

1. You are required to use Python 3.5.1 or more recent versions of Python. You are required tosubmit a single .py fi le of your code.

2. You are expected to use several modules in NLTK--a natural language processing toolkit forPython. NLTK doesn't come with Python by default. You need to install it and "import" it in fi le. NLTK's website ( provides a lot of useful information,including a book , as well as installation instructions(

3.In programming assignment 1, other than NLTK, you are not allowed to use any other non-standard Python package. However, you are free to use anything from the the PythonStandard Library that comes with Python (


You code should accomplish the following tasks:

(1) Read the 15 .txt fi les, each of which has the transcript of a presidential debate. The followingcode does it. Make sure to replace "corpusroot" by your directory where the fi les are stored. In theexample below, "corpusroot" is a sub-folder named "US_Inaugural_Addresses" in the foldercontaining the python fi le of the code.

In this assignment we ignore the difference between lower and upper cases. So convert the text tolower case before you do anything else with the text. For a query, also convert it to lower casebefore you answer the query.

import os corpusroot = './US_Inaugural_Addresses'

for filename in os.listdir(corpusroot):

if filename.startswith( '0') or filename.startswith( '1'):

file = open (os.path.join(corpusroot, filename), "r" , encoding= 'windows-1252')



doc = doc.lower()

(2) Tokenize the content of each fi le. For this, you need a tokenizer. For example, the following pieceof code uses a regular expression tokenizer to return all course numbers in a string. Play with it andedit it. You can change the regular expression and the string to observe different output results.

For tokenizing the Presidential debate speeches, let's all use RegexpTokenizer(r'[a-zA-Z]+'). Whattokens will it produce? What limitations does it have?

from nltk.tokenize

import RegexpTokenizer

tokenizer = RegexpTokenizer(r '[A-Z]{2,3}[1-9][0-9]{3,3}')

tokens = tokenizer.tokenize("CSE4334 and CSE5334 are taught together. IE3013 is an undergraduate")


['CSE4334', 'CSE5334', 'IE3013']

(3) Perform stopword removal

on the obtained tokens. NLTK already comes with a stopword list, asa corpus in the "NLTK Data" ( You need to install this corpus. Followthe instructions at

.You can also fi nd the instruction in this book: (Section 1.2 Getting Started with NLTK). Basically, use thefollowing statements in Python interpreter. A pop-up window will appear. Click "Corpora" andchoose "stopwords" from the list.

import nltk

showing info


After the stopword list is downloaded, you will fi nd a fi le "english" in foldernltk_data/corpora/stopwords, where folder nltk_data is the download directory in the step above.The fi le contains 179 stopwords. nltk.corpus.stopwords will give you this list of stopwords. Try thefollowing piece of code.

from nltk.corpus

import stopwords

print (stopwords.words( 'english'))

['i', 'me', 'my', 'myself', 'we', 'our', 'ours', 'ourselves', 'you', "you're", "you've",........]

(4) Also perform stemming on the obtained tokens. NLTK comes with a Porter stemmer. Try the following code and learn how to use the stemmer.

from nltk.stem.porter

import PorterStemmer

stemmer = PorterStemmer()

print(stemmer.stem( 'studying'))










(5) Using the tokens, we would like to compute the TF-IDF vector for each document. Given a querystring, we can also calculate the query vector and calcuate similarity.

In the class, we learned that we can use different weightings for queries and documents and thepossible choices are shown below:

from IPython.display import Image


The notation of a weighting scheme is as follows: ddd.qqq, where ddd denotes the combinationused for document vector and qqq denotes the combination used for query vector.

A very standard weighting scheme is: lnc.ltc; where the processing for document and query vectorsare as follows: Document: logarithmic tf, no idf, cosine normalization Query: logarithmic tf, idf,cosine normalization

Implement a vector space model using the lnc.ltc

weighting scheme and show the outputs for the following:

print("%.12f"% getidf('british'))

print("%.12f"% getidf('union'))

print("%.12f"% getidf('war'))

print("%.12f"% getidf('power'))

print("%.12f"% getidf('great'))


print("%.12f"% getweight('02_washington_1793.txt','arrive'))

print("%.12f"% getweight('07_madison_1813.txt','war'))

print("%.12f"% getweight('12_jackson_1833.txt','union'))

print("%.12f"% getweight('09_monroe_1821.txt','great'))

print("%.12f"% getweight('05_jefferson_1805.txt','public'))


print("(%s, %.12f)"% query("pleasing people"))

print("(%s, %.12f)"% query("british war"))

print("(%s, %.12f)"% query("false public"))

print("(%s, %.12f)"% query("people institutions"))

print("(%s, %.12f)"% query("violated willingly"))

What to Submit

Submit through Canvas your source code in a single .py fi le.

You can use any standard Pythonlibrary. The only non-standard library/package allowed for this assignment is NLTK. You .py fi lemust defi ne at least the following functions:

getidf(token): return the inverse document frequency of a token. If the token doesn't exist inthe corpus, return -1. The parameter 'token' is already stemmed.

getweight(fi lemae,token): return the TF-IDF weight of a token in the document named'fi lename'. If the token doesn't exist in the document, return 0. The parameter 'token' is alreadystemmed.

query(qstring): return a tuple in the form of (fi lename of the document, score), where thedocument is the query answer with respect to the lnc.ltc weighting scheme.


Your program will be evaluated using the following criteria:

Correctness (75 Points)

We will evaluate your code by calling the functions specifi ced above (getidf - 20 points; getweight -25 points; query - 30 points). So, make sure to use the same function names, parameternames/types/orders as specifi ed above. We will use the above test cases and other queries andtokens to test your program.

Efficiency (15 Points)

Don't be satisfi ed by exhaustive, straightforward implementation. Keep improving its effi ciency. Aneffi cient solution should be able to answer a query in a few seconds. Also, it should consider theboundary cases. Your program should behave correctly under special cases and even incorrectinput.

Clarity, organization, modularity, documentation (10 Points)

Follow good coding standards to make your program easy to understand by others and easy tomaintain/extend.

Solution Includes: Plagiarism and AI report with 100% Accuracy.

Solution will be manually developed on request only.

Please contact only serious students(Don't waste time please);

For Further Information just txt only at WhatsApp : +91-995 314 1035

68 views0 comments


bottom of page