K-nearest neighbor for text classification, Computer Engineering

Assignment Help:

Assignment 2: K-nearest neighbor for text classification.

The goal of text classification is to identify the topic for a piece of text (news article, web-blog, etc.). Text classification has obvious utility in the age of information overload, and it has become a popular turf for applying machine learning algorithms. In this project, you will have the opportunity to implement k-nearest neighbor and apply it to text classification on the well known Reuter news collection.

1.       Download the dataset from my website, which is created from the original collection and contains a training file, a test file, the topics, and the format for train/test.

2.       Implement the k-nearest neighbor algorithm for text classification. Your goal is to predict the topic for each news article in the test set. Try the following distance or similarity measures with their corresponding representations.

a.        Hamming distance: each document is represented as a boolean vector, where each bit represents whether the corresponding word appears in the document.

b.       Euclidean distance: each document is represented as a numeric vector, where each number represents how many times the corresponding word appears in the document (it could be zero).

c.         Cosine similarity with TF-IDF weights (a popular metric in information retrieval): each document is represented by a numeric vector as in (b). However, now each number is the TF-IDF weight for the corresponding word (as defined below). The similarity between two documents is the dot product of their corresponding vectors, divided by the product of their norms.

3.        Let w be a word, d be a document, and N(d,w) be the number of occurrences of w in d (i.e., the number in the vector in (b)). TF stands for term frequency, and TF(d,w)=N(d,w)/W(d), where W(d) is the total number of words in d. IDF stands for inverted document frequency, and IDF(d,w)=log(D/C(w)), where D is the total number of documents, and C(w) is the total number of documents that contains the word w; the base for the logarithm is irrelevant, you can use e or 2. The TF-IDF weight for w in d is TF(d,w)*IDF(d,w); this is the number you should put in the vector in (c). TF-IDF is a clever heuristic to take into account of the "information content" that each word conveys, so that frequent words like "the" is discounted and document-specific ones are amplified. You can find more details about it online or in standard IR text.

4.       You should try k = 1, k = 3 and k = 5 with each of the representations above. Notice that with a distance measure, the k-nearest neighborhoods are the ones with the smallest distance from the test point, whereas with a similarity measure, they are the ones with the highest similarity scores.

 

 


Related Discussions:- K-nearest neighbor for text classification

Change to palindrome, given a string S convert it to palindrome by doing ch...

given a string S convert it to palindrome by doing character replacement.convert S to palindrome with minimum character replacements

Introduction to computers, explain classification of computers in detail.al...

explain classification of computers in detail.also explain various application areas of computers

What is electronic data interchange, What is Electronic Data Interchange? ...

What is Electronic Data Interchange? Electronic Data Interchange (EDI): It is used by organizations for transactions which arise on regular basis to a pre-found format.

Explain passing parameters using pointers, Q. Explain Passing Parameters Us...

Q. Explain Passing Parameters Using Pointers ? This method overcomes the drawback of using variable names directly in procedure. It uses registers to pass procedure pointers to

Use string manipulating functions , Write a script that will first initiali...

Write a script that will first initialize a string variable that will kept x and y coordinates of a point in the form 'x 3.1 y 6.4'.  Then, use string manipulating functions to ext

Uniform memory access model (uma), Uniform Memory Access Model (UMA) In...

Uniform Memory Access Model (UMA) In this model, the main memory is uniformly shared by all processors in multiprocessor systems and each processor has equal access time to sha

Describe about the database marketing application of olap, Database marketi...

Database marketing tool or application helps a user or marketing professional in determining the right tool or plan for his valuable add campaign. This tool haves data from all sou

the entire pcd division, Will executing  SAP R/3 across the entire PCD div...

Will executing  SAP R/3 across the entire PCD division give the division with a competitive benefit?  Clarify  your answer carefully.

Explain the term- cycle based simulator, Explain the term- Cycle Based Simu...

Explain the term- Cycle Based Simulator This is a Digital Logic Simulation method which eliminates unnecessary calculations to achieve huge performance gains in verifying Bool

What is the session, What is the session.  Session is a collection of v...

What is the session.  Session is a collection of various groups of method. Every session is assigned to a single control terminal. This terminal is either a pseudo-device. or a

Write Your Message!

Captcha
Free Assignment Quote

Assured A++ Grade

Get guaranteed satisfaction & time on delivery in every assignment order you paid with us! We ensure premium quality solution document along with free turntin report!

All rights reserved! Copyrights ©2019-2020 ExpertsMind IT Educational Pvt Ltd