COMP SCI 3306, COMP SCI 7306 Assignment 4: Frequent Itemsets, Advertising

and Recommendation Systems

Formative, Weight (15%), Learning objectives (1, 2, 3),

Abstraction (4), Design (4), Communication (4), Data (5), Programming (5)

Due date: 11 : 59 pm, 30 May, 2022

1 Overview

This assignment must be done individually. This means all the rules regarding

individual submission will apply and the submission must be solely your own

work. Therefore, we will not use the groups on MyUni. You will need to submit

on the assignment page as an individual.

2 Assignment

Exercise 1 Collaborative Filtering (Exercise 9.3.1) (30 points) (Postgraduate

Students Only (COMP SCI 7306))

a b c d e f g h

A 4 5 5 1 3 2

B 3 4 3 1 2 1

C 2 1 3 4 5 3

Table 1: Utility matrix for exercise 1.

Table 1 is a utility matrix, representing the ratings, on a 1–5 star scale, of eight

items, a through h, by three users A, B, and C. Compute the following from the

data of this matrix:

1. Treating the utility matrix as boolean, compute the Jaccard distance be￾tween each pair of users.

2. Repeat Part (1), but use the cosine distance this time.

1

COMP SCI 3306, COMP SCI 7306 Mining Big Data Semester 1, 2022

3. Treat ratings of 3, 4, and 5 as 1 and 1, 2, and blank as 0. Compute the

Jaccard distance between each pair of users.

4. Repeat Part (3), but use the cosine distance this time.

5. Normalize the matrix by subtracting from each non-blank entry the aver￾age value for its user.

6. Using the normalized matrix from Part (5), compute the cosine distance

between each pair of users.

Exercise 2 Frequent Itemsets (50 points)

For this exercise, you will need to first study the Section 6.4 (up to 6.4.3) of

the text-book, Mining Massive Datasets, Leskovec, Rajaraman, Ullman (third

edition, 2020), and then follow the instructions below:

1. Implement the simple, randomized algorithm given in Section 6.4.1 of the

text-book.

2. Implement the algorithm of Savasere, Omiecinski, and Navathe (SON al￾gorithm), as explained in Section 6.4.3.

3. Compare the two algorithms implemented above on ALL the following

datasets: T10I4D100K, T40I10D100K, chess, connect, mushroom, pumsb,

pumsb star; those datasets are available at:

http://fimi.ua.ac.be/data/

Report the observed outcomes and comparisons on individual datasets and

in overall.

4. Perform different experiments on the simple randomized algorithm, using

the following sample sizes: 1%, 2%, 5% and 10%. Compare your experi￾mental results. Additionally compare the results of the above experiments

with the results produced by the SON algorithm.

Your approach should be as efficient as possible in terms of runtime and

memory requirements.

Report on challenges that you might have come across during the imple￾mentation and running the experiments.

Exercise 3 Advertising (Exercise 8.4.1 and more) (10+5+5 points)

Part 1 Explain both the Greedy Algorithm (Section 8.2.2 of the textbook) and

Balance Algorithm (Section 8.4.4 of the textbook) and explain what Competi￾tive Ratio is.

2

COMP SCI 3306, COMP SCI 7306 Mining Big Data Semester 1, 2022

Part 2 Consider Example 8.7. Suppose that there are three advertisers A, B,

and C. There are three queries x, y, and z. Each advertiser has a budget of 2.

Advertiser A only bids on x, B bids on x and y, and C bids on x, y, and z. Note

that on the query sequence xxyyzz, the optimal offine algorithm would yield a

revenue of 6, since all queries can be assigned.

1. Show that the greedy algorithm will assign at least 4 of the 6 queries

xxyyzz.

2. Find another sequence of queries such that the greedy algorithm can assign

as few as half the queries that the optimal offline algorithm would assign

to that sequence.

3 General assignment submission guidelines

Your submissions will need to include the following, at minimum:

• a PDF file of your solutions for any theoretical exercises. The solutions

should contain a detailed description of how to obtain the results, not just

the final results.

• PDF or txt file with a brief descriptions of your implementations to un￾derstand your code.

• Files containing the output results of running your algorithms on the pro￾vided datasets.

• PDF or txt file of your computation times of the algorithms on the pro￾vided datasets.

• All source files, logs, and all the project files.

• a README.txt file containing instructions to run the code, student ID,

and email address.

• the submissions that do not follow the above guidlines may lose points

accordingly.

Please do not hesitate to reach out using the discussion forum, workshops,

or the contact details of the teaching assistants on the home page of MyUni,

should you have any questions or concerns.