Requirements

• Python 3.6+

For this project, we will be leveraging poetry for our dependency manager. To get started, create a virtual environment and install poetry:

pip install poetry==1.1.7

Next, run the following command to install all dependencies from pyproject.toml:

poetry install

Note: A quick step to configure a virtual environment and handling Python versioning within this project is to simply run make penv. For more details on using pyenv, check out the docs here. Assuming pyenv is now configured, you can then install the required dependencies into the activated virtual environment by running:

make build-env

---

Data

The dataset we will be working with can be accessed from the data directory within the project root. Here’s a breakdown of the available data:

user_interests:

Keys	Description
user_handle	a unique identifier for each user
interest_tag	the he date the user viewed the course
date_followed	the date user followed/expressed interest in the tag field

user_assessment_scores:

Keys	Description
user_handle	a unique identifier for each user
user_assessment_date	the date the user completed this assessment
assessment_tag	the assessment tag
user_assessment_score	the user’s score

user_course_views:

Keys	Description
user_handle	a unique identifier for each user
view_date	the he date the user viewed the course
course_name	name of course
author_handle	a unique identifier for each course’s author
level	difficulty level assigned to this course
course_view_time_seconds	number of seconds this user spent watching per day

course_tags:

Keys	Description
course_id	course identifier for joining to user_course_views
course_tags	author-applied tags to define what topics this course covers

---

RESTful API

For this application, we will leverage Flask-RESTful, which is an extension for Flask that adds support for quickly building REST APIs in Python. Before running the Flask server, the following environment variables for configuring the database environment and results table will need to be set:

export DATABASE_ENV=dev
export RESULTS_TABLE=user_ranking

Once the above variables are set, we can then start Flask server by running the following command:

python -m server.run

Note: the default settings is to host the Flask server on localhost and listening on port 5000. To change this setting, you can modify runserver.py by setting the host and port arguments in the app.run method, respectively.

To make a API call to the hosted server and return a JSON object containing the top 5 most similar users to the user of interest, the following CURL command can be run:

curl -X GET -H "Content-type: application/json"
-d "{\"user_handle\":\"110\"}" \
"http://0.0.0.0:5000/api/similarity/"

Note: the user id for this example is 110. Simply change the value of the user_handle key the above body field -d with a user id of interest. Keep in mind that the range of user ids is between 1 and 10000.

Alternatively to using curl for making an API call, a Python utility function can also be used to facilitate returning similar users to the provided input user id. The following command can be executed to return the same JSON object:

python -m server.api --user_id 110

Output JSON for Matrix Factorization Approach

[
   {
      "similar":"5537",
      "score":0.8957
   },
   {
      "similar":"7771",
      "score":0.8567
   },
   {
      "similar":"5843",
      "score":0.8519
   },
   {
      "similar":"239",
      "score":0.8512
   },
   {
      "similar":"9530",
      "score":0.8363
   },
   {
      "similar":"687",
      "score":0.7750
   },
   {
      "similar":"854",
      "score":0.7691
   },
   {
      "similar":"1199",
      "score":0.7563
   },
   {
      "similar":"9046",
      "score":0.7409
   },
   {
      "similar":"7341",
      "score":0.7401
   }
]