Anda di halaman 1dari 2

Centennial College

MAPD725
Emerging Technology

Assignment 1: Anomaly detector on Twitter Sentiments


In a group of 2 or 3, students will collect and analyze Twitter sentiments for an automobile
company on their brands. The PR team would like to be alerted whenever there is a tweet
with a strongly negative sentiment so they may take appropriate actions. The IT team would
also like to store the tweets onto a NoSQL database for future reference. A random company
will be assigned to each group. Any request for changes must be pre-approved by the
instructor.
1. Identify a list of the company’s make and model (e.g. Accord and Civic for Honda).
2. Create a Node-RED Starter boilerplate instance on Bluemix.
3. Build a Node-RED workflow to:
a. Ingest twitter feed on the company’s brands (e.g. #Accord , #Civic, #Honda).
b. If there is a tweet with sentiment value less than -3, send a notification email
with the tweet text and the sentiment value to the PR team. Note: If email is
sent from an gmail e-mail address, the Google account setting “Access for less
secure apps” needs to be marked on.
i. Email title should be: <Brand> bad tweets (e.g. Honda bad tweets).
ii. Send email to mapd725@gmail.com .
iii. Email content should include tweet text and sentiment value.
c. Store all tweets onto a NoSQL database (cloudant) for the IT team for future
references. Note: The Cloudant database name is “nodered”.
4. Highlight all the nodes in Node-RED workflow and export the configuration
(formatted).

Submission format & requirements:

● Demo in class:
o Walkthrough the Node-RED workflow
o Show tweets sent to email inbox
o Show tweets stored in NoSQL database
● Code:
o Submit exported workflow code to e.centennial

Grading Criteria Points


Ability to ingest tweets on company’s brand 5
Ability to send email alerts with the correct format 5
Ability to retrieve tweets in NoSQL database 5
TOTAL 15 out of 15

Zip the files and name the archive as MAPD725-A2-YourLastName-ParnerLastName.doc, e.g.


MAPD725-A2-Doe-Lee.doc
Due date: Thursday, Feb 21, 2019, 11:59 PM (Week 7)

Anda mungkin juga menyukai