MK99 – Big Data
Big data
for business
Pr. Clement Levallois – 2016/2017
MK99 – Big Data
The MUSE Project:
Multilingual Sentiment Analysis for Twitter
• 40% of your total grade for MK99
• Runs fr...
MK99 – Big Data
What is the project about?
• In 2012, I created Umigon (www.umigon.com)
• A tool that guesses if tweets ar...
MK99 – Big Data
Work packages
1. Build the produt: creating the lexicons for each language we want to target
2. Study how ...
MK99 – Big Data
(each of the 6 work packages listed in the previous slide
are now detailed in the next slides)
MK99 – Big Data
1. Build the product
• Check the community “MK99 - MUSE Project” on Connections
– « Connections » is the i...
MK99 – Big Data
2. Market research / business model
generation
• What are the use cases for this product?
• Who is the com...
MK99 – Big Data
3. Take legal steps
• Meaning of our action: informal classwork? startup?
• Organization: none? corporatio...
MK99 – Big Data
4. Marketing the product
• Create a simple CRM
• How should we position the product?
– And this follows: h...
MK99 – Big Data
5. Writing an academic paper to report
on the results
• Objective: publishing in a top journal in manageme...
MK99 – Big Data
6. Exec team
• Liaise with all teams to get the project forward, solve
issues that arise (in particular, h...
MK99 – Big Data
Pick a team
• Everybody is expected to contribute to team 1.
• In team 1, we need one student per language...
MK99 – Big Data
What you will learn
• This project will give you a hands-on
experience on dealing with data to create
valu...
MK99 – Big Data
Next steps
• Go to www.umigon.com and see what it does
• Read the paper explaining how umigon works, start...
Upcoming SlideShare
Loading in …5
×

Presentation of muse project

325 views

Published on

A project for the MK99 class (Big data for business) at em Lyon Business School, 2016 / 2017

Published in: Education
  • Be the first to comment

  • Be the first to like this

Presentation of muse project

  1. 1. MK99 – Big Data Big data for business Pr. Clement Levallois – 2016/2017
  2. 2. MK99 – Big Data The MUSE Project: Multilingual Sentiment Analysis for Twitter • 40% of your total grade for MK99 • Runs from Session 1 to Session 8 • Session 8 : the teams pitch their results!
  3. 3. MK99 – Big Data What is the project about? • In 2012, I created Umigon (www.umigon.com) • A tool that guesses if tweets are negative, positive or neutral in sentiment. • Works only for tweets in English • Evaluated in 2016 as the best analyzer among 27 solutions for social data. (https://arxiv.org/pdf/1512.01818.pdf , see Table 11). • This project: creating together an innovative product: a multilingual version of Umigon and market it, for real. • To the best of my knowledge, no product does multilingual sentiment analysis on the market, beyond 5 or 6 languages. We will do better.
  4. 4. MK99 – Big Data Work packages 1. Build the produt: creating the lexicons for each language we want to target 2. Study how this product can find customers (market research / business model generation) 3. Decide which kind of of organization we want to build around the product 4. Communicate / find prospects / move them to leads / close sales. 5. Write an academic paper to report on our results. 6. An exec team to coordinate all this.
  5. 5. MK99 – Big Data (each of the 6 work packages listed in the previous slide are now detailed in the next slides)
  6. 6. MK99 – Big Data 1. Build the product • Check the community “MK99 - MUSE Project” on Connections – « Connections » is the intranet for profs and students at EMLYON, in complement to Brightspace… – The URL should be: https://apps.ce.collabserv.com , then go to « Communities ». • Contribute to the Excel files where you are a native speaker – Questions about how this works? Please check the file « instructions.docx » (not yet written as of Sept 12, but coming soon!) • Role of each language team: deliver lexicons which accurately classify tweets in pos / neg / neutral categories!
  7. 7. MK99 – Big Data 2. Market research / business model generation • What are the use cases for this product? • Who is the competition? Potential partners? • How should the product be delivered? • What pricing? • What is our cost structure? • Role of the team: research these issues and come up with decisions regarding delivery channels, pricing, recommendations for partnerships, and a financial plan. All that could be synthesized in a business model, following the handbook « Business Model Generation » – https://www.amazon.com/Business-Model-Generation-Visionaries-Challengers/dp/0470876417
  8. 8. MK99 – Big Data 3. Take legal steps • Meaning of our action: informal classwork? startup? • Organization: none? corporation? Non profit? Who leads? Who bears the risks and costs, who gets the eventual benefits? • Who owns the IP? • Role of the team: debate and consult on these issues, then implement a solution.
  9. 9. MK99 – Big Data 4. Marketing the product • Create a simple CRM • How should we position the product? – And this follows: how should be brand it? How should we communicate about it? • Find prospects, turn them to leads, close deals. • Role of the team: create a CRM (a well organized online Excel spreadsheet should be enough), define a positioning (should liaise with team 2 on this), then plan and execute branding, communication and sales.
  10. 10. MK99 – Big Data 5. Writing an academic paper to report on the results • Objective: publishing in a top journal in management (FT Rankings) • Write the paper – Method – Results (accuracy) – Managerial relevance • Role of the team – Decide on the positioning of the paper: for which journal is it the best fit? Then make a literature review, liaise with teams 1 and 2 in particular and write the research paper.
  11. 11. MK99 – Big Data 6. Exec team • Liaise with all teams to get the project forward, solve issues that arise (in particular, hicups in teams. HR skills would be great here). • Make sure the prof. is not overloaded by emails and micromanagement. • Role of the team: lead the MUSE project to completion.
  12. 12. MK99 – Big Data Pick a team • Everybody is expected to contribute to team 1. • In team 1, we need one student per language to act as group coordinator for this language. – (so, we need one student who will be the coordinator for « French », another for « Italian », another for « Chinese »…) • Pick the team (from team 2 to team 6) that you’d like to join. Teams will be effectively created on Sept 12.
  13. 13. MK99 – Big Data What you will learn • This project will give you a hands-on experience on dealing with data to create value: – Conception – Implementation – Direct contact with the market
  14. 14. MK99 – Big Data Next steps • Go to www.umigon.com and see what it does • Read the paper explaining how umigon works, start imagining how the logic could be transposed to your language (http://www.clementlevallois.net/download/umigon.pdf) • Open the file « en.xlsx » on Connections and get a feel of the data in it. • Be ready to make a choice for your team on Sept 12!

×