Survey Day 1

About the challenge

Groupes

Welcome to the Generative AI Hackathon! This event is tailored for Neoma business school students eager to explore the realm of generative AI applications. In this hackathon, participants will dive into the world of Large Language Model (LLMs) using AWS Generative AI services to develop functional apps. 

Context: Predicting product-related question responses using similar products

Predicting the response to a product-related question is an emerging research area that has recently gained a lot of attention. Responding to subjective, opinion-based questions is the most challenging due to the dependence on user-generated content. Previous work has mostly focused on response prediction based on reviews; however, these approaches fail for new or unpopular products that have little (or no) reviews available.

A new method allows for accurately predicting the response to a product question, by leveraging knowledge from similar products.

The goal is to automatically answer customer questions about products, a popular feature on e-commerce websites. However, when there are no relevant questions and answers in the archives, automatic tools struggle to provide adequate responses.

Product Question Answering (PQA) aims to automatically answer customer questions to improve their online shopping experience. Current research mainly focuses on searching for answers in unstructured text, such as product descriptions and user reviews, or in structured knowledge bases with predefined schemas. Besides these two sources, much product information is represented in a semi-structured way, for example, in the form of key-value pairs, lists, tables, JSON and XML files, etc. This semi-structured data can be a valuable source of answers as it is better organized than free text, while being easier to construct than structured knowledge bases. However, little attention has been paid to it.

To this end, the authors present semiPQA, a dataset for evaluating PQA on semi-structured data. It contains 11,243 human-written questions on JSON-formatted data covering 320 unique attribute types. Each data point is associated with a manually annotated text describing its content, in order to be able to train a neural answer presenter to present the data in a natural way (Publications).

General Learning Objectives

• Stimulate innovation and creativity capabilities to solve issues via data science and machine learning
• Foster team dynamics and intelligence through thinking, exchanges and analytical choices
• Allow students to build on cumulative learning and explore large language models (LLMs) and understand how to extend their general capabilities
• Being able to turn data to actionable insights for decision making and marketing performance
• Implement data visualization and communication skills

Specific objectives

• Offer data analysis students the opportunity to conduct a study on Amazon's verbatim customer data to detect themes and identify opportunities to make better product recommendations - including through techniques like question answering, summarization, etc.
  
• Allow these students to explore new capabilities by integrating generative AI models to analyze the data in novel ways.

Examples of expected deliverables for the jury:

• A synthesis of the learnings obtained from the analysis of this data.
• An analysis grid of the themes addressed by consumers that could be used to accelerate product knowledge.
• An actionable chatbot Q&A expérience, with natural language interactions to answer customer questions.

Hackathon Jury Profiles :

• CEO e-commerce company
• Chief Data Officer
• Chief Technology Officer
• Recommendation System Product Manager
• IA-NLP experts
• Customer experience Manager

During the hackathon, you'll learn how to use AI tools in workshop and get help from experts. Then, you'll work with your team to build your own application. At the end, you'll show your program to judges, who will decide which ones are the best.

It's a great opportunity to learn new skills and meet people in the industry!

Schedule

Day 1 - 7 hours (9h30-13h  / 15h-18h30 )

• 9:30 AM Démarrage du Hackathon : Brief et Présentation du projet : objectifs, livrables attendus et cadre de réalisation
• 10 AM - 11:00 L’IA générative avec AWS
• 11:00 AM - 01 PM Travail en groupes
• 4:00 PM : Présentation en groupe devant comité animateurs Hackathon : de l’état d’avancement (10 min Presentation + Q/A) – 5 slides Max (Optionnel) # Attendu : 1/Proposition de valeur– 2/Première approche analytique -3/Quoi & Comment construire
• 4:00 PM - 06.30 PM Poursuite du Travail en groupes

Day 2 - 8 hours (9h-13h / 14h30-18h30)

• 9:00 AM Start of Day 2 : Débrief intermédiaire et poursuite des travaux d'analyse des groupes
• 4:00 PM Démarrage des restitutions des groupes devant jury - 15 mn de présentation ppt + Q/A
• 6:00 PM Débrief Général et Annonce des Groupes Finalistes

• Creativity
  How original is the idea? Does it address a potentially valuable use case?
• Applicability
  Clarity of the use case (opportunity addressed)
• Customer Experience
  Is the solution engaging to interact with?!
• Quality of the Application
  How well does your solution use GenAi capabilities ?
• Project Contribution
  How did you contribute (Team & Individuallly) to the project ?
• Methodology
  How did you get to the solution ? Which analytical skills and methodology did you use ? relevancy of your analysis.