In the Digital Era, social media provide to the fashion industry a powerful tool to stimulate discussion and build brand awareness, along with improving consumer relationships through interactivity and networking. As a result, social platforms like Facebook, Twitter and Instagram are mostly exploited to promote engagement but also to develop digital marketing strategies to increase online sales and retail store traffic.
However, well established companies and emerging ones have different approaches on social networks, which reflect their needs and goals in relation to their size and audience; legacy brands aim to reinforce their presence by establishing an online experience that is more accessible to the customers, while young designers try to develop their own identity and to gain visibility.
In this perspective, the objective of this project was to develop, by applying quantitative and qualitative analyses, a tool useful to support emerging fashion brands in building up incisive and profitable social strategies and to measure their success in terms of brand identity and awareness creation.
As a Solution, we propose to create a tool able to categorize social media posts and to carry on descriptive statistics about the brands themselves. For instance, our architecture would be virtually able to classify each post of a brand in a category and define the principal aspects of the brand in terms of frequency of:
• Post of a certain category: which types of posts do brand A use the most?
• Marketing strategy: does the brand post more about products or fashion events?
• Relationship to other brands: is brand A similar to brand B given the type of posts?
In our context, a high level of complexity is faced since the content to be classified comes from social media and it is the result of natural language expressions, symbols and images. Hence, the system has to be able to read and see social media posts in order to extract core information used to analyze and categorize them.
This poses two main technical issues on the task: the ability of the classifier to understand natural language which is intrinsically ambiguous and the necessity to understand the key components of a picture.
Principal academic Tutor
Paola Bertola, Department of Design, Politecnico di Milano
Elena Baralis, Department of Control and Computer Engineering, Politecnico di Torino
Marco Brambilla, Department of Electronics, Information and Bioengineering, Politecnico di Milano
Chiara Colombi, Department of Design, Politecnico di Milano
Federica Vacca, Department of Design, Politecnico di Milano
Andrea Vaccarella, Fluxedo
Federico Della Bella, Wardroba
Luca Grassano [Team Controller], Mathematical Engineering, Politecnico di Torino
Umberto Di Fabrizio, Computer Science and Engineering, Politecnico di Milano
Michele Invernizzi, Communication Design, Politecnico di Milano
Manuel Impellizzeri, Communication Design, Politecnico di Milano
Pasquale Mangano, Computer Engineering, Politecnico di Torino
Marco Manino, Computer Engineering, Politecnico di Torino
Silvia Massi [Communication Coordinator], Mathematical Engineering, Politecnico di Torino