Searching for associations between social media trending topics and organizations

 João Pedro Sousa Henrique

 

Master in Integrated Business Intelligence Systems

 

December, 2020

------------------------------------

Tese em MSIAD, classificada com 18 valores

------------------------------------

Abstract

This work focuses on how micro and small companies can take advantage of trending topics for marketing campaigns. Trending topics are the most discussed topics at the moment on social media platforms, particularly on Twitter and Facebook. While the access to trending topics is free and available to everyone, marketing specialists and specific software are more expensive, therefore small companies do not have the budget to support those costs. The main goal is to search for associations between trending topics and companies on social media platforms and HotRivers prototype is designed to accomplish this. A solution that aims to be inexpensive, fast, and automated. Detailed analyses were conducted to reduced the time and maximize the resources available at the lowest price. The final user receives a list of the trending topics related to the target company. For HotRivers were tested different pre-processing text techniques, a method to select tweets called Centroid Strategy and three models, an embedding vectors approach with Doc2Vec model, a probabilistic model with Latent Dirichlet Allocation, and a classification task approach with a Convolutional Neural Network used on the final architecture. The Centroid Strategy is used on trending topics to avoid unwanted tweets. In the results stand out that trending topic Nike has an association with the company Nike and #World- PatientSafetyDay has an association with Portsmouth Hospitals University. HotRivers cannot produce a full marketing campaign but can point out to the direction to the next campaign.

Keywords: Trending Topics, Text Similarity, Text Classification, Associations.

Ler mais aqui.

  

Candidatar a MSIAD aqui.

 

OMECO: Generating personalized business card designs from images

 Nuno Francisco Castro da Gama Antunes

 

Master in Integrated Business Intelligence Systems

 

October, 2020

------------------------------------

Tese em MSIAD, classificada com 19 valores

------------------------------------

Abstract

Rising competition in the retail and hospitality sectors, especially in densely populated and touristic destinations is a growing concern for many business owners, who wish to deliver their brand communication strategy to the target audience. Many of these businesses rely on word-of-mouth marketing, delivering business cards to customers. Furthermore, the lack of a dedicated marketing team and budget for brand image consolidation and design creation often limits the brand expansion capability. The purpose of this study is to propose a novel system prototype that can suggest personalized designs for business cards, based on an existing business card picture. Using perspective transformation, text extraction and colour reduction techniques, we were able to obtain features from the original business card image and generate an alternative design, personalized for the end user. We have successfully been able to generate customized business cards for different business types, with textual information and a custom colour palette matching the original submitted image. All of the system modules were demonstrated to have positive results for the test cases and the proposal answered the main research question. Further research and development is required to adapt the current system to other marketing printouts, such as flyers or posters.

Keywords: Computer Vision, Artificial Intelligence, Marketing, Design Generation, Natural Language Processing.

Ler mais aqui.

  

Candidatar a MSIAD aqui.