Session 12 - Data & Opinion Mining

Abstract: Instagram has been the fastest growing social network in the last three years. It allows users sharing their status by uploading pictures with a description text, a location and some hashtags which not necessarily describe the content of the pictures, so this work presents a methodology to detect relevant content topics of pictures associated to a particular hashtag through text mining techniques and computer vision tools. For this purpose 7382 pictures associated with the hashtag #allyouneedisecuador were collected. Results show he most relevant topics and that the similarity between both descriptions is low.

Authors: Angel Fiallos, Karina Jimenes, Carlos Fiallos and Stalin Figueroa

Abstract: ​Entrepreneurs have been subjects of great interest among researchers since they are commonly recognized as important players for economic growth worldwide. A variety of studies have already been conducted, in which entrepreneurs motivations, emotions, choices, income, behavior, personality, and thinking, amongst others, were analyzed. However, little has been investigated about the interactions of entrepreneurs within the context of online social networks, and even less at a geographical level where they develop their duties. Two of the well-known organizations in the world, the Global Entrepreneurship Monitor (GEM) Consortium and the Global Entrepreneurship and Development Institute (GEDI), provide valuable insights about the entrepreneurial activity and the countries' entrepreneurship ecosystems, based on several dimensions. Yet, none of such dimensions have taken into account the speech of entrepreneurs in social media, and the reproducibility of such studies presents considerable limitations as they are very costly and highly demanding. The present study proposes a framework to fetch users, grouped by their country of residence, from a target community in the micro-blogging platform Twitter. By using natural language processing and data mining techniques over 219M posts, authored by 135K entrepreneurs of 65 countries, we found significant differences on the psycholinguistic dimensions present on these entrepreneurs speech. Results indicate that African entrepreneurs show higher scores regarding negative emotions than the rest of the entrepreneurs' population. In addition to this, we found that entrepreneurs from developed countries exhibit higher scores in positive emotions than other entrepreneurs in the world.​

Authors: Leonardo Kuffó, Carmen Vaca, Edgar Izquierdo and Juan Carlos Bustamante