Automated illustration of multimedia stories

Submitted in part fulfillment of the requirements for the degree of Master in Computer Science

Using text-mining-assisted analysis to examine the applicability of unstructured data in the context of customer complaint management

Double Degree

Sentiment analysis on twitter for the portuguese language

Dissertação para obtenção do Grau de Mestre em Engenharia Informática

Analysis of panoramio photo tags in order to extract land use information

In the recent past, hardly anyone could predict this course of GIS development. GIS is moving from desktop to cloud. Web 2.0 enabled people to input data into web. These data are becoming increasingly geolocated. Big amounts of data formed something that is called "Big Data". Scientists still don't know how to deal with it completely. Different Data Mining tools are used for trying to extract some useful information from this Big Data. In our study, we also deal with one part of these data - User Generated Geographic Content (UGGC). The Panoramio initiative allows people to upload photos and describe them with tags. These photos are geolocated, which means that they have exact location on the Earth's surface according to a certain spatial reference system. By using Data Mining tools, we are trying to answer if it is possible to extract land use information from Panoramio photo tags. Also, we tried to answer to what extent this information could be accurate. At the end, we compared different Data Mining methods in order to distinguish which one has the most suited performances for this kind of data, which is text. Our answers are quite encouraging. With more than 70% of accuracy, we proved that extracting land use information is possible to some extent. Also, we found Memory Based Reasoning (MBR) method the most suitable method for this kind of data in all cases.

Development of a fundus camera for analysis of photoreceptor directionality in the healthy retina

The Stiles-Crawford effect (SCE) is the well-known phenomenon in which the brightness of light perceived by the human eye depends upon its entrance point in the pupil. This physiological characteristic is due to the directional sensitivity of the cone photoreceptors in the retina and it displays an approximately Gaussian dependency which is altered in a number of pathologies. Retinal imaging, a widely spread clinical practice, may be used to evaluate the SCE and thus serve as diagnostic tool. Nonetheless, its use for such a purpose is still underdeveloped and far from the clinical reality. In this project a fundus camera was built and used to assess the cone photoreceptor directionality by reflective imaging of the retina in healthy individuals. The physical and physiological implications of its development are addressed in detail in the text: the optical properties of the human eye, illumination issues, acquiring a retinal image formed by the eye, among others. A full description of the developmental process that led to the final measuring method and results is also given. The developed setup was successfully used to obtain high quality images of the eye fundus and in particular the parafoveal cone photoreceptors. The SCE was successfully observed and characterized. Even though considerable improvements could be done to the measurement method, the project showed the feasibility of using retinal imaging to evaluate the SCE thus motivating its usage in a clinical environment.

Text mining e twitter : o poder das redes sociais num mercado competitivo

Actualmente, com a massificação da utilização das redes sociais, as empresas passam a sua mensagem nos seus canais de comunicação, mas os consumidores dão a sua opinião sobre ela. Argumentam, opinam, criticam (Nardi, Schiano, Gumbrecht, & Swartz, 2004). Positiva ou negativamente. Neste contexto o Text Mining surge como uma abordagem interessante para a resposta à necessidade de obter conhecimento a partir dos dados existentes. Neste trabalho utilizámos um algoritmo de Clustering hierárquico com o objectivo de descobrir temas distintos num conjunto de tweets obtidos ao longo de um determinado período de tempo para as empresas Burger King e McDonald’s. Com o intuito de compreender o sentimento associado a estes temas foi feita uma análise de sentimentos a cada tema encontrado, utilizando um algoritmo Bag-of-Words. Concluiu-se que o algoritmo de Clustering foi capaz de encontrar temas através do tweets obtidos, essencialmente ligados a produtos e serviços comercializados pelas empresas. O algoritmo de Sentiment Analysis atribuiu um sentimento a esses temas, permitindo compreender de entre os produtos/serviços identificados quais os que obtiveram uma polaridade positiva ou negativa, e deste modo sinalizar potencias situações problemáticas na estratégia das empresas, e situações positivas passíveis de identificação de decisões operacionais bem-sucedidas.