The Internet allows organizations managers access to large amounts of data, and this data are presented in different formats, i.e., data variety, namely structured, semi-structured and unstructured. Based on the Internet, this data variety is partly derived from social networks, but not only, machines are also capable of sharing information among themselves, or even machines with people. The objective of this paper is to understand how to retrieve information from data analysis with data variety. An experiment was carried out, based on a dataset with two distinct data types, images and comments on cars. Techniques of data analysis were used, namely Natural Language Processing to identify patterns, and Sentimental and Emotional Analysis. The image recognition technique was used to associate a car model with a category. Next, OLAP cubes and their visualization through dashboards were created. This paper concludes that it is possible to extract a set of relevant information, namely identifying which cars people like more/less, among other information.

ResumoA internet fez com que os gestores das organizações tivessem acesso a grandes quantidades de dados e esses dados são apresentados em diferentes formatos, em concreto, estruturados, semiestruturados e não estruturados. Esta variedade de dados é essencialmente proveniente das redes sociais, mas não só, também são provenientes da Internet of Things. Verifica-se para os dados estruturados que existem técnicas validadas, estudadas e maduras, mas para os outros tipos de dados, ou seja, semiestruturados e não estruturados tal já não se verifica. Neste poster, é apresentado um conjunto de técnicas de análise de dados para os dados semiestruturados e não estruturados, utilizando como principal bibliografia conferências de investigação na área de análise de dados. Palavras-chave: Análise de dados, Variedade de dados, Tipo de dados AbstractThrough the Internet, the organizations managers had access to massive amounts of data and these data are presented in different formats, namely, structured, semi-structured and unstructured. These variety of data is essentially generated from social networks, but not only, they also are generated from the Internet of Things, from machines, sensors, among others. While the structured data has techniques well studied, mature and validated, otherwise the other types of techniques, semi-structured and unstructured, this is no longer true. In this poster, a set of data analysis techniques is presented for the semi-structured and unstructured data by using as main bibliography data analytics conferences.