| Grant number: | 13/05757-2 |
| Support Opportunities: | Regular Research Grants |
| Start date: | October 01, 2013 |
| End date: | December 31, 2015 |
| Field of knowledge: | Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques |
| Principal Investigator: | Leandro Augusto da Silva |
| Grantee: | Leandro Augusto da Silva |
| Host Institution: | Faculdade de Computação e Informática (FCI). Universidade Presbiteriana Mackenzie (UPM). Instituto Presbiteriano Mackenzie. São Paulo , SP, Brazil |
| Associated researchers: | Leandro Nunes de Castro Silva |
Abstract
The advances in computation and communication technology have produced a problem of data overabundance. Just to give an overview: the number of Internet users in worldwide have a considerable rises from sixteen million people in 1995 to nearly two billion in 2011; the number of papers published only in English Wikipedia exceeded five hundred thousand in 2005 to nearly four million in 2011; the time necessary for the radio audience to reach fifty million people was thirty-eight years while the TV audience needed thirteen years and for Internet just four years; the number of daily searches using Google surpasses one billion; two hundred million of Tweets are written per day and three billion videos are viewed on YouTube daily. The number of hour of video sent to Youtube in 2010 was thirteen million which corresponds to approximately eight years of content uploaded every day.The problem with the large amount of data consists of system capability to collect and to store data, which has surpassed the ability of the human to analyze and to extract knowledge from them. This effect is mainly caused by the growth of the social networks, mobile devices and the availability of data storage as services and remote processing from data (cloud computing). The database management systems (DBMSs) have also improved the storage, not only textual data, but also multimedia data, i.e., those that use one or more content forms like text, audio, images, video and interactivity. Thus, the consolidated Structured Query Language (SQL) used to access the database is not sufficient and efficient in applications that uses multimedia data.With the reality of multimedia data, the traditional query using text as input parameter must also be reviewed to allow a query using multimedia data as an input parameter. The Google search engine, the most used in the world, allows in its portal search for images from photos, but still retains the option to search for images from texts. In the same way, Facebook has available in its social network website an tool to search people using marked face image.However, the major problem of doing a query for images is the semantic gap between what the user wants to find and what the system retrieve as an answer. The image representation for characteristics as contour, areas and shapes, or any relationship between these features cannot guarantee that the search system can distinguishes foot from hand, for example. One reason maybe because the high degree of similarity between the shapes. On the other hand, the image representation by keywords, the meaning defined from peoples can be different. Using the example above, the hand image can be a body part or a clothing company. In this case, the efficiency of the query image by a hand could be achieved in a two-step process: in the first a query from text (annotations, keywords, etc) and, in the second a query by image characteristics (contour, area, texture, etc). However, in other applications the best efficiency could be achieved with the combination of the two representations.In summary, the main issue to be investigated in this project is the use of text and image features to represent multimedia data and to extract knowledge from this kind of data. In this context, it is necessary to apply techniques and tools to transform, intelligently and automatically, the available multimedia data to useful information that represents knowledge to assist strategic decision making in business. (AU)
| Articles published in Agência FAPESP Newsletter about the research grant: |
| More itemsLess items |
| TITULO |
| Articles published in other media outlets ( ): |
| More itemsLess items |
| VEICULO: TITULO (DATA) |
| VEICULO: TITULO (DATA) |