Insights about media awareness are by-products of reasoning about online social networks (SNs) focusing on brands or target communities, helping in strategy decision-making on publishing business advertisement and preventing rumor and the spread of illegal information. The key issue we will present is the recently created methodology for finding insights using SNs. The process involves collecting all kinds of unstructured data: from brands to weird life styles, political position to nonsense questions, and combining everything based on a specific subject. Target communities are explored using member's information and their relationships with other communities. Therefore, a distinct model for classification allows the creation of rankings and relationships between products, lifestyles and members. Insights can be generated through rankings and bi- or tri-dimensional association maps, using correspondence analysis and visualization methods. In addition, many demographic variables collected from user profile or even generated by an association expert system allows different cross combinations and filtering. These relationships eliminate bias introduced in a conventional focus groups and/or Internet. Real cases will be presented, demonstrating the potential of this new insight generation tool compared to traditional qualitative analysis.
The use of the 'recall' methodology for radio audience measurement is widely adopted for good quality results and positive cost-benefit relationship. Nevertheless, although very effective for the evaluation of the station's performance, it presents important limitations for planning advertising for this medium as it does not allow the simulation of campaign results (reach, frequency and frequency distribution).Research using 'panel' methodology is an effective alternative, but with high implementation and maintenance costs, considering the necessary sample size to represent a very fragmented behavior. This paper delves into a creative alternative for the use of fusion algorithms to build a virtual panel from a group of individual respondents using a 'recall' survey.This proposal allows the joining of the advantages of the recall research (practicality and low cost) with the benefits of the 'panel' survey (possibility of campaign simulations), preserving the quality of the results.