in information. The data-users (in this case, the companies) need to ensure the quality and trustworthiness of data and be able to trust in it in their businesses. At first, when collecting data, the user wants to ensure the reliability of the data and the data source, leaving out suspicious data. Secondly, when further processing and analyzing the data, the user wants to ensure that the quality and relevancy of data are appropriate for the specific situation. Reliable and valuable data enhances business decision making in sever always,enabling, for example, real-time demand predictions, the estimation of trends, and innovation of potential new products/services. The usage of unreliable data, such as data from suspicious sources, or corrupted, subjective, inaccurate or incomplete data,has a high risk for a company’s business,and may lead to poor or incorrect business decisions. Furthermore, the usage of value less and irrelevant data for certain situations causes a lot of unnecessary effort and expenses for companies. The evaluation of data quality has relevance in one or more data processing phase(s) of big data architecture (i.e.big data pipeline);in data extraction,data processing and analysis, and finally in decision making. Therefore, quality evaluation of big data must be considered during architecture design, when designing how the data goes through the pipeline of a big data system.Difficulties in quality evaluation are determined by the fact that data quality cannot be judged without considering the context at hand[10];the same quality attribute is applicable to different situations but the evaluation metric is different. In addition,there are no agreed definitions of quality attributes or classification of their applicability to certain contexts. Furthermore, the characteristics of big data, [6], [11], and [12] as such, set special challenges for quality evaluation. The growing amount of semi-structured and unstructured data, new ways of delivering information and user’s changed expectations and perceptions of data quality have been recognized as new challenges in data quality research [8]. Thus, it is obvious that new means are required for data quality evaluation for such kinds of big data. The purpose of this paper is to describe how to ensure the quality and trustworthiness of social media data for company’s business decision making. We introduce a novel solution for data evaluation, in which the data consumer can select the applicable quality attributes and evaluation metrics for the context and situation at hand, and evaluate the quality attributes with evaluation metrics. The solution follows the pipeline of the big data reference architecture of [7]. This paper is organized according to the following: Section 2 defines the basic terms used in this work, and provides state-of-the-art of the big data architectures, and the application of metadata,quality attributes,quality metrics and quality policies in business usage. Section 3 introduces our solution for data quality evaluation in big data architecture. Section 4 provides a case example of how the developments are used in practice; an industrial case company achieves insight into customer needs utilizing social media data. Section 5 provides the validation of the trial
usage of the solution and identifies the shortcomings and development targets. Finally, section 6 concludes the work.