An Approach for Truth Discovery by Resolving the Conflicts on Categorical Data

Data is playing an important role in the world for competitive advantages among nations, organizations and decision making for business development. It is indeed that every day quintillion bytes of different types of data from variety of sources are created and complex in processing (from Medical data, Business transactions, Data captured by sensors, Social media/networks, Banking, Marketing, Government data, etc.). The different sources produce large amount of structured, semi structured and unstructured data with varieties of data. Normally, the unstructured content collected by an organization is in the form of textual format, corporate documents to web pages and social media content. When number of sources produces various descriptions for a particular object, it leads to data conflict. It can be processed, when numerical data and measurement data are produced for the objects using big data tools and statistical tools for having high achievements. But, it is the challenging one to extract and analyse conflicted text from corpus data, collected from various sources. The objective of our work is to produce the true information even when distance between observed values and mean values are closest to each other.