Normalization Methods for Multiple Sources of Data

Data is important part of any kind of applications, specifically in World Wide Web and Electronic documents. As the ICT (Information & Communication Technology) industry intensifies, the need for managing real data processing exceeds. In the other sense, with multiplying of mission critical applications in Artificial Intelligence like machine translation and text to speech conversion, speech to text conversion, data supplication etc. which are anxious with raw text data, data processing has become one of vital part in the research fields Normalization is the process of converting a series of real-valued numeric attributes into a 0 to 1 scale. In machine learning, data normalization is used to enable model testing less reactive to feature size. Data Normalization is the job of discovering these specific words and transferring them to the subsequent series of words that is defined as normalized words. A novel characteristic-based mining technique is used in similarity propagation framework and point wise mutual information pruning is proposed to extract characteristics systematically and precipitously. Primarily, characteristics named entity sets are acquired from the corpus corresponding to predefined syntactical guidelines. Then, the characteristics named entity sets are represented by frequency counting word vectors and calculating the similarity of characteristics words to cluster the characteristics named entity sets.