Authors : B. SunilSrinivas, P.N. Santosh Kumar, A. Govardhan and C. Sunil Kumar
Abstract: Text-Mining (TM) refers generally to the practice of extracting attractive and non-trivial information and facts from unstructured text. TM includes several Computer Science (CS) regulations with a strong direction towards Artificial Intelligence (AI) in general including but not limited to Pattern Recognition (PR), Neural Networks (NN), Natural Language Processing (NLP), Information Retrieval (IR) and Machine Learning (ML). A significant variation with search is that search requires a user to identify what he or she is looking for while TM attempts to realize information in a model that is not known earlier. TM is mainly motivating in domains where users have to invent new information. This is the case for, e.g., in criminal enquiries and legal findings. Such examinations require 100% evoke, i.e., users can not meet the expense of missing relevant data. In distinction, a user searching the internet for background information using a benchmark Search Engine (SE) simply requires any data as long as it is reliable. Increasing evoke almost positively will decrease accuracy involving that users have to browse huge collections of documents that that may or may not be relevant. Standard procedures use language expertise to increase accuracy but when text collections are not in one language are not domain specific and or contain variable size and type documents either these schemes fail or are so complicated that the user does not understand what is happening and loses control. A different technique is to combine standard significance ranking with Adaptive Filtering (AF) and Interactive Visualization (IV) that is based on characteristics that have been mined earlier.
B. SunilSrinivas, P.N. Santosh Kumar, A. Govardhan and C. Sunil Kumar, 2014. A New Text Mining Approach in Search Technology. Asian Journal of Information Technology, 13: 93-98.