Abstract: Unstructured data mining has gotten to be topical as of late because of the accessibility of high-dimensional and voluminous computerized substance (known as "Big Data") over the venture range. The Relational Database Management System (RDBMS) have been utilized over the previous decades for substance stockpiling and administration, be that as it may, the constantly developing heterogeneity in today's data requires another capacity approach. Subsequently, the NoSQL database has developed as the favored storeroom these days since the office bolsters unstructured data stockpiling. This makes the need to investigate proficient data mining procedures from such NoSQL frameworks since the accessible devices and systems which are planned for RDBMS are regularly not straightforwardly relevant. In this paper, we concentrated on points and terms mining, in light of bunching, in archive based NoSQL. This is accomplished by adjusting the engineering configuration of an investigation as-an administration system what's more, the proposition of the Viterbi calculation to improve the exactness of the terms order in the framework. The outcomes from the pilot testing of our work show higher exactness in examination to some already proposed procedures, for example, the parallel search.

Keywords: Unstructured Data Mining, Big Bata, Viterbi algorithm; Terms, NoSQL, Association Rules, classification, clustering.