Published Paper Details:

Search engines act as a vital role in information retrieval and knowledge discovery. Due to the huge amount of data on web and social sites it is a challenging task to find suitable information as requested by user efficiently even by using search engines. Also, the above frameworks specified will gather large amount of information, which could be helpful for end users in different manners yet would have neglected the original meaning, henceforth showing irrelevant results. In Order to reduce those problems, topic modelling plays a vital role in filtering data to improve efficiency in information retrieval. Topic modelling is an unsupervised machine-learning model which is one of the important tools for information retrieval as it plays a vital role in clustering the documents. In this modelling technique each document represents a set of / collection of bags of words and labelling appropriate word distributions to the defined topic is a challenging task. Traditional topic model such as Latent Dirichlet Allocation, which is a model that predicts topic space in a document using Dirichlet Distribution. The major drawbacks of LDA are that the topics are fixed and must be known before the time of running the model, Dirichlet topic distribution cannot capture correlations, etc. To overcome these problems, the model used Correlated Topic Modelling (CTM) to find the correlation between words and phrases, to speed up the processing and to increase the scalability. Correlated Topic Model (CTM) helps in finding the correlation between the hidden topics in the collection, and enables the construction of graphs that are related to topics collected and document that allow a user to go through the collection of topics in a standard manner. In this paper experimental results show that how CTM plays a important role than LDA in extraction of accurate information, and to enhance the optimization of the search query on a smaller scale and demonstrate its use as an major tool for large document collections.