Efficient query clustering technique and context well-informed document clustering

Abstract

Data clustering plays a crucial role in extracting useful information based on the user interest. Traditional query clustering algorithms work on the collection of previously available data from the query stream. As we observe day by day the topic of interests, popularity, query meaning is changing. However, it is quite challenging as the queries are incomplete, ambiguous and short. Existing clustering methods like k-means or DBSCAN cannot assure to perform well in such fully measurable environment. There is high demand for enhancement of algorithms that can indulge in the prediction of characteristics, as the new data is being added to the data mob without implementing a complete re-clustering. So, proposing a new enhancement paradigm for query and context well-informed query document clustering. Even through analysis of user’s click-through log and hierarchical agglomerative clustering, we can achieve good results, but, however, it is computationally quite expensive. In order to overcome the problem, the proposed enhancement model attains both the query and document cluster quality. This model in regular intervals updates the new information which is being produced and can be applied in a distributed environment. And also, the suggested paradigm can be related to the outcome of hierarchical query clustering algorithms which produces query clusters and as well as document clusters. This proposed system not only concentrates on achieving accuracy, but also can show a remarkable speedup.