Objects are characterized by a collection of relevant features, and are represented as points in a multi-dimensional feature space. Existing Tree based techniques are all applicable to various queries such as spatial queries, range queries and many more. The techniques are used only for those queries that have coordinates. But the same technique is not applicable to those queries which don’t have coordinates. One of such query is Nearest Keyword Set (NKS) query. It is the set of userprovided keywords, and the result of the query may include k sets of data points each of which contains all the query keywords and forms one of the top-k tightest clusters in the multidimensional space. Keyword-based search in multi-dimensional datasets have facilitated many applications. Various algorithms have been contributed for queries with coordinates but rarely for the queries without coordinates. A method called Projection and Multi Scale Hashing i.e. ProMiSH is studied that uses a set of hash-tables and inverted indexes to perform a localized search. It also enables fast processing for NKS queries and retrieves the optimal top-k results.
Nearest Keyword Set (NKS) queries on multidimensional database are used in many applications such as photo sharing social network, graph pattern search and many more. Existing Tree based techniques are all applicable to various queries such as spatial queries, range queries and many more. The techniques are used only for those queries those have coordinates. Same technique is not applicable to those queries which don’t have coordinates. There are many existing techniques using Tree-based indexes for NKS queries, but the performance of these algorithm deteriorates with the increase in size in datasets.
Keyword based search in multi-Dimensional dataset is involved in many applications in today’s world. Therefore, solutions to the problem of Top-k nearest keyword set search in multi-Dimensional dataset are studied. Efficient search algorithms are studied that work with indexes for fast processing. A novel model called as Projection and multi-scale Hashing (ProMiSH) that uses random projection and hash based index structures and achieves high scalability and speedup is introduced.
Main focus – Hashing, Indexing, Multi-dimensional data, Querying.