Innovative Feature Selection for Effective Context Resolution Using Natural Language Query Interface

Any system that supports human interaction through natural language has high utility and ease of use. The challenge in natural language arises due to difficulty in correct interpretation, disambiguation and context resolution. Use of natural language for information retrieval and other related activities enhances effectiveness of the process and provides greater flexibility to the users in terms of document access. To do so, use of a feature vector with respect to different perspectives in addition to metadata is proposed. The work presented here encompasses a generic architecture of context resolution and categorization of document through use of natural language to achieve the intended goal. The architecture encompasses various document indices along with methodology for lexicon analysis. It also uses metadata. The proposed document features (indices) along with lexical analysis will help in correctly determining the context through the limited query keywords. The architecture is domain independent and can be used for various applications in vernacular languages. To demonstrate the application of the architecture and its methodology, necessary discussion is also included in this paper with required technical details.