An Intelligent hybrid tool for finding and organizing relevant text
Sanjay Sharma, Arpana Rawal, Ani Thomas · 2009
work in intelligent text retrieval systems have shown improvements in knowledge representation techniques, irrespective of user-specific tasks viz. classification or categorization, summarization, indexing and ranking. Text miners have begun extracting and aggregating key concepts by slowly shifting from utilizing the explicitly available ontologies towards machine generated ones. In the present communication, the authors present text mining experiments in the closed-world domain to rank the documents, here chosen as academic realm. The proposal offers a two-stage hybrid tool, where a confusion matrix obtained from suitably chosen naive-bayes classifier is used to arrive at similarity matrix, that is put to hierarchical agglomerative clustering procedures. This is extended to render an accurately precise hierarchical topic and sub-topic sequencing in the considered domain of context. The resulting accuracy of term-to-term arrangements in topic hierarchy were found promising as the same was found to be preferred, when consulted with subject experts.