Journal of Intelligent Systems (Sep 2018)
Cognitively Motivated Query Abstraction Model Based on Associative Root-Pattern Networks
Abstract
This paper attempts to investigate some aspects related to problems involved in textual inter-cognitive communication in the context of search queries. Furthermore, it aims at stressing on the root-pattern and morpho-phonetic dimension of a word meaning within a query, and its effects on understanding and predicting the intended information conveyed by some search patterns in a human language. As humans are inclined to use very few words possibly pervaded with vague and uncertain interpretational potential for requesting information, misinterpreting conveyed information in a query term might critically influence an inter-cognitive communication, particularly in case of Arabic- and Semitic-based computer systems. Furthermore, as phonetic patterns are involved in the mental word perception, an abstract morpho-phonetic query model is proposed based on the non-linearity of the morpho-phonetic characteristic of Arabic word cognition. This model suggests forming the intended query information by constructing morpho-phonetic query patterns relying on the most associative root-pattern subnetworks. An important advantage of this model resides in introducing the concept of query abstract morpho-phonetic vectors expressing query vector space. Furthermore, this approach suggests employing the fuzzy subsethood theorem as an assessment reflecting model accuracy and the closeness to human associative word-networks. Finally, it opens the discussion to consider indexing based on a higher level of abstraction, such as utilising patterns as cognitive search variables. Furthermore, as this model is capable of predicting most human associative query key terms, integrating these within certain human-machine interaction would improve inter-cognitive communications.
Keywords
