Motive-based Search using a Recommendation-driven Visual Divide and Conquer Approach

A novel generation of (e.g. touch-driven) applications leads to a new universe of interaction paradigms and a growing need for simple, inspiring and smart interfaces while the size of searchable data sets is increasing permanently (big data). A system intended for non-experts should only present information the user needs to solve his task, instead of confronting him with the large and complex underlying data structure. In this paper, we focus on users wanting to perform a product search driven by a vague information need. We call this kind of search motive-based search, which is often initiated by unconscious motives and expectations that are difficult to transform into a specific search query at the beginning of the search process. Hence, the user needs guidance to fulfill his search task. A search approach will be developed in this paper, which allows a step-by-step reduction of the result set by selecting (visualized) concepts such as "beach", "relaxing" and "culture". Concepts are often organized as multiple faceted hierarchies (polyhierarchies) to represent different views on things. Hierarchies can also be used as navigation paradigm, named faceted search. We will present the significant flaws of this approach concerning larger knowledge bases. Alternatively, we propose a selection-based recommendation-driven search, based on the principle of divide and conquer. An experiment compares both approaches proving that the proposed approach allows to solve the given search tasks in shorter time and with less effort.