Granular　computing　is　a　general　computing　paradigm　of　problem　solving　for　effectively　using　granules　in　problem　solving.　From　the　viewpoint　of　granularity,　this　paper　presents　a　new　granular　computing　data　cycle　model　in　which　information　granule　construction,　granular　space,　information　granule　operations,　and　knowledge　acquisition　approach　are　proposed　respectively.　To　do　a　more　user　oriented　way,　a　multi-strategies　model,　namely,　Base-Level　learning　strategy,　user　interest　retention　strategy,　and　starting　point　strategy,　introduces　to　unify　search　and　reasoning　for　effective　problem　solving.　Then,　we　discuss　granularity-based　unification　using　three　strategies　mentioned　above　and　show　three-levels　of　granularity　from　human　problem　solving　to　a　wide　variety　of　user　needs　satisfied.　Furthermore,　on　the　basis　of　user-centric　multi-strategies　and　granular　information　processing,　we　develop　a　conceptual　framework　of　granularity-based　knowledge　retrieval　model,　which　enlarges　the　application　areas　of　granular　computing.　©　2013　ACADEMY　PUBLISHER.