With　the　continuing　penetration　of　sensor　devices　and　the　development　of　wireless　communication　techniques,　increasing　number　of　applications　involving　context　awareness　in　ubiquitous　computing　have　been　used　in　daily　life.　How　to　collect　data　from　mobile　devices　at　a　low　energy　cost　and　to　mine　contextual　habits　of　users　remains　a　key　challenge　for　ubiquitous　computing.　We　proposed　an　efficient　context-　awareness　computing　middleware　system,　Senz.　By　leveraging　high-efficiency　mobile　data　transmission　method,　using　high-performance　context　recognition　algorithms,　and　combining　mobile　data　with　online　third-party　data,　this　middleware　system　can　recognize　various　user　behavior　patterns　reliably,　accurately　and　efficiently.　Experiments　show　that　the　Senz　recognition　accuracy　of　context　activity　is　above　83%　on　average　and　the　energy　cost　is　relatively　low.　By　integrating　Senz　SDK　and　cloud　computing　engine,　developers　can　build　rich　user　experience　apps　with　better　understanding　of　users＇　behavior　data　and　providing　various　personalized　contextual　services.　©　2017　IEEE.