An　eye　movement　data　based　online　user　state　recognition　method　and　device,　　the　method　comprising　the　following　steps　:　　gathering　eye　movement　data　when　an　online　user　watches　a　webpage,　　and　pre-processing　the　eye　movement　data　to　acquire　the　pre-processed　eye　movement　data　(S1);　　performing　feature　extraction　on　the　pre-processed　eye　movement　data　to　acquire　a　plurality　of　eye　movement　feature　parameters　(S2);　　inputting　a　plurality　of　eye　movement　feature　parameters　of　the　online　user　to　a　preset　state　classifier　to　compare　so　as　to　recognize　the　current　working　state　of　the　online　user　as　a　searching　state　or　a　browsing　state　(S3).　The　present　invention　recognizes　two　working　states　of　online　user,　　searching　and　browsing,　　and　adopts　an　active　sensing　technology　to　extract　two　eye　movement　features　as　well　as　a　classification　method　based　on　a　support　vector　machine,　　and　has　a　fast　learning　speed,　　a　high　classification　accuracy,　　and　a　high　recognition　efficiency.