Discovering　and　seizing　technology　opportunities　is　key　to　innovation　at　all　levels.　However,　there　are　several　open　issues　in　the　existing　research　into　the　discovery　of　technology　opportunities,　such　as　the　insufficient　specification　of　technology　opportunities,　defining　the　features　of　opportunities　in　a　way　that　may　lead　to　the　exclusion　of　some　valuable　opportunities,　and　a　lack　of　empirical　support　for　evaluation　criteria.　This　study　proposes　a　new　approach　to　technology　opportunity　discovery　that　attempts　to　address　these　issues.　Our　approach　uses　patents　as　a　data　source　and　constructs　domain　knowledge　networks　(DKNs)　automatically　based　on　the　syntactic　dependencies　of　technological　words.　We　represent　technology　opportunities　as　connected　subnetworks　within　DKNs　and　use　a　regression　analysis　of　historical　patents　to　obtain　significant　variables　that　affect　the　value　of　technology　opportunities.　These　are　then　used　to　form　an　objective　function　for　searching　for　and　interpreting　optimal　opportunities.　Ant　colony　optimization　is　applied　to　discover　the　optimal　set　of　technology　opportunities.　The　feasibility　and　effectiveness　of　the　proposed　approach　are　demonstrated　by　empirical　research　into　a　technology　for　measuring　mechanical　vibrations　or　sound　waves　by　electromagnetic　means.