This paper presents GREN, a method to recognize and normalize mobile phone names from domain-specific Internet forums. Users often use informal abbreviations or nicknames for products, making feedback collection challenging. GREN first generates candidate names to capture variations and short forms, then uses a Conditional Random Field (CRF)-based recognizer to predict if a candidate name refers to a specific phone model. A rule-based component then normalizes the recognized name to its formal form. Tested on over 4000 sentences, GREN achieves precision and recall of 0.918 and 0.875, respectively, outperforming baseline methods.