# Final Version ## 耗時 ### With pinyin | Step | Volume | Time | | ------------------------------------------------------------------------ | ------- | ------------------- | | Pinyin preprocess | 42150it | 00:08, 5070.43it/s | | Decoding Message | 9987it | 00:00, 34194.37it/s | | Detecting account | 9987it | 00:04, 2245.44it/s | | Determining language | 9987it | 00:21, 464.02it/s | | Predicting zh-message label, detecting rule-based word, detecting pinyin | 4480it | 35:07, 2.13it/s | ### Without pinyin | Step | Volume | Time | | ------------------------------------------------------------------------ | ------ | ------------------- | | Decoding Message | 9987it | 00:00, 34194.37it/s | | Detecting account | 9987it | 00:04, 2245.44it/s | | Determining language | 9987it | 00:21, 464.02it/s | | Predicting zh-message label, detecting rule-based word, detecting pinyin | 4480it | 26:10, 2.85it/s | ## 優化 ### 拼音表預處理 | Step | Volume | Time | | ------------------------------------------------------------------------ | ------- | ------------------- | | Pinyin preprocess | 42150it | 0.03 sec | - 直接讀取拼音表,省去拼音轉換的時間 (減少近8秒) ### pinyin detect 階段將已判斷過的label存起來 | Step | Volume | Time | | ------------------------------------------------------------------------ | ------ | ---------------- | | Predicting zh-message label, detecting rule-based word, detecting pinyin | 4480it | 27:20, 2.73it/s | - 此方法可在遇到相同拼音或是相同詞時,不需要再進到拼音檢測的環節 - 耗時減少8分鐘左右(+0.6it/s)