CAJ | 학술논문

为提高逆向最大匹配算法的分词精度,本研究利用词频阙值,单字函数等方法取得了较好的消歧效果.实验结果表明:该分词算法既能遵循长词优先的原则,又能进一步识别和消除覆盖歧义.改进的RMM不仅在速度上仍保持较大优势而且在分词准确率上有了进一步的提高,对使用机械分词算法的中小型搜索引擎在提高分词精度方面具有一定的实用价值.
위제고역향최대필배산법적분사정도,본연구이용사빈궐치,단자함수등방법취득료교호적소기효과.실험결과표명:해분사산법기능준순장사우선적원칙,우능진일보식별화소제복개기의.개진적RMM불부재속도상잉보지교대우세이차재분사준학솔상유료진일보적제고,대사용궤계분사산법적중소형수색인경재제고분사정도방면구유일정적실용개치.
In order to enhance the accuracy of chinese word segmentation, using term frequency value and single character function, the present study has made great progress on the ambiguity resolution. The experiment shows this method is able to follow the long-word-first principle, and can further detect and resolve ambiguity. The improved RMM not only has a greater advantage in speed, but also increases the accuracy. It has practical value in the aspect of ambiguity resolution to the middle and small-scale search engines which adopt mechanical method.