基于自然语言处理及语音识别方法的电信业务软件设计

发布时间：2018-10-09 13:41

【摘要】：一直以来,如何识别语言及处理都是人工智能开发者们潜心研究的重要课题。而一些如定理证明、问题解答、模式识别、机器博弈和机器人科学等虽属于自然语言处理的其他领域,但也都有着一致性,其最根本的都是信息表达和如何利用问题。也就是说如何对获取到的各种不尽相同的信息,最后通过计算机以一种可以理解的方式方法来将信息输出。在我们日常使用的通信系统中,只要加入自然语言和语音识别等技术,可以把传统通信服务转变成为一个服务的“提供者”和生活“伙伴”。使用具有语言处理和语音技术能力的通信网络,人们可以非常便捷地从远端的数据库系统中查询与提取有关的信息。自然语言处理技术和语音识别技术正逐步成为信息技术中人机接口的关键技术,在不久的将来,通过语音识别技术与语音合成技术的结合,人们就能够通过语音命令来进行原需要利用键盘才可以进行的操作。本文探讨如何通过对语音识别及语义理解能力平台的建设,来降低智能技术在电信业务中应用的门槛,提高先进技术复用能力。文中首先对业内以苹果公司的语音控制功能为代表的人工智能技术应用的根本意义进行探讨,对原有语义理解能力平台进行分析,找出可优化部分,增加能力平台在业务调用,数据挖掘,理解能力几个方面来完善平台能力,以此解决面向电信业务的行业语料积累的问题,提升语音识别及语义理解的准确程度;通过对平台原型进行测试,使查全率、查准率,达到一个平衡,可完成电信自身的先进技术积累,节约新技术在电信业务应用开发成本。最后,对加载业务能力的语义平台进行性能测试与分析。通过设置不同的仿真场景,评估该系统在多用户并发时的处理性能、响应时间和可靠性等数据指标,结果表明:设计方案有效可行。
[Abstract]:All the time, how to recognize language and process is an important research topic for artificial intelligence developers. Some of them, such as theorem proof, problem solving, pattern recognition, machine game, robot science and so on, belong to other fields of natural language processing, but they all have consistency. The most fundamental problem is how to express information and how to use it. That is to say, how to output the information in a way that can be understood by computer. In our daily communication system, we can transform traditional communication services into "providers" and "partners" of services by adding technologies such as natural language and speech recognition. With a communication network with language processing and voice technology, people can easily query and extract information from remote database systems. Natural language processing technology and speech recognition technology are gradually becoming the key technology of man-machine interface in information technology. In the near future, through the combination of speech recognition technology and speech synthesis technology, People can use voice commands to perform operations that could have been done using the keyboard. This paper discusses how to reduce the threshold of application of intelligent technology in telecommunication service and improve the reuse ability of advanced technology through the construction of speech recognition and semantic understanding ability platform. In this paper, we first discuss the fundamental significance of artificial intelligence technology application, which is represented by the voice control function of Apple, and analyze the original semantic understanding ability platform, find out the part that can be optimized, and increase the ability platform in business call. Data mining, understanding ability to improve the platform capability, to solve the problem of industry language data accumulation for telecom services, improve the accuracy of speech recognition and semantic understanding, through testing the platform prototype, make the recall rate, The precision rate can reach a balance, which can complete the advanced technology accumulation of telecom itself and save the development cost of the new technology in telecommunication business application. Finally, the performance of the semantic platform loaded with business capability is tested and analyzed. The performance, response time and reliability of the system are evaluated by setting different simulation scenarios. The results show that the design scheme is effective and feasible.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.34

【参考文献】