Research And Implementation Of Related Issues About Instruction-Oriented Knowledge Reasoning In Smart Home

With the development of information society, every individual in a society hopes to lead the wisdom of high quality life, so the need for Smart Home arises at the historic moment, and new sensing technology with the growing progress of intelligent information processing technology offers possibility to the implementation of such demand. At present, research and related products on Smart Home is still in the initial stage. In the process of using, although it has achieved the effective control of equipment, it has higher requirements of operating personnel, and there is great limitation in the aspect of users interacting with system, especially for people with disabilities. Based on these reasons, we design a way for users interacting with the system, that is, through the natural language. This kind of design not only meets the needs of people with disabilities, but also explores a new area which is the combination of the natural language and Smart Home.The key of control Smart Home by natural language is mapping natural language instructions into machine instructions with some relevant knowledge of the reasoning. First, the paper constructs a domain ontology for Smart Home. The ontology contains relevant concepts and relationship between concepts in Smart Home domain. It is used to provide the basis for knowledge reasoning. Secondly, it is necessary to find an effective sentence similarity matching algorithm to finally realize natural language instructions mapping into machine instructions. Based on this, the paper chooses the sentence similarity matching algorithm based on TFIDF through in-depth analysis of the sentence similarity matching algorithms and combined with the application of Smart Home. This not only ensures the accuracy and efficiency of instruction matching, but also guarantees the simplicity of the algorithm realization. Finally, it is also the core of this paper namely the design and implementation of reasoning rules based on Drools. Reasoning rules is mainly used to analyze and reason the user input natural instruction events. After a series of system reasoning, it will produce the events to execute machine instruction or no matching events of machine instruction, and according to the corresponding events, the rule system takes the next action and gives the user related feedback about reasoning results.Finally, combining related theories of ontology, rules and sentence similarity matching algorithm, the paper realizes a well interactive ontology management tool and a knowledge reasoning subsystem for instruction mapping. Then, through designing specific test cases, it verifies the efficiency and the accuracy of instruction mapping in this system.