Research On Semantic Retrieval Based On Link Sensor Data

The Internet of Things(IoT)is an important part of the new generation of information technology and has received a widespread attention in recent years.With the development of the Internet of Things technology,more and more intelligent devices have been putting into practical applications.As a result,With the generation of rich information resources,how to obtain useful information from the the raw IoT data through the semantic retrieval technology is an important issue that the Internet of Things field is facing.Because the original sensor data obtained by the Internet of Things sensing devices itself is with the characteristics of massive,heterogeneous,and multi-sourced,and the original sensor data presents the feature of singularity in content,which greatly increases the retrieval and utilization difficulty of data resource in Internet of Things.Most of the current retrieval systems are only focused on the basic information about Internet of Things devices,such as device name,device ID,deployed location,manufacturer,and the gateway the device belongs.Considering the above situation,by using the advantages of semantic technology and semantic data association methods in information description,representation,processing,and data content expansion,an IoT semantic retrieval model based on linked sensor data is built,aimed to solve the heterogeneity and unity issues of IoT data resources.Moreover,correlated with the related content in other fields to improve the ability of IoT information retrieval and services.The main research work and results are as follows:(1)A linked sensor data generation method based on the integrity of Internet of things data observation attribute is proposed.Aimed at the problem of the incomplete observation attributes of original Internet of things perception data,the word similarity comparison of the observation attributes based on the morphological corpus is used to form sensor data with complete observation attributes.The classic semantic sensor network ontology(SSN)in the existing IoT field is researched,and the related concept classes covered in the application field are abstracted,and the object attribute relationships between the classes are abstracted to build an IoT application oriented to the specific application field Ontology.Based on the linked sensor data generation model and the constructed IoT application ontology,the sensor data is linked with the associated web data resources to form linked sensor data.(2)A semantic retrieval model based on linked sensor data is constructed.Based on linked sensor data and ontology structure,the feature mapping method of similar structure based on RDF triples is adopted to fuse linked sensor data and ontology structure,so that to form a semantic retrieval database with "Device-Observation-Document" as the core content.Aimed at the inaccuracy of query words,the semantic similarity algorithm based on Chinese WordNet was used to attach the purpose of semantic preprocessing,and the TransE model was used to implement semantic matching in the semantic retrieval model.(3)A semantic retrieval system based on linked sensor data is designed and implemented.The system based on the retrieval library that was built through the linked sensor data to implement query word preprocessing and semantic retrieval functions,and to return the retrieved documents according to the degree of relevance.The test results show that the system can realize the relevant retrieval of IoT data resources,and has a good performance and certain practical value.In response to the needs of the Internet of Things retrieval service,based on the existing related link sensor data generation methods,the sensor data is subjected to attribute integrity processing to form linked sensor data,and the semantic retrieval model and retrieval system for processing these kind of data are constructed.An empirical study was conducted with agricultural tobacco cultivation as the background,and rigorous experimental results in this paper prove that this system can provide automatic information retrieval services for tobacco cultivation,besides,the method is not only applicable to the tobacco field,but also can be extended to other Internet of Things applications.